unquestion

unquestion is an AI-powered conversation builder designed to replace static forms with adaptive dialogue. A creator arranges questions, publishes the resulting conversation through a public URL or embeds it on a site, and the AI can respond conversationally and ask follow-up questions while the platform records the resulting answers in structured tables.¹²

The current homepage positions the product around customer feedback, newsletter signup, event registration, demo requests, cancellation surveys, product surveys, feature requests, waitlists and customer support. Results can be monitored as complete or partial responses and exported as CSV.¹

The template library reinforces that positioning. Current templates span product, office, personal and marketing workflows, including customer support, event registration, tourist Q&A, newsletter signup, demo qualification, waitlist signup, customer feedback and feature requests.³

The core product idea is stronger than generic “AI forms” branding suggests. Traditional forms are predictable but rigid; open chat is flexible but produces messy transcripts. unquestion tries to combine adaptive dialogue on the respondent side with structured data on the operator side.

This is a desk review. AiToolMap did not create an account, build or publish a form, embed a conversation, submit responses, measure completion rates, test branching/follow-up behavior, inspect exports, test the MCP server or compare output quality against Typeform, Tally, Fillout or another form builder.

unquestion’s strongest product decision is to preserve structure. The respondent gets a more natural interaction, but the operator does not have to analyze a long free-form chat log after every submission. The homepage explicitly promises structured answers in clean tables, real-time visibility into completed and partial responses and CSV export.¹

That makes the product more directly useful for operational workflows than a generic embedded chatbot. A feedback conversation can still end as rows that can be filtered, analyzed or moved into another system.

Trend Hunter’s current exact-product coverage describes the same thesis: static forms can feel repetitive, while unquestion adapts its conversation to respondent answers and can ask follow-ups to obtain more detail; the collected output is then presented as structured data rather than raw conversation transcripts.⁶

That editorial coverage confirms the product positioning, but it is not a hands-on comparative test and does not provide a reliability or satisfaction score.

The current Terms state that AI inference is provided by Cerebras and that the provider may change in the future.² The Privacy Policy is even more explicit: all inputs submitted to conversational forms are sent to Cerebras for inference.⁴

AiToolMap found no current first-party unquestion documentation identifying the exact underlying Cerebras-served model or model version. It would therefore be incorrect to import a benchmark score from any particular Cerebras-hosted model and present it as unquestion performance.

This distinction matters because a conversational-form product should ultimately be evaluated on the behavior experienced by respondents: relevance of follow-up questions, instruction adherence, tone, latency, safety, consistency and extraction quality. Those properties depend on prompts and orchestration as well as the foundation model.

unquestion’s Privacy Policy says that, as of July 4, 2026, Cerebras does not use customer API data to train its models.⁴ Cerebras’s own current cloud/privacy material is directionally consistent: it says inference inputs and outputs are not retained as described in its policy and its Terms do not grant Cerebras the right to train or fine-tune models on Service Content.⁵

That is a useful cross-check, although customers should still treat Cerebras’s own current terms as authoritative for the downstream processor relationship.

The Terms state plainly that conversational responses are generated by AI without human review and warn that they may be inaccurate, false, inappropriate, biased, inconsistent or outdated.²

This disclosure is important because the product is designed to be placed in front of third parties. A hallucination inside a private drafting assistant affects the operator; a hallucination inside a published customer-support or registration conversation can directly affect a customer, prospect or participant.

The Terms therefore place responsibility on the creator to test conversations before publishing, monitor AI behavior, adjust system prompts and ensure suitability for the use case.²

That is a reasonable allocation for a young builder product, but it means unquestion should not be deployed as an unsupervised source of medical, legal, financial or other high-stakes advice. The Terms explicitly say its AI output is not professional advice.²

The Terms say unquestion provides an MCP server through which authorized third-party AI clients can interact with the account after an OAuth consent flow.²

This can be useful for creators who want an agent to inspect workspaces, forms or conversation results without manually exporting CSV files.

It also adds another trust boundary. Once a third-party AI client receives account data, that client’s own privacy, retention and security rules apply.⁴²

For organizations, MCP access should therefore be treated as a privileged integration rather than a convenience toggle. Only approved AI clients should be authorized, and revoked clients should be removed promptly when no longer needed.

The Terms say unquestion does not offer subscriptions. Instead, credits are sold as one-time packs.²

One credit is consumed per conversation. Credits are non-transferable, have no cash value and expire 12 months after purchase. Credits and fees are non-refundable except where law requires otherwise.²

The homepage says a user can create a first conversation for free.¹

AiToolMap did not locate current public dollar prices for the credit packs in the pages accessible during this review cycle. Those prices may be exposed only after sign-in or in checkout. They are therefore not inferred or copied from snippets of uncertain provenance.

The pricing model itself is attractive for intermittent use because there is no recurring subscription commitment. Its value cannot be fully assessed without the current pack prices, however.

The one-credit-per-conversation rule is also unusually simple compared with token-based AI pricing. Buyers should confirm exactly what constitutes a “conversation” for credit consumption, especially for partially completed sessions, retries and embedded workflows.

AiToolMap audited all 50 members of `ai-review-panel-2026-08-v4` using exact-domain searches anchored to `unquestion.ai`. That constraint mattered because “unquestion” is an ordinary English word and broad searches would otherwise create false positives.

None of the 50 fixed-panel sources produced current exact-product evidence. The model benchmark sources are also a poor match: unquestion is an application whose quality depends on conversation design, Cerebras-served inference, data capture, structured output, publishing and embedding behavior. A benchmark score for an underlying model family would not measure those product-level properties.

The absence of fixed-panel coverage is therefore treated as missing evidence rather than a negative score.

Beyond the panel, AiToolMap substantively read the current homepage, templates, Terms and Privacy Policy, current Cerebras privacy/terms material relevant to its inference service, and an August 21, 2026 Trend Hunter article describing the exact product.¹³²⁴⁵⁶

When a conversation is published, the Terms say it becomes accessible through a unique public URL without authentication until the creator unpublishes it; it can also be embedded on external websites.²

The Privacy Policy draws a useful controller/processor distinction. For account and operational data, Unquestion LLC acts as controller. For end-user responses collected through a customer’s published conversations, the customer is the controller and unquestion acts as processor.⁴

That means a business embedding an unquestion conversation is responsible for its respondent-facing notice, lawful basis/consent where required, and the purpose for which it collects the answers.

The Terms prohibit using published conversations to collect HIPAA-regulated protected health information, payment-card information, Social Security numbers/government IDs, data from children under 13 without verifiable parental consent, or data the customer is not legally authorized to collect.²

This is a meaningful guardrail, but it also narrows the use cases. A company should not treat unquestion as a generic secure intake form for any category of sensitive data.

The Privacy Policy was last updated July 4, 2026 and identifies the operator as Unquestion LLC, a Wyoming LLC.⁴

It lists the principal data categories: account information, conversation designs/system prompts, respondent text and interaction metadata, IP address, browser/device information, analytics and diagnostic data.⁴

It also names major subprocessors and their roles: Supabase for hosting/database, Stripe for payments, Loops for transactional email, PostHog in cookieless mode for product analytics, Langfuse for AI performance monitoring and Upstash for rate limiting.⁴

All conversation inputs are sent to Cerebras for AI inference.⁴ The policy says services and providers process data in the United States and potentially other countries, so non-US users should expect international transfer to the United States.

The policy says it does not use advertising cookies, third-party tracking pixels or cross-site behavioral tracking, and that PostHog runs in cookieless mode.⁴

For MCP, unquestion exposes an OAuth-authorized server that can let compatible third-party AI clients access account data such as workspaces, forms and conversations. The Privacy Policy and Terms both warn that once data is transmitted to the authorized third-party client, that client’s handling is outside unquestion’s responsibility.⁴²

For active accounts, the Privacy Policy says account information, forms and conversations are retained while the account remains active.⁴

After account deletion, account information is deleted immediately, while conversation data and responses are deleted within 90 days. Backups and logs may persist for up to 180 days, while billing/transaction records may be retained as required by law, often seven years.⁴

The Terms repeat the broad account-deletion structure: account information is deleted immediately and conversation data within 90 days.²

That is reasonably transparent. Businesses should nevertheless understand that deleting the operator account does not mean respondent content disappears from every backup immediately.

The Privacy Policy provides access, correction, deletion, export and objection rights and says requests are answered within 30 days.⁴

The Privacy Policy describes HTTPS/TLS in transit, secure authentication, access controls, regular security monitoring and webhook signature verification.⁴

AiToolMap did not locate a separate public security/trust center, SOC 2 report, ISO certification, public penetration-test summary or detailed encryption-at-rest/key-management specification during this review cycle.

That does not establish that the implementation is weak. It limits procurement confidence. A public conversation builder can hold customer-created prompts and respondent data, and enterprise buyers will normally want more than a generic list of security measures.

The Terms also reserve the right to implement or modify rate limits and disclaim guaranteed uptime or uninterrupted availability.² No independent uptime/reliability record was located.

unquestion is best suited to small teams, founders, product managers and marketers who want an adaptive alternative to a static survey/form but still need structured data at the end.

Good candidate workflows include customer feedback, feature requests, demo qualification, cancellation interviews, event registration, waitlists and low-risk customer-support intake.¹³

It is less suitable for regulated or highly sensitive intake, high-stakes advice, workflows that require deterministic question wording, or enterprises that cannot adopt a vendor without stronger assurance artifacts.

Before deploying a public conversation, teams should test representative and adversarial respondent inputs, inspect how follow-ups behave, set clear respondent notices, avoid prohibited sensitive data, confirm current credit-pack prices, document deletion/retention requirements and review any downstream MCP client separately.