Evaluation Profiles
Evaluation profiles (formerly Prompt Profiles) define what the AI looks for when analyzing a conversation. They control which aspects are evaluated and what the output includes. Different projects can use different evaluation profiles.
Not the same as Criteria. A profile is the scorecard Aelo applies to a whole category of conversations. Questions asked of every conversation — yes/no, a score or a choice — are a separate feature — see Criteria.
Owner Admin
Aelo provides base categories that serve as starting points:
| Category | Best For |
|---|---|
| Sales Call | Outbound/inbound sales phone conversations |
| Support Call | Customer service and technical support calls |
| Sales Chat | Text-based sales conversations (web chat, messaging) |
| Complaint Resolution | Complaint handling and escalation calls |
The base category chooses the built-in worked examples the model sees and the modules a new profile starts with. Chats always use the Sales Chat example set, whichever base category the profile has. You choose the category in the Base category field when you create a profile (New Profile on the Profiles tab); it cannot be changed afterwards.

Owner Admin
Aelo automatically detects the category of every call (Purchase, Support, Complaint, etc.). Each organization gets a set of category-bound profiles — one per category — so every conversation is analyzed with metrics tuned for its type:
- 10 call categories: Product Inquiry, Purchase, Scheduling, Support, Billing, Onboarding, Complaint, Cancellation, Follow-up, Other.
- Chats: detected into the same categories as calls, and analyzed with their category’s profile. The channel also changes what is evaluated (and a chat skips the deal-stage step when its profile is chosen): metrics that apply only to speech (filler words, interruptions, diction, timestamped key moments) don’t apply to chats, and a chat is calibrated on the Sales Chat built-in examples, whichever base category its profile has.
- Missed and short calls are not analyzed, so they have no profiles. The same applies to a chat with no rep reply, or with fewer real messages than the project’s Min Chat Messages setting (4 by default).
When a record is analyzed, the profile is selected in a fixed order, described step by step in Full Resolution Order below.
Assigning, changing, or removing a category binding happens only in the matrix: Analysis Setup → Categories and Profiles → Categories tab. Each row of the matrix is a category with its current bound profile (or “Base profile” if unbound):
- Assign — pick a profile for the category.
- Restore default — recreates the pre-configured profile for a category after a binding is removed.
- A row’s picker offers only profiles not bound to another category, plus the row’s current profile, so one profile cannot hold two bindings.

The profile editor itself (Profiles tab, formerly the Prompt Builder) shows the binding read-only, with a link back to the matrix (“Binding is changed on the “Categories” tab”).
Each category can have at most one bound profile per organization.
Owner Admin
Category-bound profiles are convenient, but some projects need to be locked to one profile no matter what category the AI detects — for example, a project dedicated entirely to a single sales motion. Pinning does that: it opts a project out of the org’s category auto-selection.
Open Projects → select a project → Settings → Analysis & Quality tab → Pinned analysis config:
- Pin analysis profile — always use the project’s selected analysis profile for every analysis in this project, regardless of the detected call category. There is no pin for scripts, and none is needed: a script you assign to a project already beats the category binding on its own (see the selection order below).
The toggle requires the project to already have a prompt profile selected — it stays disabled with a hint until you pick one above it. Pinning without a selection is rejected.

With pinning in the picture, a record’s prompt profile is chosen in this order:
- Explicit override — a manual choice made when re-analyzing a specific record; always wins.
- Pinned project profile — used when Pin analysis profile is on.
- Deal-stage mapping — for orgs with stage-based scorecard selection turned on and connected to Bitrix24 or Uspacy: an exact mapping for the linked deal’s current stage, falling back to a mapping for the stage’s semantic phase when no exact mapping exists. Skipped entirely when the call isn’t linked to a Bitrix24 or Uspacy deal.
- Category-bound profile — the org profile bound to the detected call category.
- Project profile — the profile linked to the project (fallback when nothing above applies).
- Default profile — used when none of the above is configured.

Script selection for adherence scoring follows a similar shape minus the deal-stage and override steps (re-analysis doesn’t let you choose a script, and script adherence is not stage-aware) — but with the project ahead of the category binding, unlike profiles: project script → category-bound script → no script (adherence scoring is skipped for that conversation). A script you assign to a project is a deliberate choice, so it wins; category bindings apply to projects that have no script of their own.
- Deactivating a pinned profile does not reliably unpin it. If you deactivate the profile that’s currently pinned, the pinned step is skipped on the next analysis — but if no category binding matches that call, the project-profile fallback step can still pick up that same (inactive) profile. If you want a project to stop using a profile, unpin it explicitly rather than deactivating the profile.
- Only one control stops adherence scoring for a project: Enable rule-compliance scoring in its Analysis & Quality tab. Switch it off and the step is skipped, whatever any binding says.
- Clearing a project’s script, or deactivating it, changes WHICH script is used — not whether scoring happens. Both make resolution fall through to the category-bound script, so a conversation whose category has a binding is still scored, just against that script. Deactivating also reaches beyond this project: a script belongs to the organization, so it stops being used everywhere it was assigned, and in its category binding too.
- Deleting a pinned profile also falls through cleanly: category binding → project profile → default.
- A script assigned to nothing is never applied — and the list says so. A script that is bound to no call category and is not the default script of any active project matches no step of the cascade, so script adherence is skipped for every conversation. In Script rules, such a script carries a Not assigned marker; the others show the category they are bound to and the projects that use them. An attachment that cannot work — the project uses another script as its default, or the project is not active — is dimmed and explains itself on hover.
Owner Admin
For organizations connected to Bitrix24 or Uspacy, Aelo can select the evaluation scorecard based on the linked deal’s current CRM pipeline stage, on top of the detected call category — useful when, say, a discovery call and a closing call in the same category deserve different evaluation profiles. This applies to calls linked to a Bitrix24 or Uspacy deal; leads are not covered in this version.
Deal-stage mapping sits between the pinned project profile and the category-bound profile in the resolution order above: an exact stage mapping wins first, falling back to a mapping for the stage’s semantic phase when no exact mapping exists for that stage (the phase comes from the CRM’s own marker — Bitrix24’s won/lost marker, or the closing success/fail stage of a Uspacy funnel — else from the phase you chose for the stage, else is guessed by AI from your pipeline and stage names — see Deal Phases below). It is skipped entirely — falling straight through to the category-bound profile — for chats, calls not linked to a Bitrix24 or Uspacy deal, and orgs that haven’t connected Bitrix24 or Uspacy or turned this on.
Owner Admin
Open Analysis Setup → Categories and Profiles → Deal Stages tab to configure the mapping:
- Deal Phases — set a default evaluation profile for each of the 7 coarse semantic phases (Discovery, Qualification, Proposal, Negotiation, Closing, Post-sale, Other). They apply automatically at analysis time to a stage with no exact mapping: Post-sale and Other from the CRM’s own won/lost marker (the deal’s won/lost status in Bitrix24, the closing success/fail stage of a Uspacy funnel), the rest from the phase you chose for the stage under Pipelines & Stages, or — until you choose — the phase AI guesses from your pipeline and stage names, in any language and also for stages you renamed (for example, a stage named “Discussing terms” counts as Negotiation). Only the names of your pipelines and stages are sent to the AI model for this — no customer or conversation data; Aelo pays for it, no minutes are spent, and the names are sent again only when a stage is added or renamed, or an earlier attempt gave no answer. The phase is stored on the call; a guessed one is marked “AI guess” on this tab and as a guess on the Objections screen.
- Pipelines & Stages — each stage has two settings. Phase: choose the stage’s phase, or leave it on the AI guess (a won or lost stage shows the phase the CRM sets, with no choice). Changing a phase also moves the calls already recorded at that stage to it — on the Objections screen; their scores are not recalculated. Profile: map the stage to a profile directly; an exact mapping always takes priority over the phase default.
- Keep guessed phases — stores every phase still guessed by AI as your own choice in one action, so renaming a stage in the CRM no longer changes its phase.
- Accept suggested — for every unmapped stage with a usable guessed phase, creates an exact binding to that phase’s default profile in one action, so the stage keeps that profile even if its name changes later.
- Sales methodology — on Pro and higher plans the methodology of each pipeline (the framework and optional company rules that the advice in the deal state follows) is set on its own step, Analysis Setup → Sales methodology; the tab links there.
The Pipelines & Stages list requires an active Bitrix24 or Uspacy connection (a Uspacy one only while its processing is active). Without a connection the tab prompts you to connect a CRM source first; with a Uspacy portal whose processing is paused or withdrawn it says so and reads nothing from the portal.

When a call is scored using the deal-stage mapping, the analysis panel shows a Stage: … badge next to Profile: …, so you can see which mapping applied. This is visible in the standalone Aelo record view; it is not yet shown inside the embedded Bitrix24 panel.
Each profile consists of analysis modules. Enable or disable modules to control what the AI evaluates. The “said / not said” checks — mandatory phrases, profanity and the recording disclaimer — are not profile modules and are no longer part of the analysis: built-in criteria do them, see Checks Done by Criteria below:
Produces high-level metrics:
- Quality Score (0–100)
- Sale Probability (0–100%)
- Attention Flag — marks conversations needing manual review
Extracts conversation content:
- Topic Detection — main subject of the conversation
- Keywords — important terms and phrases
- Key Moments — critical events (objection raised, pricing discussed, agreement reached)
- Summary — natural language overview of the conversation
- Structured Summary — the same in three parts: situation, actions taken, outcome
Evaluates speaking patterns:
- Filler Words — how many hesitation sounds the rep made (“um”, “uh”, “hmm”), counted in the rep’s own speech only. Words are not counted, even common fillers like “like” or “well”: they often carry meaning, and a count without context would blame the rep for them. The score depends on the rate per minute: each sound a minute takes 10 points off 100, so ten or more a minute score 0. The call transcript hides these sounds to stay readable; the count is made on the verbatim text. If the rep can’t be identified in the call, the metric is not shown
- Rep Uncertainty — the rep hedging on their own price, deadline, or promise (not retelling the customer’s words, instructions, or estimates). The signal list is built into the analysis and is not configurable
- Rep Errors — mistakes the rep made (wrong information, a customer request left open, and the like), each with a verbatim quote as evidence
Evaluates conversation structure:
- Greeting & Intro — proper opening and introduction
- Needs Qualification — how fully the customer’s needs were explored, by the framework of the conversation’s category (BANT — Budget, Authority, Need, Timeline — for a purchase)
- Objection Handling — how the rep responds to customer objections
- Closing Techniques — effective conversation conclusion and next steps
Analyzes emotional aspects:
- Sentiment Analysis — positive, neutral, negative classification
- Politeness — rep courtesy and professionalism
- Interruptions — frequency and handling of speaker interruptions
- Speech Clarity — overall speaking clarity score
Checks regulatory and policy adherence:
- CRM Completeness — only assessed when there is a connected CRM and collecting the customer’s identifying details was actually this call’s job (a new customer not yet linked to any record); calls tied to an existing deal or known contact are skipped rather than scored low, and so are calls from organizations with no CRM integration, where there is no CRM to measure hygiene against. When applicable, it checks that the customer/contact name and contact info (phone or email) were captured — plus the company name for business customers only, since a private individual has none. This is analyzed and shown as its own compliance indicator and does not contribute to the quality score.
Generates coaching output:
- Strengths — what the rep did well
- Improvements — specific areas to improve with actionable suggestions
Picks up business signals mentioned in passing: competitor mentions, product feedback, market signals and compliance flags.
The “said / not said” checks — mandatory phrases, profanity, the recording disclaimer and the compliance checks (personal data handling, US TCPA) — are not part of the analysis profile or of the main analysis: the profile has no switches for them. They are done only by built-in criteria worded by Aelo. Up to eight of them are visible to every organization in Analysis Setup → Criteria, in the Built-in section; which of them are asked depends on the organization and the conversation, and ones paused by Aelo are not asked (see Pauses).
| Check | Built-in criterion |
|---|---|
| Mandatory phrases | Named the company; Agreed on the next contact; Explained the satisfaction guarantee |
| Recording disclaimer | Disclosed call recording |
| Profanity | Rep used profanity |
| Compliance requirements | Broke personal data rules; Met TCPA requirements |
A new organization gets these criteria set up from Aelo’s defaults: Named the company, Agreed on the next contact and Explained the satisfaction guarantee apply to Product Inquiry, Purchase, Scheduling and Follow-up conversations; Rep used profanity applies to all conversations; Disclosed call recording starts paused — switch it on in Analysis Setup → Criteria when you need it. Organizations that already have these criteria keep their scope and pauses as they were. There you pause or resume a check and narrow it to projects, categories and channels. Aelo can pause a built-in criterion itself (for example, when its wording does not fit the jurisdiction), and the owner cannot resume such a pause.
Personal data. The old personal-data check has left the analysis. Personal data is now checked only by the built-in criterion Broke personal data rules, on calls of organizations with a concrete jurisdiction: it counts a request for full card details, a password or a code, and the customer’s data named, after a sign that the person who answered is not the customer, before finding out who they are talking to — even if it later turns out a relative placed the order; an ordinary call about an order to a person who accepts the order as theirs is not a violation.
Earlier analyses. Conversations analyzed before this change keep the checks they stored, and the call panel, the PDF report and the reports show them as before. New analyses do not write them. In the frequent dips on the dashboard (Rep’s frequent dips on the Team tab and My frequent dips on Mine), new analyses no longer list the rows Mandatory phrases, Profanity, Recording disclaimer and Compliance.
Forbidden words. The list of forbidden words is not a profile switch and is not set in project settings: it belongs to the built-in criterion Rep used a forbidden word (see The Forbidden Words List).
- Open Analysis Setup → Categories and Profiles → Profiles tab in the main menu (formerly the Prompt Builder).
- Pick a profile in the toolbar, or click New Profile. In the panel that opens, enter a name (required) and an optional description, choose the Base category and click Save profile. A new profile is not bound to any conversation category until you assign it on the Categories tab.
- Switch modules on or off in the Analysis Modules panel on the left: four essential ones (Quality Score, Summary, Keywords, Sentiment Analysis) sit at the top, the rest are in collapsible groups with an on/total counter. Select All and Reset to Base are at the bottom of the panel. The “said / not said” checks are not in the profile — set them up in Analysis Setup → Criteria.
- Click Save to Profile to keep the changes (the button is active only while there are unsaved ones).
- To see what the modules produce before you rely on them, edit the sample transcript in the Live Preview card and click Analyze (see Live Preview).

The middle of the Profiles tab is a test bench for the selected profile:
- Sample transcript — the Live Preview card holds an editable transcript (a sample dialogue to start with, with a character counter). Paste a real conversation or write your own.
- Analyze — runs a test analysis of that transcript with the modules exactly as switched on now, saved or not. The panel on the right (Analysis Results) shows what came back, with Export JSON, Copy and View Raw JSON. The test is not stored as a record.

The prompt text and the token and cost estimates are shown only to Aelo platform administrators, not in customer organizations.
Use the test to find the right balance between analysis depth and how much output a profile produces.
By default the preview is computed at organization level, and project rules are not part of it. The toolbar’s Preview scope selector (by default No project (org level)) lets you pick a project — Analyze then applies that project’s settings: the report language, and the project’s name and description. Those are the prompt inputs configured on the project; data a real analysis pulls in as it runs (CRM fields for autofill, a tailored analysis configuration) is not part of the preview.
To assign an evaluation profile to a project:
- Go to Projects > select a project > Settings.
- Find the Analysis profile section.
- Select the desired evaluation profile from the dropdown.
- Save.
All new records in that project will use the selected profile. Existing records are not affected unless you manually re-analyze them.
The project profile acts as a fallback: it applies when no category-bound profile exists for the conversation’s detected category (see Category-Bound Profiles) — unless the project has pinned its own profile, which always takes priority over category bindings.

| Feature | Owner | Admin | Supervisor | Rep |
|---|---|---|---|---|
| View profiles | ✅ | ✅ | ❌ | ❌ |
| Create/edit profiles | ✅ | ✅ | ❌ | ❌ |
| Delete profiles | ✅ | ✅ | ❌ | ❌ |
| Link to projects | ✅ | ✅ | ❌ | ❌ |
| View live preview | ✅ | ✅ | ❌ | ❌ |
- Start with a base category and customize — don’t build from scratch.
- Enable only the modules you need. More modules = higher cost per analysis and longer processing. The “said / not said” checks (mandatory phrases, recording disclaimer, profanity) are not in the profile: built-in criteria do them.
- Use Analyze in the Live Preview on a sample transcript to check a new profile before you deploy it.
- Create different profiles for different conversation types (e.g., one for sales calls, another for support).
- After changing a profile, re-analyze a few recent records to verify the results meet expectations.
- For regulated industries, set up the built-in criteria, not the profile. The recording disclaimer and mandatory phrases are checked by built-in criteria only: open them in Analysis Setup → Criteria, make sure they are not paused (the recording disclaimer criterion starts paused for a new organization), and narrow them to the projects and channels that must carry them.