Tool Reference

Chat directly with an individual digital twin. Give explicit situational context (where the asset is seen, price, alternatives) and ask for a verbal reaction before any numeric rating; replies are stated opinions, not behavior predictions. Twin endpoints are stateless — for multi-turn follow-ups pass the full history via `messages`.

Ask a twin group one question and get every twin's answer plus an aggregated summary. Single-question wrapper over the same endpoint as chat_with_twin_group. Prefer a deliberate panel of 5-10 twin_ids (omitting twin_ids fans out to all twins via an extra list_twins call; max 20). Ask for verbal reactions before numeric ratings; answers are stated opinions, not behavior predictions.

Start a conversation with an entire demographic twin group. Prefer a deliberate panel of 5-10 twin_ids (omitting twin_ids fans out to all twins via an extra list_twins call). ONE stimulus per call — never batch items to compare. Answers are stated opinions, not behavior predictions. Twin endpoints are stateless — for multi-turn follow-ups pass the full history via `messages`.

Start an async job that creates N digital twins for a target audience. Returns the job; twins appear as the job processes.

Invite a user to join a workspace by email address

Cancel/delete a pending workspace invite

Create a new workspace with the current customer as the owner

Update a workspace's name

Permanently delete a workspace (destructive)

Create a workspace member directly (admin/support only)

Update a workspace member's role

Remove a member from a workspace

Accept a workspace invite using its confirmation code

Reject a workspace invite using its confirmation code

Resend the invitation email for a workspace invite

Update a subscription (admin)

Partially update a subscription (admin)

Increment the word-usage counter on the workspace's current subscription

Update the billing contact for a workspace (name and email are required; other fields sync from Stripe)

Create a Stripe checkout session so the customer can create or update a subscription

Create a new brand voice in a workspace

Update an existing brand voice. Accepts the full attribute set from import endpoints. Name is always required; all other fields are optional. JSON array/object fields (perspective, vocabulary, glossaries, generalTone, etc.) accept Python lists and dicts in execute_code — they are serialised automatically.

Delete a brand voice from a workspace

Create a brand voice by analysing a URL

Create a brand voice by analysing a sample text

Create a brand in a workspace. All fields are required by the API; pass empty strings or empty arrays for unknown values.

Update a brand. PUT semantics — all fields required on every call.

Delete a brand from a workspace

Duplicate a brand voice under a new name

Upgrade a legacy brand voice to the current schema, attaching an audience model

Convert a brand voice into the DS prompt format used by data-science backends

Calculate the contradiction score for a brand voice against an audience model

Record a usage event for a brand voice (used for analytics/quotas)

Create a brand voice by importing the contents of a base64-encoded file

Create a new target audience in a workspace. Provide a full demographic profile: name (3–100 chars), residence (2–50 chars — short location label like 'Berlin' or 'Western Europe', NOT a full descriptor), country, age range (min/max 0–200), gender distribution (female/male/non-binary/prefer-not-to-say as 0–1 fractions summing to 1), pain points, interests, and needs (1–15 entries each, max 250 chars per entry). Returns the created audience including its ID.

Update an existing target audience. PUT semantics — full demographic profile required on every call, including gender distribution (female/male/non-binary/prefer-not-to-say as 0–1 fractions summing to 1). Same field caps as create_target_audience apply (name 3–100, residence 2–50, painPoints/interests/needs 1–15 entries × 250 chars).

Delete a target audience from a workspace

Infer a target-audience profile from any combination of free-form text, URLs, brand voices, and uploaded files. Returns a PREPARED profile (name, residence, painPoints, interests, needs, minAge, maxAge, genderDistribution, country, nativeLanguage, additionalInformation) — does NOT persist the audience and does NOT return an audience ID. Forward the returned fields to create_target_audience to save it (then use the resulting ID for create_twin_creation_job). Trim 'residence' to ≤ 50 chars before forwarding; minAge < 18 is auto-clamped to 18 upstream and maxAge ≤ minAge is auto-reset to 100.

DEPRECATED — use import_audience instead. Analyse one or more URLs and return a PREPARED target-audience profile. Does NOT persist or return an ID — forward the returned fields to create_target_audience. Trim 'residence' to ≤ 50 chars before forwarding (fieldValidation.error.max otherwise).

Duplicate an existing target audience under a new name

Upload or update the image associated with a target audience (base64-encoded)

Delete the image associated with a target audience

Record a usage event for a target audience (used for analytics/quotas)

Generate marketing or creative text content using neuroflash AI

Create a new persistent conversation session

Delete a conversation and all its messages. workspace_id is required — the service enforces workspace-scoped ownership.

Send a message in an existing persistent conversation. The tool replays the whole stored history on every call and condenses it via the service once it outgrows the context budget.

Condense a long conversation: the service summarises everything except the recent tail and stores the summary, which later chat calls send in place of the summarised messages. Nothing is deleted — the full history is still readable. workspace_id is required — the service enforces workspace-scoped ownership.

Generate an image from a text prompt using AI (text-to-image)

Edit or transform an existing image using a text prompt (image-to-image)

Delete a previously generated image from the history