Get a verdict on the proposition before the budget is committed.
A bounded synthetic audience answers your hypothesis and returns GO, NO-GO or CONDITIONAL with a confidence score. Personas calibrate on the real advisor data of the market you name, and a run takes two to three minutes.
Where it runs today: your team reads it in the Emable workspace, signed in and your own AI client calls it as a tool over MCP.
What changes
Who inside your organisation is better off, and what they can show for it.
Head of Product at an insurer or distributor
Walks into the steering committee with a verdict, a confidence score and the audience it came from, not an opinion.
get_result returns GO, NO-GO or CONDITIONAL with a confidence score, a one-sentence summary and a dashboard link. Source: the get_result tool contract in the emable capability catalogue.
Pricing Lead at an insurer
Tests the tier and the price point against a bounded audience before the tariff is locked, and can raise the stakes setting for a bigger bet.
run_validation takes a hypothesis, an audience and a decision_size of S, M, L or XL, where larger sizes trigger deeper analysis. Source: the run_validation argument schema in the emable capability catalogue.
Distribution Director at an adviser network
Asks the question of an audience calibrated on the market's real advisor data, instead of a workshop of eight people in one office.
create_audience takes a market and calibrates the personas on that market's real advisor data. An audience holds 5 to 50 personas, default 10. Source: the create_audience argument schema.
Head of Insight at a bank or distributor
Keeps a defensible trail: which audience, which sources, which date, per run, with the evidence labelled directional and never dressed as a survey.
Audience and source snapshots are recorded with each Validator run. The manifest states plainly that synthetic-persona evidence is directional, not observed customer behaviour. Source: the capability evidence_freshness and limitations fields.
AI Product Lead at a fintech or distributor
Puts the same verdict rail inside the assistant their teams already use, with no new screen to roll out.
it runs in the Emable workspace and through the connector, and six Validator tools are wired and callable on the connector today. Source: the emable capability catalogue.json.
Two ways to get it
Have it delivered, or let your own AI call it.
As a service
We run the study for you. You send one sentence: the decision you are about to fund, the market, and who you sell through. We build the audience and calibrate it on that market's real advisor data. We run it at the decision size the stakes deserve, S through XL. You get back the verdict, the confidence score and the audience it came from, in a form a steering committee can read in one page. Write to us about the decision, not about credits. We reply with the audience we would build and the decision size we would use, before anything is charged. There is no self-serve run on this page. Every route starts with a person or a connected client.
Tell us what you needOne message, no form. We answer with what it would take.
Through your own AI
Market Validator is live and reachable over MCP today. Connect the Emable connector once and six Validator tools appear in Claude, ChatGPT, Codex or any MCP client. Your assistant can list the audiences you already hold, inspect who is inside one, create a new audience calibrated on a named market, start a run and poll it through to a GO, NO-GO or CONDITIONAL verdict with a confidence score. Tools only appear when the credential carries account:read and validator:run. A run takes two to three minutes, so the assistant polls get_result about every fifteen seconds. What comes back is the verdict, a one-sentence summary and a dashboard link, not a persona dump. Two of the six tools write, create_audience and run_validation, so the assistant should confirm the audience and the spend with a person before it calls them.
- Connector
- https://api.emable.ai/mcp
- Tools it adds
create_audienceget_audience_detailsget_resultlist_audienceslist_validationsrun_validation
What your AI can answer
Connect it once, and your assistant answers these.
Questions in your own words, each one landing on a published tool. Ask us and we answer the same ones for you.
- 1Would advisers in Spain pay for this at the price we are testing?
create_audience - 2Is that a GO, a NO-GO or a CONDITIONAL, and how confident is the verdict?
get_result - 3Who was actually in the audience that verdict came from?
get_audience_details - 4Which audiences do we already have saved, and how many personas does each hold?
list_audiences - 5Have we tested a hypothesis like this one before, and what did it return?
list_validations - 6Can you build me an audience of Czech insurance branch managers running teams of five to fifteen advisers?
create_audience - 7Run this pricing hypothesis against our Italian audience and our Spanish one, then tell me where the two verdicts differ.
run_validation - 8This one is a company-defining bet. Can you run it at XL and tell me the moment the verdict lands?
run_validation
Market Validator answers these through the tools above. A question outside them is refused, not guessed.
What it rests on
The evidence, and the edges.
Evidence
- Status live, exposure public, provider emable, EU coverage, derived data class, write access. Source: the emable capability catalogue.
- Six wired Validator tools on the connector today, four read and two write: list_audiences, get_audience_details, list_validations, get_result, create_audience, run_validation. Source: the emable capability catalogue.json.
- get_result returns GO, NO-GO or CONDITIONAL, a confidence score, a one-sentence summary and a dashboard link. Source: the get_result tool contract.
- A run takes two to three minutes, polled roughly every fifteen seconds. Source: the run_validation and get_result latency fields.
- Personas calibrate on the named market's real advisor data. An audience holds 5 to 50 personas, default 10. Source: the create_audience argument schema.
- Every run records the audience and source snapshot it used. Source: the capability evidence_freshness field.
- Charging unit is one run over 10 personas, with a hard maximum of 500 credits. Credits are the platform's metering unit, not a price. Source: the capability pricing policy.
- Verified usage: Not enough verified runs yet. Public statistics appear only after 20 verified runs across 5 different accounts. Source: the platform statistics rule.
Audience and source snapshots are recorded with each Validator run.
What it does not do
- Synthetic-persona evidence is directional, not observed customer behaviour.
- Trial runs are limited to 10 personas.
- A trial run exports no raw personas.
- It needs a configured audience and a Validator execution entitlement.
- There is no self-serve run on this page. Every route starts with a person or a connected client.
- It does not tell you what a named customer did. It tells you what a bounded synthetic audience answered, and records which audience that was.
Verified usage
Not enough verified runs yet
Public statistics appear only after 20 verified runs across 5 different accounts.
The contract, for the people who will wire it
Status, permissions, tools, schemas and where it runs. The same values the API returns, so nothing here can drift from what your client will actually get.
Status, access and price
- Status
- Livepublic
- Provider
- emable
- Manifest version
- v1.0.0
- Countries
- EU
- Data class
- derived
- Access
- write
- Permissions
account:readvalidator:run
- Charging unit
- One run over 10 personasHard cost ceiling: 500 credits per run.Credits are the platform's metering unit, not a price. We quote the engagement.
Tools and schemas
List audiences
list_audiences
List your saved audiences so you can find the one to validate against.
Use it when
- You need an audience_id for run_validation or get_audience_details but do not have one yet.
- The user asks what audiences already exist.
Returns
Audience id, name and persona count for each.
Get audience details
get_audience_details
Inspect one saved audience's personas so you can confirm it targets the right people before spending credits on a run.
Use it when
- You are about to call run_validation and want to confirm the audience matches the hypothesis first.
- The user asks who is actually in a given audience.
Returns
Audience profile plus up to 10 personas with age, role and bio.
Arguments
| Name | Type | Required | Description |
|---|---|---|---|
| audience_id | string | yes | UUID of the audience. Get this from list_audiences. |
List validation runs
list_validations
Browse recent validation runs so you can avoid re-running a hypothesis someone already validated.
Use it when
- The user asks about past results before starting a new validation.
- You want to check whether a similar hypothesis was already tested.
Returns
Verdict, decision size, date and dashboard link per run.
Arguments
| Name | Type | Required | Description |
|---|---|---|---|
| limit | number | no | How many recent validations to return (1–50). Defaults to 10. |
Get validation result
get_result
Poll a validation run so you can return its GO/NO-GO verdict the moment it finishes.
Use it when
- You just called run_validation and are waiting for its simulation_id to complete.
Returns
GO/NO-GO/CONDITIONAL verdict, confidence score, summary and a dashboard link.
Typical latency
Polls immediately; the underlying run itself takes 2 to 3 minutes.
Arguments
| Name | Type | Required | Description |
|---|---|---|---|
| simulation_id | string | yes | The simulation_id returned by run_validation |
Create audience
create_audience
Create a reusable audience from a plain-language description so you can validate against a real, calibrated population instead of an ad-hoc guess.
Use it when
- No saved audience from list_audiences matches the target group.
- The market is known and specific — pass it, so personas calibrate on real advisor data rather than a generic default.
Do not use it when
- A close-enough audience already exists — reuse it with list_audiences instead of fragmenting history across near-duplicate audiences.
Returns
The new audience id.
Arguments
| Name | Type | Required | Description |
|---|---|---|---|
| name | string | yes | Short display name, e.g. 'Spanish insurance advisors'. |
| description | string | yes | Who these people are — role, demographics, context. More specific = better personas. |
| market | string | no | Country or market, e.g. 'Spain' — enables calibration on that market's real advisor data. |
| size | number | no | How many personas the validation should generate (5–50, default 10). |
Example: Create a market-specific audience
{
"name": "Spanish insurance advisors",
"description": "Independent insurance advisors registered with DGSFP, serving mass-affluent retail clients.",
"market": "Spain"
}{"audience_id": "..."}
Run a validation
run_validation
Validate a business hypothesis against a bounded synthetic audience so you can get a GO/NO-GO verdict before committing real budget.
Use it when
- The hypothesis, audience, market, and decision size are all confirmed with the user.
- The user has approved the credit spend for a run at this decision_size.
Do not use it when
- No audience_id is confirmed — call list_audiences or create_audience first.
Returns
A simulation id to poll with get_result.
Typical latency
2 to 3 minutes; poll with get_result roughly every 15 seconds.
Arguments
| Name | Type | Required | Description |
|---|---|---|---|
| hypothesis | string | yes | The business decision or hypothesis to validate. Be specific — e.g. 'Should we launch a 990 Kč/mo premium tier targeting Raynet power users?' |
| audience_id | string | yes | UUID of the audience to validate against. Get this from list_audiences. |
| market | string | no | Country/market of the audience (e.g. 'Spain'), so personas are calibrated on that market's real advisor population. |
| decision_size | string | no | Stakes of this decision. S = quick experiment, M = standard (default), L = significant investment, XL = company-defining bet. |
Example: Validate a pricing decision
{
"hypothesis": "Should we launch a 990 Kč/mo premium tier targeting Raynet power users?",
"audience_id": "...",
"decision_size": "M"
}{"simulation_id": "..."}
Errors a caller must handle
invalid_api_keyThe credential is absent, invalid, expired, or revoked.Stop and reconnect with a valid key or OAuth grant.missing_scopeThe identity lacks the `validator:read` scope this tool requires.Request `validator:read` and reconnect.422Input does not match this tool's schema.Correct the input using the schema returned with this tool.capability_runs_disabledExecution is temporarily off for this capability.Keep discovery available and retry later; this is not a permanent failure.
Input schema
{
"type": "object",
"properties": {
"hypothesis": {
"type": "string",
"minLength": 10,
"maxLength": 2000
},
"market": {
"type": "string",
"minLength": 2,
"maxLength": 120
},
"audience_id": {
"type": "string",
"format": "uuid"
}
},
"required": [
"hypothesis",
"market"
],
"additionalProperties": false
}Output schema
{
"type": "object",
"properties": {
"status": {
"type": "string"
},
"verdict": {
"type": "string"
},
"confidence": {
"type": "number"
},
"result_reference": {
"type": "string"
}
},
"required": [
"status",
"result_reference"
],
"additionalProperties": false
}Surfaces and connectors
Emable Capabilities
https://api.emable.ai/mcp
Streamable HTTP · Emable OAuth · dedicated tool allowlist
Sources, limits and support
Sources
The customer-selected Emable audience and Validator synthetic personas.
Limitations
Trial runs are limited to 10 personas.
Synthetic-persona evidence is directional, not observed customer behavior.
Dependencies
- A configured audience and Validator execution entitlement