Privacy Policy
1. Introduction
This Privacy Policy explains how Auspex Streamline S.L. ("Auspex", "we", "us") processes personal data in connection with Aelo — AI Communication Analyzer ("Service", "Platform").
Aelo is a B2B SaaS platform that provides AI-powered call analysis for businesses. We act as a data processor — our customers ("Customers", "Controllers") record calls and decide how the data is used. We process the data on their behalf, according to their instructions.
This policy covers two types of data processing:
- Section A: Data we process on behalf of Customers (call recordings, transcripts, analyses) — we act as Processor
- Section B: Data we process for our own purposes (platform accounts, website visitors) — we act as Controller
Section A: Processing on Behalf of Customers (Processor)
A.1 Our Role
When Customers use Aelo to analyze calls, the Customer is the Controller and we are the Processor. This means:
| Responsibility | Who |
|---|---|
| Deciding to record calls | Customer (Controller) |
| Obtaining consent from call participants | Customer (Controller) |
| Informing call participants about recording and AI analysis | Customer (Controller) |
| Determining retention periods | Customer (Controller) — configurable 1–365 days; 365 by default |
| Responding to data subject rights requests | Customer (Controller), using the tracked workflow and applicable product or support procedure |
| Securely processing and storing data | Auspex (Processor) |
| Providing rights-request tracking and applicable product or support procedures | Auspex (Processor) |
| Ensuring sub-processor compliance | Auspex (Processor) |
Pro Trial. A free (Pay-as-you-go) organisation receives the features of the Pro plan once, for 14 days from its next analysed call or chat (Terms of Service, section 9.6). While its Pro Trial lasts, the processing this policy describes for the Pro plan and above also applies to that organisation; when the trial ends, that processing stops for new conversations, and what it already produced is kept and deleted as described for it in this policy.
A.2 Data We Process on Behalf of Customers
| Data Category | Description | Retention |
|---|---|---|
| Audio recordings | Call recordings supplied by the Customer — uploaded through the app, sent through the public API (directly, or as a URL for Aelo to fetch — see A.3), or received via webhook from connected telephony/CRM | Customer's retention period (365 days by default, configurable 1–365); deleted together with transcripts |
| Transcripts | Text transcriptions generated from audio by ASR providers | Customer's retention period (365 days by default, configurable 1–365) |
| Search vectors | Numerical vector representations (embeddings) of segments of call transcripts and chat conversations, computed after the pattern-based PII redaction and the contextual masking described in A.3 for search by meaning (vectors computed before release 26.50.0: after the pattern-based redaction only) and stored in Cloudflare Vectorize so that users can find records by meaning. Each vector is stored with the identifiers of the record, project, organization and transcript it belongs to, the record type (call or chat) and the segment's length, and — when known — an identifier of the employee who handled the conversation (their user ID in Aelo, in the connected CRM or telephony, or the value supplied through the public API) and the record's date — and without any transcript or chat text; the text shown next to a search result is read from the stored transcript, masked the same way. | Follow the record: rebuilt from the current transcript when the record is transcribed again (if that build fails, the earlier vectors stay until the next index rebuild) or the index is rebuilt, and removed from the index after the record is deleted — by a user, at the end of the retention period or together with its project — normally within a few hours; while a large volume of deletions is being processed, removal can take several days |
| AI analyses | Quality scores, sentiment analysis, key points, sales indicators | Follows transcript retention |
| Criteria and their answers | Questions the Customer defines to be asked of every conversation — the name, the question, optional guidance on what counts, the answer values, and where the criterion applies (conversation categories, projects, channels) — and, for each analysed conversation within a criterion's scope, its answer: yes or no, a 0–100 score or one of the Customer's values, or that no confident answer was possible, with verbatim quotes taken from the PII-redacted transcript, a short explanation written by the model and — when known — an identifier of the employee who handled the conversation (their user ID in Aelo, in the connected CRM or telephony, or the value supplied through the public API). Criteria also include a standard set worded by Aelo (built-in criteria): standard questions (for example, whether the company was named, a next contact was agreed, call recording was disclosed or the employee used profanity) that are the only place where these checks of what was or was not said are made — the analysis itself no longer performs them; the jurisdiction the Customer selected may appear in their wording, the Customer pauses them and narrows where they apply in its criteria settings, within the limits Aelo sets (Aelo may also pause one), and their answers are stored and shown to the Customer in the same way — unless Aelo has switched built-in criteria off for the Customer, in which case they are not shown and the Customer has none of these checks. Analyses made before this change keep the checks they stored, and those are still shown | Answers follow transcript retention and are deleted together with the record; criteria themselves are kept for the duration of the Customer's account (an archived criterion is kept so that past answers stay readable) and are cascade-deleted when the account is deleted |
| Customer questions and decline reasons | For an organisation with the Voice of Customer section with its customer questions and decline reasons (Pro plan and above), for each conversation analysed after this was introduced: the questions the customer asked — a one-sentence summary written by the model, the customer's verbatim words taken from the PII-redacted transcript, for a call the second at which they were said and for a chat the number of the message they were written in, the question's subject (from a fixed list), whether the employee answered it and at which stage of the conversation it was asked; and, for a sales conversation, how it ended (progressed, postponed, declined or no decision) and, when the customer declined, the reason (from a fixed list) in the customer's words with their verbatim quote, and whether the customer stated the reason or the model inferred it. Phone numbers, emails, card numbers and IBANs in them are masked, and a summary or quote that may contain an identification number (such as a contract or order number) is not stored | Follows transcript retention; deleted together with the record and its analysis |
| Conversation themes | For an organisation with conversation themes (Pro plan and above, together with the Voice of Customer section and its customer questions and decline reasons): the organisation's own set of themes — the names and short descriptions written by its owner or administrators, and a record of who changed the set and when; for each conversation analysed while the set has an active theme, one to three themes of the set (or a mark that none fits) and, where the conversation has a subject the set lacks, a short suggested theme written by the model (one in which a phone number, email, card number or IBAN is detected, or that may contain an identification number, is not stored); for a starter set the organisation asked for, the themes the model proposed and the list of short conversation-topic phrases with their call categories that was sent to it; and the weekly suggestions of new themes — a name and a short description written by the model (not stored where the same check finds such data), the number and share of conversations, the suggested-theme phrases the suggestion was grouped from, and the identifiers of up to three example conversations | The set of themes and the record of its changes: until the account is deleted (a theme is archived, not deleted). The themes and the suggested theme of a conversation: follow the record — deleted together with it and its analysis. The phrase list of a starter set: until the set is approved or a new one is requested, and at most 30 days; the proposed themes stay until a new starter set replaces them. A weekly suggestion: at most 30 days after the last week in which it reached the threshold; a rejected one 90 days after the decision, without its phrases, which are erased when it is accepted or rejected; an example is removed when that conversation is deleted. The phrases of a starter set or a suggestion are not linked to a record, so a phrase from a conversation deleted in the meantime remains until these periods end |
| Deal states and CRM deal snapshots | For an organisation with the deal state (Pro plan and above), after each analysed call or chat linked to a deal or lead in the connected CRM (Bitrix24 or Uspacy), the accumulated state of that deal or lead, written by the model from its previous state and the new conversation: a status (on track, needs attention or at risk), a chance of closing, what the conversation changed, the reason for the status in the model's words with the customer's verbatim quote taken from the PII-redacted transcript and, for a call, the second at which it was said, the next action from a fixed list with its due date and reason, points for the next contact, commitments (who made them, what, by when and whether they are open, done or overdue), objections (reason from a fixed list, whether open or resolved, the customer's last verbatim quote), what is known of the budget, the decision maker, the need and the timeline with a short fact for each, the customer's interest, the sales methodology applied, and identifiers of the conversations the items came from. For a Bitrix24 or Uspacy deal converted from a lead, its states also keep the identifier of that lead, and, when the lead has a state in the same project, the deal's first state continues the lead's state as of the deal's first conversation. A new state is added after each such conversation and the earlier ones are kept, so the states form the deal's history. For a Bitrix24 deal or lead, a snapshot read from Bitrix24 at that moment is stored with the state: its title (which can contain a person's or company's name), amount and currency, pipeline and stage, whether the stage is final, and the identifier of its responsible employee in Bitrix24, with the time it was read; For a Uspacy deal or lead, only the identifier of its responsible employee in Uspacy is stored with the state, and only while the organisation’s Uspacy processing is active; no title, amount or other field is read. Also stored: whether the comment about the state was posted to the CRM, and whether a to-do was created from it, with the to-do's identifier in the CRM. Phone numbers, emails, card numbers and IBANs in the state's texts are masked | Customer's retention period, counted from the earliest conversation whose facts the state may carry: a state is deleted once that conversation falls outside the period (the next conversation then starts the deal's state anew), when its project is deleted, or with the account. A deal state is not deleted when a call or chat it was built from is deleted — its text, quotes and links to that record stay as the employee saw them, until the state itself reaches the end of the retention period. Comments and to-dos already written to the Customer's CRM are under the Customer's control there and are not removed by Aelo |
| Voice-delivery metrics | Objective acoustic measurements derived from call audio — speaking rate, pause patterns, pitch variation, and voice stability. Not voiceprints and not used to identify individuals. | Follows transcript retention |
| Communication metadata | Phone numbers, call duration, timestamps, caller IDs | Follows transcript retention |
| Alert data | Notifications triggered by analysis thresholds. Always shown in-app; additionally delivered by email and the recipient's configured personal channels. Personal email and webhook messages may include a masked caller reference — first name with surname initial, or the last 3 digits of the phone number — and an AI-generated reason when present. Personal Telegram receives only business alerts, in a contact-free form with neither the caller reference nor the AI-generated reason; no other notification type is sent to Telegram. Project team channels receive a reduced summary — rule, severity, call category and record link — with no identity for either the other party to the call or the platform user associated with the record, no phone number, transcript excerpt or AI-generated reason. | An in-app notification is deleted 180 days after it was created, together with every user’s status of it (whether it was delivered, read or archived); earlier, together with the record it concerns (including at the end of the Customer’s retention period) or when the Customer’s account is deleted |
| Notification delivery log | For each attempt to deliver a notification through a channel (for example email, Telegram or a webhook): the channel, whether it was sent or failed, the time, the error message if it failed, and the identifier of the recipient user and technical details of the delivery, such as the email provider’s message identifier | Deleted 30 days after the attempt, whatever retention period the Customer sets |
| Ingest activity log | For each incoming webhook, the decision on where to route it — routed, rejected, skipped or error — with its reason, the Customer’s routing rule that matched, the external identifier of the call or chat and technical details of the decision, such as its source; the Customer sees it in Analysis Setup → Distribution → Log | Deleted 90 days after the entry was written, whatever retention period the Customer sets; the entry of a call or chat still waiting to be processed is kept until its receipt is processed |
| Ingest receipts | For each incoming call or chat taken in for processing, a receipt used to recognise a repeated delivery: the external identifier, the source vendor and installation, a hash of the content and the identifier of the record created from it, and technical identifiers of the delivery (version timestamp, idempotency key); for a call whose recording and details Aelo cannot fetch again from its source by the call’s identifier (such as one from Uspacy), also, while the call is waiting to be processed, the details that came with the call: the address of the recording, the employee’s identifier in the source, the other participant’s phone number, the call’s direction, duration and subject, its link to a CRM deal, lead, contact or company, and whether it was missed or ended without a recording | A processed receipt is deleted 30 days after it was processed; a receipt still waiting to be processed is kept until it is, and the call details are removed from it when it is processed. Afterwards a repeated delivery is recognised by the record itself, while it exists |
| Emailed reports and digests | Email addresses the Customer configures for a project to receive analysis reports (which need not belong to Aelo users), and the reports, notifications and digests emailed to them and to the Customer's users — see A.3 and, for what each email carries, the Sub-Processor List | Recipient addresses: until the Customer changes them or the account is closed. Emails already sent are outside Aelo's control: copies stay in the recipients' mailboxes and in the email provider's delivery logs, and deleting a record does not recall an email sent about it |
| CRM deal-stage data | When the Customer connects Bitrix24 or Uspacy and enables stage-based evaluation, the linked deal's CRM stage ID, pipeline/funnel identifier, and stage phase (e.g. discovery, negotiation) — captured at analysis time to select the matching evaluation scorecard. Where the CRM gives the stage no phase (in Uspacy only a funnel's closing success or fail stage gives one), Aelo uses the phase the Customer set for that stage or, failing that, the phase determined by AI from the names of the Customer's pipelines and stages (see A.3), and marks it as derived. For each such stage Aelo keeps the determined phase, the model that determined it and a hash of the stage’s name — not the name — until it is replaced or the account is deleted (the row of a stage removed from the CRM stays until then). Applies only to calls linked to a Bitrix24 or Uspacy deal. For a Uspacy deal Aelo reads the deal's current stage identifier and the portal's list of deal funnels with their stages, only while the Customer's Uspacy processing is active. For a Bitrix24 deal, additionally, once per call, Aelo reads the deal's stage history to determine whether the deal advanced, held, or regressed in the ~7 days after the call (or by deal close), storing that per-call result to show a progression indicator in the Customer's dashboard. Uspacy keeps no stage history, so Aelo keeps its own: at each scheduled re-check (see A.3) it stores the deal's pipeline/funnel and stage identifiers with the times it first and last saw the deal on that stage, and for a Uspacy deal the same per-call result compares the stage at the call with the stage the deal was seen on when the ~7-day window after the call ended; a call whose window ended at a time Aelo had not seen the deal at (before stage-based evaluation was switched on, or between two re-checks) gets no result. | Follows transcript retention. The stage records kept for a Uspacy deal: deleted 90 days after the deal was last seen on that stage, and when the Customer’s account is deleted |
| CRM org-structure data | When the Customer connects Bitrix24 or Uspacy and runs the org-structure import (Settings → Team → Org Structure), Aelo reads the portal's department tree and employee directory and stores: department identifiers, names, parent-department links and sort order; which employees belong to which departments; and the department head named by the portal. Employees are matched to existing Aelo accounts by email address, and the resulting pairing of the portal's user ID with the Aelo account is stored. For Uspacy, the import runs only while the organisation’s Uspacy processing is active, and a match by email address makes the employee a member of their departments, with the portal's identifier stored on the membership, but stores no such pairing: it is made only by the employee, from the portal, and a Uspacy department head receives access to the department's records only once it exists, unless the organisation’s owner assigns a head. For portal employees with no matching Aelo account, Aelo stores the portal's own identifier for that employee, their email address when the portal provides one, their name, and their department membership — shown to the organisation’s owner by name so the gap can be resolved — until an Aelo account for them is created (when an invitation is sent to them, or through their first sign-in inside the connected CRM's embedded app), at which point Aelo links the two automatically and deletes this interim record. The same applies to a department head the portal names who has no Aelo account yet: the head's portal identifier and name are stored on the department. If Aelo can identify that head's new account using the portal's own identifier, access to the department's records is granted automatically, unless the organisation’s owner has manually assigned a different head, which always takes precedence and survives later imports. This data determines which call records a department head may open — a head sees the records of everyone below them in the tree. The import replaces the previously imported tree in full. | Duration of the Customer's account (cascade-deleted when the account is deleted); the tree is replaced in full on each import, and an individual not-yet-matched employee's record is deleted sooner, as soon as an Aelo account for them is linked |
| CRM deal-outcome data | When the Customer connects Bitrix24 or Uspacy and enables stage-based evaluation, the linked deal's outcome status (open/won/lost), pipeline/stage identifiers, and close and sync timestamps — synced periodically from the CRM to show a deal-outcome funnel (call volume by won/lost, broken down by average call-quality score) in the Customer's dashboard. Applies only to deals in a connected Bitrix24 pipeline or Uspacy funnel; a Uspacy deal is won or lost on its funnel's closing success or fail stage and open on any other, and is read only while the Customer's Uspacy processing is active. If the organisation also has the Voice of Customer section with its customer questions and decline reasons (Pro plan and above), a lost deal also carries the code of the reason the customer gave in the most recent conversation of the deal's current loss cycle that ended in a decline (a code from a fixed list, without the customer's words or quote), with the version of the code list, the identifier of that conversation and the time it took place; only conversations held after this was introduced are used. | Duration of the Customer's account (cascade-deleted when the account is deleted). The decline-reason code is kept only while the deal stays lost and the conversation it was taken from exists: it is removed when that conversation is deleted, including at the end of the retention period, and when Aelo sees the deal reopened or won; it is replaced or removed when a newer conversation of the same loss cycle ends in a decline or when that conversation's analysis no longer confirms the reason. While stage-based evaluation is switched off, Aelo does not re-check the deal, and the code stays until that conversation is deleted |
| Assistant conversations | Questions a Customer's user asks the AI assistant about a call or chat record, or about a CRM deal, lead, contact or company across the records linked to it, or about all the records the user may see; for a question about all records, the filters it was asked under (period, project, agent, record type, category, customer name); the generated answers with the transcript excerpts and timecodes they cite; the list of records each answer was built from; and the identity of the asking user. Questions receive the same pattern-based PII redaction as transcripts and are checked by the LLM provider for personal data it missed before they are stored (see A.3), and only the stored text is sent to the LLM provider. | Customer's retention period (365 days by default, configurable 1–365); a question and its answer are also deleted when the record they were asked about, or any record the answer was built from, is deleted — usually together with that record, otherwise by a routine scheduled cleanup |
| Coaching feedback | Feedback the Customer’s supervisors, administrators and owner leave about its employees’ work: notes about an employee (up to 2,000 characters, optionally naming one of the employee’s calls), coaching sessions scheduled with an employee (the scheduled time, a status and notes of up to 2,000 characters, optionally naming a call and, for a session scheduled from an alert, that alert), annotations on a passage of a call or chat transcript (a type, a severity and a comment of up to 5,000 characters; an employee may also annotate their own conversation) and a review of a conversation’s analysis (a status, the points marked and an overall comment of up to 2,000 characters), each with its author and time. Annotations and reviews are shown to the Customer’s users who may see that conversation’s analysis, including the employee whose conversation it is; notes are shown to the Customer’s supervisors, administrators and owner and to the employee they are about; a coaching session is shown to the user who scheduled it and to the employee it is with. Not sent to AI models. The employee is notified of new feedback as described in A.3 | Annotations and reviews: follow the record — deleted together with it, including at the end of the retention period. Notes: follow the retention period — a routine scheduled cleanup deletes a note once it is older than the Customer’s retention period, counted from when it was written; a note is deleted earlier when the Customer’s account is deleted, when its author’s user account is deleted or, for a note about an employee with an Aelo account, when that account is deleted; when the call a note names is deleted, the note stays without the link until its own period ends. Coaching sessions: follow the retention period — a routine scheduled cleanup deletes a session once both its creation and its scheduled time are older than the Customer’s retention period, so a session scheduled for a future date is never deleted before it takes place (a session can be scheduled at most 365 days ahead); a session is deleted earlier when the Customer’s account, the user account of the person who scheduled it or the Aelo account of the employee it is with is deleted; when the call or the alert a session names is deleted, the session stays without the link until its own period ends |
A.3 How We Process This Data
- Customer supplies audio — uploaded through the app, sent through the public API, or received via webhook from connected telephony/CRM
- Alternatively, on the Customer's instruction, Aelo retrieves the audio itself: the Customer supplies a URL through the public API, and Aelo opens an outbound HTTPS connection to that address and downloads the file. The Customer designates the host; Aelo resolves the hostname through Cloudflare's DNS-over-HTTPS resolver (1.1.1.1) before connecting, and stores the downloaded file as in the next step
- Aelo validates and stores audio (encrypted, Cloudflare R2 with AES-256-GCM server-side encryption)
- Audio sent to the ASR provider (ElevenLabs) for transcription
- PII redaction applied to the transcript (phone numbers, emails, payment card numbers, IBANs) on the current application repository write paths — before it is stored; see the next step for the scope and limits of this control
- For new transcripts processed through the current application repository write paths, the redacted transcript is stored in Neon PostgreSQL and the unredacted transcript text is not retained through those paths. This control was not applied retroactively: earlier transcript, preview, metadata and word-timestamp rows were not rewritten and may retain the text originally written. All such rows remain subject to the Customer's configured retention period. Original audio is not text-redacted and is retained as described in A.2.
- If the Customer has connected Bitrix24 or Uspacy and enabled stage-based evaluation, and the call is linked to a Bitrix24 or Uspacy deal, Aelo reads the deal's current CRM stage (stage, pipeline, and phase — where the CRM gives none, the phase the Customer set for the stage, else the phase determined by AI from the names of the Customer's pipelines and stages (see A.3)) to select the evaluation scorecard used in the next step
- Transcript (with PII redaction applied) sent to OpenAI for analysis
- If the Customer has defined criteria, or built-in criteria worded by Aelo apply to the conversation, the redacted transcript is sent to OpenAI again, in a separate request, together with the wording of the criteria that apply to it (question, guidance and answer values — the Customer's as the Customer wrote them, at most with phone numbers, emails, card numbers and IBANs masked; built-in criteria as Aelo worded them, with the Customer's jurisdiction where it applies), to answer them; the answers and their quotes are stored (see A.2). Aelo does not charge for built-in criteria, and they are not asked for a short call that receives no analysis, or when an already analysed conversation is re-analysed (a failed analysis that is retried asks them as usual); their answers are shown to the Customer in the call panel and the PDF report, the dashboard, the attention flag, alert rules and the report written to the Customer's CRM, and these checks of what was or was not said are made only through them, because the analysis no longer performs them (where Aelo has switched built-in criteria off for the Customer, they are not shown and the Customer has none of these checks). When the Customer checks a criterion's wording before saving it, the same is done for up to 20 recent analysed conversations within the criterion's scope with the draft wording; those answers are shown to the Customer and not stored. When a user describes a criterion in their own words, that description is sent to OpenAI (at most with phone numbers, emails, card numbers and IBANs masked) to write a draft wording; and when a user marks answers of such a check as wrong, the draft wording, the marked answers with the user’s optional reasons and, for each marked conversation (up to 10), a fragment of its transcript around the quoted lines, redacted again before it is sent, are sent to OpenAI to propose an edit of the wording. The draft and the proposed edit are shown to that user and are not stored until the user saves the criterion
- If the Customer's organisation has the Voice of Customer section with its customer questions and decline reasons (Pro plan and above), the redacted transcript is sent to OpenAI again, in a separate request, to find the questions the customer asked and, for a sales conversation, how it ended and the reason the customer declined; the results and their quotes are stored (see A.2). Where the organisation also has conversation themes and its set has an active theme, the same request carries the names and descriptions of those themes (as the Customer wrote them, at most with phone numbers, emails, card numbers and IBANs masked) and also returns one to three themes of the set for the conversation, or that none fits, and, where the conversation has a subject the set lacks, a short suggested theme; they are stored with the results. Conversations analysed before this was introduced, or before a theme entered the set, are not processed in bulk; a conversation the Customer re-analyses is processed like a new one. Aelo does not charge for it
- When a Customer's user asks for a draft FAQ in the Voice of Customer section (Pro plan and above), the most frequent subjects of the questions customers asked in the period and projects the user selected, with up to 40 of the stored one-sentence summaries of those questions (see A.2) and the number of conversations each subject was asked in, are sent to OpenAI in a separate request to write a draft FAQ — questions and answer templates for the Customer to complete — in the language set for the Customer's organisation. Before they are sent, the summaries pass the same masking of phone numbers, emails, card numbers and IBANs again, and a summary that may contain an identification number is left out; no transcript, quote, record or person is sent. The draft is shown to the user who asked for it and is not stored. It is generated only on that request, at most 10 times per organisation and 5 times per user in 24 hours; Aelo does not charge for it
- When the owner or an administrator of the Customer's organisation asks for a starter set of conversation themes (Pro plan and above), the short topic phrases the analysis stored for the organisation's conversations of the last 90 days, each with the call category of the conversation — as a list of distinct phrase and category pairs, up to 1,500 of them, without demo conversations — are sent to OpenAI in a separate request to propose up to 25 themes with short descriptions. Before they are sent, the phrases are checked again for phone numbers, emails, card numbers and IBANs, and a phrase in which the check finds one, or that may contain an identification number, is left out; no transcript, quote, record or person is sent. A name or description the model returns in which the same check finds such data is not stored. The proposal is shown to the organisation's owner and administrators, who edit and approve it; only the themes they approve enter the set. The proposal and the list of phrases that was sent are stored as described in A.2. It is generated only on that request — a new one at most once in 24 hours per organisation, and after a failed attempt not sooner than an hour later; Aelo does not charge for it
- Once a week, for an organisation with conversation themes whose set has an active theme and that had newly analysed conversations in the last 7 days, the suggested-theme phrases stored for its conversations of the last 30 days (see A.2) — up to 300 distinct phrases, without demo conversations — are sent to OpenAI in a separate request to group them into at most five candidate themes, each with a name and a short description. Before they are sent, the phrases are checked again for phone numbers, emails, card numbers and IBANs, and a phrase in which the check finds one, or that may contain an identification number, is left out; no transcript, quote, record or person is sent. A name or description the model returns in which the same check finds such data is not stored. A candidate that came up in at least 10 conversations and 3% of the period's conversations is shown to the organisation's owner and administrators, with its counts and up to three example conversations, to accept into the set or reject; it is stored as described in A.2. It is not done while the organisation's analysis is paused, or, for an organisation connected to Uspacy, while its Uspacy processing is not active. Aelo does not charge for it
- If the Customer's organisation has the deal state (Pro plan and above) and a newly analysed call or chat is linked to a deal or lead in the connected CRM, Aelo builds the updated state of that deal or lead (see A.2). For a Bitrix24 deal or lead, Aelo first reads it from Bitrix24 — its title, amount, currency, pipeline, stage, whether the stage is final and its responsible employee, and, for a deal, the lead it was converted from — and stores all but the lead as the snapshot. For a Uspacy deal Aelo reads from Uspacy the deal's current stage identifier and the portal's list of deal funnels, to find the deal's funnel, and the leads linked to the deal, of which it uses only their identifiers and whether Uspacy marks a lead as converted, to find the lead the deal was converted from; for a Uspacy deal or lead it also reads the identifier of its responsible employee, while the organisation’s Uspacy processing is active, and stores that identifier with the state — no other snapshot is stored with the state (the stage records described under CRM deal-stage data are kept separately). Then the transcript of the conversation (the stored, redacted transcript, with the same PII redaction applied again before it is sent), a summary of its stored analysis, the previous state of the deal or lead (for a Bitrix24 or Uspacy deal that has no state of its own yet, converted from a lead that has one in the same project, the lead's state as of the deal's first conversation), the sales methodology set for the pipeline of a Bitrix24 deal or the funnel of a Uspacy deal with the Customer's company rules for it (at most with phone numbers, emails, card numbers and IBANs masked; a lead, and a deal whose pipeline Aelo does not read, get the standard methodology without company rules), the number of days since the previous conversation and, for a Bitrix24 deal or lead, its stage, amount and currency, and, for a Uspacy deal, its stage identifier are sent to OpenAI in a separate request to write the updated state; the title of the deal or lead and its responsible employee are not sent. A quote is kept only where it matches the transcript word for word. Aelo does not charge for it; it is not done for a short call that receives no analysis or when an already analysed conversation is re-analysed, and conversations analysed before the deal state was available to the organisation are not processed in bulk
- Where the Customer has connected Bitrix24 or Uspacy and switched on stage-based evaluation, the names of its deal pipelines (in Uspacy, funnels) and of their stages (not of the stages the Customer has given a phase itself), in their order and with the CRM’s own mark of the won and lost stages (in Uspacy, a funnel’s closing success or fail stage), are sent to OpenAI in a separate request to determine the sales phase (discovery, qualification, proposal, negotiation, closing or post-sale) of each stage that neither the CRM nor the Customer has given a phase; no customer, deal or conversation data is sent. This is done only while stage-based evaluation is switched on: when a call at such a stage is analysed, when a user opens the deal stages settings or clears the phase of a stage there, and when Aelo gives a phase to calls recorded before stages had one; other screens that show the stages use the phases already determined and send nothing. It is asked when such a stage is new or has been renamed, or an earlier attempt gave no answer for it; the request then contains the names of all the organisation’s pipelines and stages. The phase of each stage is stored with a hash of the stage’s name, not the name (see A.2); a phase the Customer sets for a stage always takes precedence. Aelo does not charge for it
- For search by meaning, segments of the transcript — or of the chat conversation — and the text a user types into search are sent to OpenAI to compute numerical vector representations (embeddings). Both receive the same redaction before they are sent: the pattern-based PII redaction (phone numbers, emails, payment card numbers, IBANs) and, in addition, a contextual masking of a passport or other identity-document number, a tax number, a date of birth, bank account details and an address when a label names them (such as “passport”, “date of birth” or “address”; the label of an address must be followed by a colon, a dash or “is”, and such an address is also masked where it continues after a line break within one turn), and of an address written as a street word from a fixed list (“street”, “avenue”, “calle”, “avenida” and the like, in English, Spanish, Ukrainian and Russian), the street’s name, a comma and a house number (such as “calle Mayor, 7” or “avenida Diagonal, 12”). An address or a date of birth without such a label or form is not detected (for example “my address 123 Broadway”, with no colon, dash or “is” after the label, or “Hauptstraße 12”, with no comma), and names are not removed. The masking applies to the text sent to compute the vectors and to the text shown next to a search result; the stored transcript keeps the redaction described above. Search vectors computed before release 26.50.0 were computed after the pattern-based redaction only and are replaced when the record is transcribed again and its vectors are built, or when the index is rebuilt. Only the vectors of the segments are stored, in Cloudflare Vectorize, with the identifiers listed in A.2 and without the text; the vector of a search query is used for that search only and is not stored
- When a Customer's user asks the AI assistant a question about a call, the question (with the same PII redaction applied) is sent to OpenAI together with the redacted transcript and the stored analysis of that record to generate the answer; the question, the answer and the excerpts it cites are stored as an assistant conversation (see A.2)
- Before a question to the AI assistant is stored, it is sent to OpenAI on its own in a separate request to find personal data the pattern-based redaction missed, and what it finds is masked; the question is stored, kept in the conversation’s history and sent to generate the answer only as masked, and a question whose check fails is not accepted
- Before any assistant answer is generated, the redacted transcripts and analyses it will be generated from (and, for a question about a CRM deal, lead, contact or company, the entity's name and the one-line index of its records), and the earlier questions and answers of the same conversation sent with the question, are sent to OpenAI in a separate request to find personal data the pattern-based redaction missed — such as an address, a date of birth, an identity or tax number, card or account details, a phone number or an email, also where it is split between turns or spelled out in words; what it finds is masked before the answer is requested. The generated answer, with the excerpts it cites and its suggested follow-up questions, is sent to OpenAI the same way, together with the question, before it is stored and shown. The checks are made with the same OpenAI account as the answer; a record whose check fails is left out of the answer, earlier questions and answers whose check fails are left out of the request, and when the check of the answer or of every record fails, no answer is given
- When a Customer's user asks the AI assistant a question about all the records they may see, the question (with the redaction a search query receives, described above) is sent to OpenAI's embeddings endpoint to find the closest records by meaning in the search index, and the redacted transcripts and analyses of up to 12 of the closest records the user may see, within the filters they chose, are sent to OpenAI to generate the answer, together with the filters the question was asked under (period, record type, category and customer name); the question's vector is not stored, and when no record matches, no answer is generated by the model
- When the question is about a CRM deal, lead, contact or company, what is sent to OpenAI is drawn only from the records linked to it that this user is permitted to see (up to the 2,000 most recent): the name of the CRM entity (with the same PII redaction applied) and, for a deal, its outcome in the CRM where known (won, lost or still open); a one-line index of those records with their date, type and a short summary of their stored analysis; and, within a fixed size limit and most relevant to the question first, their stored analyses and redacted transcripts. The conversation is stored the same way (see A.2)
- To evaluate the quality of the analysis, Aelo measures how accurately it answers on samples of conversations: the stored redacted transcripts of a sample are sent to a language model that labels them without seeing the analysis’s answers (blind labelling) or checks the stored analysis against its transcript, and the analysis a candidate model or wording produces on them is compared with the current one. Depending on the measurement, the stored analysis or answers of those conversations and the wording of the criterion or instruction being measured are sent too. These requests go to OpenAI and, from 2026-12-03, to Anthropic (see A.4); the redaction does not remove names. The results are used to decide changes to the analysis, are not shown to the Customer and are not used to make decisions about the people in the conversations.
- Call audio is additionally processed by Aelo's self-hosted acoustic service (hosted on Hetzner infrastructure in Germany, EU) to derive objective voice-delivery metrics (speaking rate, pauses, pitch variation, voice stability); the audio is processed transiently and not retained by that service
- Analysis results stored and delivered to Customer's CRM. Where the Customer has connected a CRM, Aelo also delivers the call transcript there: attached to the call record when the Customer switches that on; posted to the timeline of the call item when delivery targets a Bitrix24 Smart Process, alongside the analysis report; and written into a CRM field the Customer designates only if the Customer switches that on (off by default). Each carries the transcript as stored by Aelo, with the PII redaction described in A.3 already applied; the field copy is truncated when it exceeds the field's capacity
- If the Customer's organisation has the deal state, after each new state Aelo posts a separate comment about it to the timeline of the deal or lead in the Customer's CRM — the CRM that receives the Customer's call reports: Bitrix24, or Uspacy while the organisation's Uspacy processing is active. The comment carries the date of the conversation and, for a call, its duration; the status and chance of closing and what they were before; what changed; the reason for the status with the customer's verbatim quote and a link to the conversation in Aelo; the next action with its due date and reason; points for the next contact and the methodology; the commitments without who made them; the objections by their standard names; which of the budget, decision maker, need and timeline are known, without the facts; the customer's interest; and a link to the deal or lead in Aelo. It is posted only for a conversation held within the last 7 days and not for a conversation analysed after a newer one; the Customer can switch these comments off (they are on by default). When a Customer's user asks for it, Aelo creates a to-do in the Bitrix24 deal or lead, assigned to its responsible employee, with the next action as its title, its reason and, except when it is created from the Aelo tab in the Bitrix24 card, a link to the deal or lead in Aelo as its description, and its due date; in Uspacy, while the organisation's Uspacy processing is active, Aelo adds the same as one planned task (a CRM activity) to the deal or lead, assigned to its responsible employee where the latest state knows one and otherwise without a responsible employee set by Aelo (Uspacy then assigns it by its own default), and reads the activities of that deal or lead only to find a task already added for the same state
- When a Customer's user who may see the texts of a deal's state asks for a draft message to the customer in the deal card (Pro plan and above, for an organisation with the deal state), Aelo sends to OpenAI, in a separate request, the current stored state of that deal or lead — its status, chance of closing, the reason for the status, the next action with its due date and reason, points for the next contact, what the latest conversation changed, commitments (who made them, what, by when and whether they are open, done or overdue), objections (the reason code and whether it is open or resolved), what is known of budget, decision maker, need and timeline, and the customer's interest — with those texts masked again for phone numbers, emails, card numbers and IBANs, and the sales methodology set for the pipeline of a Bitrix24 deal or the funnel of a Uspacy deal (the standard methodology and, for a deal, the Customer's company rules for it, masked again the same way), to write one short follow-up message to the customer in the analysis language of the deal's project. The customer's quotes, the identifiers of conversations and records, the title of the deal or lead, its amount, stage and responsible employee are not sent, and neither is any transcript; the model's reply is masked again before it is shown. The draft is shown to the user who asked for it, who can edit and copy it; it is not stored and is not written to the CRM. It is generated only on that request, at most 5 times per minute, 60 times per hour and 200 times in 24 hours per organisation, and 3, 30 and 60 times in the same periods per user (a request is refused when these counters cannot be read). Aelo does not charge for it
- Aelo emails analysis reports (the record's quality metrics) to the addresses the Customer configures for a project, which need not belong to Aelo users. Daily or weekly role digests and a monthly Voice of Customer email go to the Customer's users by default, and each user can switch them off; a supervisor's digest names employees with their quality scores. These emails are sent through Resend (A.4); what each one carries is listed in the Sub-Processor List. Emails never carry transcripts or audio
- While coaching is switched on for the Customer’s organisation (it is by default), when a supervisor, an administrator or the owner leaves feedback about an employee — a note, an annotation other than a correction of the AI analysis, or an overall comment added to a review that had none — Aelo notifies that employee in the app and by email through Resend (A.4), unless the employee has switched these emails off in their notification settings; and, when the employee’s Aelo account is linked to their user in the Customer’s Bitrix24 portal and the organisation’s Bitrix24 installation is active, by a Bitrix24 notification to that user in the portal, which the email switch does not stop. The notification, the email and the Bitrix24 notification carry the author’s name, the kind of feedback and, for a call or chat, a link to it (in the Bitrix24 notification, a link that opens Aelo inside the portal) and, for a call, the moment the feedback is about; never the feedback text and no data of the other party to the conversation. Several pieces of feedback from the same author on the same conversation within 30 minutes send one email and one Bitrix24 notification, the rest are notified in the app only; feedback left by Aelo staff through platform access notifies no one
- If the analysis triggers a business alert rule, Aelo delivers a notification: always in-app; additionally by the channels the recipient or organization configured. Personal email and webhook messages may include a masked caller reference and the model's short written reason. Personal Telegram receives only business alerts, with neither field; no other notification type is sent to Telegram. Project team channels receive a reduced message containing rule, severity, call category and record link, with no identity for either the other party to the call or the platform user associated with the record, no transcript excerpt or model-generated reason
- On a recurring schedule, if the Customer has connected Bitrix24 or Uspacy and enabled stage-based evaluation, Aelo re-checks each linked deal's status in the CRM (pipeline or funnel, stage, and won/lost outcome) to keep the deal-outcome funnel in the Customer's dashboard current, and determines whether the deal advanced, held, or regressed after the call — from the deal's stage history in Bitrix24, and in Uspacy, which keeps no stage history, from the stages Aelo recorded at its own re-checks (see A.2) — storing that per-call result to display a progression indicator alongside the call. For Uspacy this happens only while the Customer’s Uspacy processing is active, and only the deal’s stage identifier and the portal’s deal funnels are read. If the Customer's organisation also has the Voice of Customer section with its customer questions and decline reasons (Pro plan and above), Aelo copies onto each lost deal the code of the reason the customer gave in the deal's most recent conversation of the current loss cycle that ended in a decline (see A.2); the code is taken from the stored decline reason, and no further request to the AI model is made for it
- Data retained for the Customer's configured retention period — 365 days by default, adjustable by the Customer between 1 and 365 days (see A.2) — then automatically deleted, except the data for which A.2 states another period and the action log, which B.1 keeps for 365 days whatever period the Customer sets
A.4 Sub-Processors
We use third-party sub-processors to provide the Service. The complete list is maintained at our Sub-Processor List.
| Provider | Purpose | Location |
|---|---|---|
| ElevenLabs | Speech-to-text transcription | US |
| OpenAI | AI analysis of transcripts, answers to criteria (the Customer's own and those worded by Aelo) and to assistant questions with a check of their input and answers for personal data the pattern-based redaction missed, a draft criterion wording from a user’s description and a proposed edit of it from the answers the user marked as wrong, extraction of customer questions and decline reasons, a draft FAQ from those questions on request, the conversation themes of an organisation (assigning them to conversations, a starter set on request and weekly suggestions of new themes), the state of a CRM deal or lead after each conversation linked to it, a draft message to the customer from that state on request, the phase of CRM deal stages from their names, and search embeddings (transcripts pattern-redacted; Customer configuration and names sent without anonymisation — at most phone numbers, emails, card numbers and IBANs in them are masked; see the Sub-Processor List); the evaluation of the quality of the analysis on samples of conversations (A.3) | US |
| Anthropic | Evaluation of the quality of the analysis on samples of conversations (A.3), from 2026-12-03; also authorised for the analysis of conversations, which is currently made with OpenAI | US |
| Neon | Database storage | EU (Frankfurt) at rest; US vendor access |
| Cloudflare | Infrastructure, audio storage, search index (vectors and identifiers, no transcript text) | EU at rest (audio in Western Europe, metadata in Eastern Europe); Cloudflare Workers are not pinned to the EU: a request is received at the Cloudflare edge location nearest to it; background processing (queues, scheduled tasks, Workflows), which carries most of the processing of calls and chats, runs at locations Cloudflare chooses, mostly outside the EU; and from release 26.58.0 the Aelo API and the request handling of the processing service are asked to run in a Cloudflare location near Frankfurt (Germany), which Cloudflare chooses and does not guarantee, while the Worker that serves the app and passes its requests to the API, and the other Workers, are not placed, so processing may occur outside the EU; the search index region is not pinned; Workflows state, Queues, Workers Logs/Traces, Workers Analytics Engine, KV and Durable Objects are not pinned to the EU — Workflows state and Queues hold transcript and chat text while a record is processed and for a few days after (see the Sub-Processor List) |
| Hetzner | Self-hosted acoustic voice-metrics service (audio processing) | EU (Germany) |
| Clerk | Platform authentication | US |
| Resend | Email delivery of analysis reports, alerts, record notifications and digests — see the Sub-Processor List for what each email carries; no transcripts or audio | US |
| SendPulse | Marketing email automation (onboarding series and the usage signals that steer it) — Aelo's own, as controller; everything is sent only for a person who gave separate marketing consent (see B.5); no call content | EU (Germany) — endpoint verified; storage region not contractually confirmed |
| Langfuse | LLM observability (optional; traces only, no transcripts) | EU |
| Sentry | Error tracking, performance tracing, and session replay for the web application and API — not the call-processing pipeline. Receives the signed-in platform user's name, email, and account identifier; the connecting IP address at the network level; page/API URLs; breadcrumbs of recent navigation and network activity; error text, which can incidentally include data being processed at the time of the error; session replay recordings for a sample of sessions (~10%) and for every session in which an error occurs, with on-screen content masked. See the Sub-Processor List for the full breakdown. | EU (Germany) — storage; US vendor access |
| Telegram | Opt-in alert delivery through Telegram's Bot API. Personal and team Telegram messages contain no identity for either the other party to the call or the Aelo user associated with the record, no phone number in any form, no transcript excerpt and no model-generated reason. Only business-alert messages are sent to Telegram; other notification types are not. No DPA or SCCs are offered by Telegram for the Bot API — see A.5 and the Sub-Processor List | British Virgin Islands (group: BVI + UAE) |
Customers are notified at least 30 days before any new sub-processor is added. The complete, authoritative list (with transfer mechanisms) is maintained at our Sub-Processor List.
A.5 International Data Transfers
Auspex Streamline S.L. is established in Spain (EU). Some sub-processors are located in the United States.
For transfers of personal data outside the EU/EEA, we rely on:
- EU-US Data Privacy Framework (DPF) — where the sub-processor is DPF-certified
- Standard Contractual Clauses (SCCs) — the 2021 version approved by the European Commission
- Transfer Impact Assessments (TIAs) — conducted for each transfer
One exception, stated plainly: Telegram alert delivery. Telegram Messenger Inc. is incorporated in the British Virgin Islands, publishes no data processing agreement for its Bot API and offers bot operators no Standard Contractual Clauses; neither the British Virgin Islands nor the UAE holds a European Commission adequacy decision. None of the three mechanisms above is therefore available for this one destination. The channel is off by default and sends nothing until you or one of your users explicitly connects a chat; what is sent is reduced and limited to business alerts (see the Sub-Processor List for the exact fields). If your organisation requires a contractual safeguard for every recipient, do not enable it — the other delivery channels are unaffected.
A.6 Security Measures
| Category | Measures |
|---|---|
| Encryption | TLS 1.2+ in transit; AES-256-GCM server-side encryption at rest (R2 SSE active by default); database encryption (Neon native); provider API credentials (speech recognition, AI analysis, email delivery) and the credentials of connected telephony and CRM integrations additionally sealed by the application with AES-256-GCM and bound to their own record, so a database read alone does not yield a usable key |
| Access control | Role-based access (RBAC); organization-level tenant isolation |
| Data minimization | For new transcripts processed through current application repository paths, pattern-based PII redaction (phone, email, payment card, valid IBAN) is applied on write and before LLM processing. Original audio is unchanged; word-level timestamps are not retained for records processed after this control was introduced, while earlier transcript, preview, metadata and word-timestamp rows were not migrated; Customer-configured retention (365 days by default, 1–365 range) |
| Integrity | Webhook authentication: per-installation secret, HMAC signature or sender IP allowlist, depending on the provider; Zod input validation; SQL injection prevention (ORM) |
| Availability | Cloudflare edge network; database replication; idempotent processing |
| Monitoring | Production logs capture denied authorization attempts and selected security-relevant events. Coverage does not extend to a customer-facing history of every action. |
| Authentication | Interactive access: multi-factor authentication via Clerk. Programmatic access (public API): bearer API keys issued per organization, which do not use Clerk or multi-factor authentication — a key is stored only as a SHA-256 hash (the plaintext is shown once at creation and never retained), is valid solely for the organization that issued it, carries separate read and write scopes, may be given an expiry, and takes effect as revoked on the next request after revocation |
A.7 Data Subject Rights
Call participants (Data Subjects) should contact the Customer (Controller) to exercise their rights. The Customer is responsible for verifying the identity of the Data Subject, recording and reviewing the request in Aelo where applicable, fulfilling the approved request through the available product or support procedure, and responding within the legally required timeframe. Aelo's DSR workflow tracks requests and decisions but does not yet execute approved requests automatically.
| Right | GDPR Article | How Aelo supports the request |
|---|---|---|
| Right of access | Art. 15 | Request intake, review and status tracking; approved request fulfilled through available export/support procedures |
| Right to erasure | Art. 17 | Request intake, review and status tracking; approved deletion performed through available deletion/support procedures |
| Right to rectification | Art. 16 | Request intake, review and status tracking; approved correction handled through the applicable product/support procedure |
| Right to data portability | Art. 20 | Request tracking plus available machine-readable exports; scope confirmed per request |
| Right to object | Art. 21 | Request intake, review and status tracking |
| Right to restrict processing | Art. 18 | Handled through the Controller's documented instruction and applicable support procedure |
| Withdraw consent | Art. 7(3) | Consent event can be recorded; downstream action depends on the applicable processing context |
A.8 Recording and Interception of Communications (ePrivacy)
The recording and interception of electronic communications is regulated by the ePrivacy Directive (2002/58/EC, Article 5(1)) and its national implementations. The Customer (Controller) is solely responsible for complying with applicable ePrivacy/telecommunications laws when recording calls, obtaining any required consent for recording under national law, and providing required pre-recording announcements or notifications.
The Customer, as Controller, determines the legal basis for recording and analysing communications under the laws applicable to it. To assist Controllers in demonstrating compliance, Aelo provides a consent journal — an audit trail of consent events recorded per organization and per call record, including consent withdrawal. However, Aelo does not record calls — the Customer's telephony system performs the recording and transmits audio to Aelo for processing.
A.9 AI Processing
Aelo uses artificial intelligence to analyze call transcripts. Important limitations:
- Employee emotion recognition in the workplace is a prohibited practice under the EU AI Act (Art. 5(1)(f)), in force since 2 February 2025. Aelo does not perform emotion recognition or emotion scoring of agents or call participants.
- Sentiment analysis is text-only and customer-side only. The sentiment values in an analysis are computed by the language model from the transcript text alone, and they measure the customer's expressed sentiment during the conversation. The system does not produce a sentiment or emotion score for the agent/employee.
- Voice-delivery metrics are call-mechanics measurements, not emotion signals. Of the acoustic measurements, only three operational values — speaking rate, speech-to-silence ratio, and pause count — are shared with the language model, under an explicit instruction not to infer any participant's emotional state from them. Pitch variation, voice stability, and loudness are never included in model prompts, and no acoustic measurement contributes to any score.
- AI assistant answers are AI output, not verified statements. The assistant answers only from the transcripts and analyses it is given — of the record, of the linked records of a CRM deal, lead, contact or company, or of up to 12 records matching a question about all records, in each case among those the user may see — and cites the excerpts it relied on; a claim it cannot match to the transcript is marked as not confirmed. The limitation on automated decision-making below applies to these answers as well.
- AI analysis results are not used for automated decision-making with legal or significant effects without human review (GDPR Art. 22).
- Customers must inform call participants that AI is used to analyze calls (AI Act Art. 50 transparency obligation).
A.10 Data Breach Notification
In the event of a personal data breach affecting Customer data:
- We notify the Customer without undue delay and in any event within 48 hours of becoming aware
- We provide: nature of breach, categories and approximate number of records affected, likely consequences, measures taken
- The Customer is responsible for notifying the supervisory authority (within 72 hours for GDPR) and affected Data Subjects
Section B: Processing for Our Own Purposes (Controller)
B.1 Platform User Accounts
| Element | Description |
|---|---|
| Data subjects | Customer employees who access the Aelo platform |
| Data collected | Name, email address, role, organization, login history; for a web sign-up that started from a link with UTM tags, the traffic source of that visit (utm_source, utm_medium, utm_campaign, utm_content, utm_term) and which sign-up button on our website was clicked — kept with the account to measure which channels bring new customers, and passed to the lead described in B.6 |
| Purpose | Platform authentication, authorization, usage analytics |
| Legal basis | Performance of contract (Art. 6(1)(b)) |
| Retention | Duration of service + 30 days post-termination |
| Action log | Aelo keeps a log of security-relevant actions in the platform: opening call records and analyses and refused access to them, refused access to another organisation or from a deactivated account, role changes, changes to a project’s quality weights, analysis version changes and authorised re-analyses, linking a CRM portal and administrative recalculations. An entry holds the user’s account identifier, the organisation, the action and its parameters (identifiers, roles, the settings changed; for an authorised re-analysis, the approver’s name and reference as entered), its time and severity, and for security events and quality-weight changes the IP address and browser user agent of the request. Changes to access — a user’s role, organisation membership and project assignments, the organisation structure, analytics access grants and the merging of two accounts of one person — are also written to a separate access log, whose entry holds the organisation, the identifiers of the user who made the change and of the user whose access changed, the action, the values before and after, and its time — for a change that comes from the CRM, also the CRM user identifiers linked to users and the CRM identifiers of employees in the imported organisation structure who have no account yet — without names, email addresses, conversation content, IP address or user agent. Legal basis: legitimate interest in the security of the platform and in an account of changes made to the Customer’s settings (Art. 6(1)(f)). Retention, for both logs: 365 days from the action, and no longer than the user’s account or the organisation’s account exists; entries older than 365 days are deleted daily. When a user’s account is deleted, the entries of an organisation that still exists are kept for that organisation, until the end of those 365 days, with the user’s identifier, the action and its time, and their IP address and browser user agent are erased; entries without an organisation, and entries of an organisation already deleted, are deleted together with the account. Access log entries of an organisation that still exists are likewise kept for it after a user’s account is deleted, until the end of those 365 days. On termination the organisation’s entries are deleted together with its other data |
| Third parties | Clerk (authentication), Cloudflare (infrastructure), Sentry (error tracking and performance monitoring — records the signed-in user's name, email address, and account identifier; see A.4), Resend (account and invitation emails — the user's or inviter's name, the organisation name, the role and the invitation link; US, EU-US Data Privacy Framework) |
B.2 Website Visitors
| Element | Description |
|---|---|
| Data subjects | Visitors to our website and documentation |
| Data collected | IP address, browser information, pages visited; when you send a contact or newsletter form after following a link with UTM tags, the traffic source (utm_source, utm_medium, utm_campaign, utm_content, utm_term) travels with your request |
| Purpose | Website functionality, analytics, security |
| Legal basis | Legitimate interest (Art. 6(1)(f)) |
| Retention | 90 days; pages and traffic source attached to a request are kept as long as the request |
| Third parties | Cloudflare (CDN, security); Umami (self-hosted, cookieless analytics; when you send a contact form, the pages you viewed are attached to your request) |
B.3 Support Conversations
| Element | Description |
|---|---|
| Data subjects | Platform users who contact us through the in-app support chat |
| Data collected | Name, email address, avatar image URL (when set), organization name, subscription tier — with the subscription status appended when it is not active, e.g. "Pro (past_due)" — role, organization identifier, the page the app was on when that identity was loaded (address and path only, without query parameters), and the content of the conversation itself. Passed automatically when the user is signed in, so the conversation does not begin with identification questions. |
| Purpose | Answering support requests, and recognising a returning user so their earlier conversations remain available to the operator |
| Legal basis | Performance of contract (Art. 6(1)(b)) |
| Retention | Duration of service + 30 days post-termination |
| Third parties | Bitrix24 (Aelo's own portal, used as our CRM and support desk) |
B.4 Billing
| Element | Description |
|---|---|
| Data subjects | The platform user who initiates a payment action (subscribing, topping up prepaid credit, or saving a card for auto-recharge) on behalf of their organization |
| Data collected | Organization name, the initiating user's email address, and Aelo's internal organization identifier — passed to our payment processor when a billing customer record is created. This applies to every self-serve payment, including prepaid top-ups by Pay-as-you-go organizations that have no subscription. Card details are collected directly by the payment processor on its own pages and never reach Aelo's systems. |
| Purpose | Subscription billing, prepaid credit top-ups, auto-recharge, invoicing, and payment records required by tax and accounting law |
| Legal basis | Performance of contract (Art. 6(1)(b)); legal obligation (Art. 6(1)(c)) for tax/accounting records |
| Retention | Duration of service; payment records thereafter as required by applicable tax and accounting law |
| Third parties | Stripe (payment processing; US — EU-US Data Privacy Framework). Resend (budget, balance and auto-recharge notices and minute-usage emails — amounts spent, budget limits, the prepaid balance, the top-up amount and number of failed attempts, and minute usage with the organisation name; the pause notice of an eligible free-tier organisation also offers bonus minutes with a deadline; US, EU-US Data Privacy Framework). Stripe processes payment data for our own billing as our processor; it is not a sub-processor of Call Data — see the Sub-Processor List |
B.5 Marketing Email (Newsletter & Onboarding Series)
| Element | Description |
|---|---|
| Data subjects | A person who asks for the newsletter, in any of the three places we offer it: the marketing checkbox at web sign-up or in the embedded Marketplace flow; the one-time question shown when a CRM portal account is linked to a web account; or Settings › Notifications, where it can be switched on and off at any time. Recorded per PERSON, not per organization: the consent is attributed to that user, and nobody else in the organization is enrolled by it — installing the CRM app seeds an organization's whole staff, and none of them are subscribed by that. Outside the embedded consent screen only an account owner or administrator is offered the newsletter; the checkbox on the embedded consent screen is shown to every user who signs in there with a deliverable e-mail address, whatever their role, and ticking it subscribes that person alone. An account with no deliverable e-mail address — as a CRM portal user has before linking — cannot be subscribed under any circumstances |
| Data collected | Email address, plus the contact variables that decide which message is sent: display name, interface locale, self-declared role, team size and industry (both optional, absent from the embedded flow), the signup source and date, the traffic source of the visit that led to a web sign-up (UTM tags and the sign-up button clicked), when there was one, the plan (free/paid), and two behavioural milestone flags — whether a data source has been connected and whether an analysis has run. Beyond sign-up, three updates reach the same contact, each only for a person who holds this consent: the plan changes to paid when the organization moves to a paid plan; the two flags are set when the organization sends its first record and gets its first analysis; and, when the organization has used 80 % of its included minutes, one event carrying your email address, the organization's identifier, the minutes used and included, and the language, sent for each administrator who consented. If you consent after your organization is already active (at sign-up, in Settings or in the CRM embedding), your plan and flags are raised to what the organizations where you are an account owner show, never lowered, and nothing is written for what they do not show. No call content, no transcripts, no customer data of any kind |
| Purpose | The onboarding email series: setup guidance, feature introductions, and product news — and the signals above, which only decide which message is sent (for example, no reminder to connect a source once one is connected, and no upgrade message once the plan is paid or the 80 % threshold is reached) |
| Legal basis | Consent (Art. 6(1)(a)). Always a separate, unticked opt-in or an explicit action of its own — never bundled with the Terms or this Policy, which are a precondition of using the product, and never a default. The grant is written to the consent journal with the user, timestamp and document versions, and neither the series nor any of the signals above is sent without it: the organization's first client or administrator is not enough, each person must hold the consent themselves. Withdrawing it stops all of them |
| Retention | Until consent is withdrawn or the account is closed; the withdrawal event itself is retained as evidence of when consent ended |
| Third parties | SendPulse (marketing email automation, receiving only what is listed under Data collected; see A.4 and the Sub-Processor List) |
| How to withdraw | Withdraw it in the platform under Settings › Notifications (consent type marketing_email) — the same place it is switched on, so withdrawing is no harder than consenting — use the unsubscribe link in any message of the series, or write to privacy@aelo.cloud — each one stops delivery. A withdrawal recorded in the platform is written to the consent journal against your user and unsubscribes your address from the series; because it is attributed to you personally, it affects only your subscription and no colleague's. Withdrawing marketing consent changes nothing about the service — it is not a precondition of anything |
B.6 New Organisations (Sales Follow-up)
| Element | Description |
|---|---|
| Data subjects | The person who creates an organisation in Aelo by a web sign-up; for an organisation created by installing Aelo from a Bitrix24 or Uspacy portal, no person — only the portal |
| Data collected | For a web sign-up: the user’s name and email address (not a placeholder address of an account without a mailbox), the organisation name, the self-declared role, team size and industry from the sign-up wizard, the traffic source of the visit that led to the sign-up (UTM tags and the sign-up button clicked) and Aelo’s organisation identifier. For an installation from a portal: the portal address, the organisation name and identifier |
| Purpose | A lead in our own CRM so that our team can follow up with a new organisation (onboarding help, questions about its plan). It is not marketing email: the onboarding series is sent only with consent (B.5) |
| Legal basis | Legitimate interest in following up with organisations that start using the product (Art. 6(1)(f)); you can object at any time (B.7) |
| Retention | While we pursue the commercial relationship with the organisation; the lead is deleted on request or objection |
| Third parties | Bitrix24 (Aelo’s own portal, used as our CRM and support desk) |
B.7 Your Rights (Section B Data)
For data we process as Controller (Section B), you can exercise the following rights by contacting us at privacy@aelo.cloud:
- Access — Request a copy of your personal data
- Rectification — Correct inaccurate data
- Erasure — Request deletion of your data
- Restriction — Limit how we process your data
- Portability — Receive your data in a portable format
- Object — Object to processing based on legitimate interest
- Withdraw consent — For marketing communications
We will respond within 30 days (GDPR) or 45 days (CCPA). You also have the right to lodge a complaint with your local supervisory authority. For Spain: Agencia Española de Protección de Datos (AEPD), www.aepd.es.
4. Children's Data
Aelo is a B2B service not directed at children. We do not knowingly process personal data of children under 16.
5. Changes to This Policy
We may update this Privacy Policy from time to time. Material changes will be communicated to Customers via email at least 30 days before they take effect.
6. Contact
Data Processor / Controller:
Auspex Streamline S.L.
C.I.F.: B56341829
Calle Velarde 13, 4B
35010 Las Palmas de Gran Canaria
Canarias, Spain
Privacy inquiries: privacy@aelo.cloud
For call participants: Please contact the company that recorded your call. They are the Controller of your data and responsible for responding to your requests.
Document ID: PP-Aelo-2026-001 · Version: 1.77