Terms of Service

By creating a ChatSights account, or by calling the API at https://api.trychatsights.com, you agree to these terms. If you are agreeing on behalf of an organization, you confirm you may bind that organization. If you do not agree, do not use the service.

These terms may change. Material changes will be reflected in the “Last updated” date above; continuing to use the service after a change means you accept the revised terms.

ChatSights is a data-access layer. You submit a prompt and a list of AI surfaces; we route the request to our upstream provider and return the resulting answer records — answer text, citations and sources, provider record IDs and provider metadata — in a normalized envelope.

We do not interpret provider records. ChatSights does not generate scores, rankings, sentiment, comparisons, confidence intervals, recommendations or reports, and it does not judge the content of any answer. Any interpretation of the records is performed by you or by your own agent, on your own infrastructure. This boundary is not a stylistic preference: it is enforced in the API contract, which does not accept brands, competitors or analysis settings as inputs.

We do reshape the record, and you should know exactly how, because “raw” has limits. Normalization copies the provider's answer text through verbatim — we never rewrite, summarize or truncate it — and then: reads the answer from whichever field the surface used; reduces each citation to its title, URL and position; keeps five provider fields (url, prompt, model, web_search_triggered, prompt_sent_at) and drops the rest; and marks the record delivered only if answer text came back. Provider fields outside that list do not reach you.

Webhook events are operational only — they report delivery outcomes (job.completed, job.partial, job.failed) and never make a semantic claim about an answer.

Accounts are created and authenticated through Clerk, our identity provider, using email or Google sign-in. You are responsible for activity under your account.

API keys are shown in plaintext exactly once, at creation, and are stored by us only as a SHA-256 hash — we cannot recover a lost key, and you should treat one as a secret. You may revoke a key at any time; revocation stops it authenticating immediately. Keep your workspace webhook signing secret confidential and rotate it if you believe it is exposed.

Billing is metered in credits. The five rules below are invariants enforced in our code — not aspirations:

• One credit per delivered record. A run charges exactly the number of records the provider successfully returned. One prompt sent to three surfaces costs at most three credits.
• Failed records charge zero. A surface that fails delivers no record and consumes no credit. A run where nothing arrives is billed nothing at all.
• Demo records charge zero. Runs in demo mode — where the provider is unconfigured, or the calling key has test scope — are labelled "mode": "demo" and never bill.
• The free tier hard-stops. Once a live run would cross the free tier's included credits, it is refused before any provider call with a 402. It does not silently accrue a bill.
• Paid tiers do not hard-stop. Unlike the free tier, a paid plan does not refuse a run for crossing its included credits — the allowance wall is skipped for any plan with an overage rate. Your account spend cap and per-key budgets are therefore your own brakes on a paid plan, and those are enforced before any provider call.

Overage itself is the one part of this that is not automated, and we would rather say so than let the sentence above imply otherwise. Records beyond your included credits are metered and visible to you, but no system invoices them: as section 5 explains, ChatSights has no payment processor. What you actually owe for a paid plan, overage included, is whatever we have agreed with you in writing. The per-credit rates on the pricing page are the basis for that conversation, not a charge that happens on its own.

Current prices, included credits and overage rates are on the pricing page; the mechanics are documented in the developer docs. Your consumption is visible to you at any time via GET /v1/usage and in the console.

Plans are not self-serve. The API deliberately refuses to change your plan: PATCH /v1/settings rejects plan_id with a 422. Every account starts in the free workspace with its included credits and no card. To move to a paid tier, you email keronelau@gmail.com and the change is arranged with you directly and applied to your workspace by hand.

Because there is no automated checkout, there is no automated subscription, renewal, cancellation or refund mechanism to describe here. Payment terms, invoicing and any refund for a paid arrangement are whatever we agree with you in writing at the time. To stop using a paid plan, or to close your account, email keronelau@gmail.com.

You agree not to use ChatSights to:

• break any applicable law, or infringe anyone's rights;
• submit prompts containing other people's personal data, credentials or confidential information — prompts are transmitted to a third-party provider (see section 7);
• attempt to circumvent the billing meter, the plan allowance wall, the spend caps or the per-key budgets, or to access a workspace that is not yours;
• probe, load-test or disrupt the service or its upstream providers beyond your plan's limits;
• present raw provider records in a way that implies ChatSights, or any AI surface operator, authored, endorsed or verified a conclusion you drew from them.

Records you retrieve originate from third-party AI surfaces and may contain material subject to the rights of others. You are responsible for your own use of what you retrieve.

ChatSights does not operate any AI surface. To fulfil a live run we transmit your prompt text, and your country and language hints, to our upstream data provider, Bright Data, which retrieves the answer from the requested surface and returns the record to us. Your prompt text therefore leaves our systems and is processed by a third party.

We also rely on Clerk for identity and Cloudflare for hosting, compute and data storage.

Do not submit anything through ChatSights that you are not willing to have transmitted to these providers.

The service is provided “as is”, without warranties of any kind to the fullest extent permitted by law.

In particular: ChatSights depends on third-party AI surfaces and an upstream provider that we do not control. We do not warrant the availability, accuracy, completeness, timeliness or truthfulness of any provider answer, citation or source. AI surfaces are non-deterministic and change without notice — the same prompt may return different answers, or no answer, at different times. A record reflects only what the provider returned at that moment; we neither verify nor endorse its content. We do not warrant that the service will be uninterrupted or error-free, and no plan on the pricing page carries a service-level agreement unless we have agreed one with you in writing.

To the fullest extent permitted by law, neither party is liable for indirect, incidental, special, consequential or punitive damages, or for lost profits or lost data, arising out of these terms or your use of the service.

Our total aggregate liability arising out of or relating to the service is limited to the amounts you actually paid us for the service in the 12 months preceding the event giving rise to the claim.

You may stop using the service at any time; to close your account, email keronelau@gmail.com. We may suspend or terminate access for a breach of section 6, or where required to protect the service, its users or our upstream providers.

ChatGPT is a trademark of OpenAI. Google, Gemini and Google AI Overviews are trademarks of Google LLC. Perplexity is a trademark of Perplexity AI, Inc. Copilot is a trademark of Microsoft Corporation. Claude is a trademark of Anthropic PBC. All other names, logos and marks are the property of their respective owners.

ChatSights is not affiliated with, endorsed by, sponsored by or otherwise connected to any of these companies. Their names and logos appear on this site as nominative reference and nothing more, in one of two roles: naming a publicly available AI surface a record can be retrieved from (ChatGPT, Perplexity, Gemini, Google AI Overview, Google AI Mode, Copilot), or naming a developer tool that can call ChatSights over MCP (Claude Code, Cursor, Codex). Anthropic appears only in the second role. Claude is not one of the surfaces this service fetches from, and no record is ever retrieved from it.

ChatSights is an independent project operated by an individual developer. There is no company, no office and no postal address to write to — all contact is by email.

• General, billing, support and security: keronelau@gmail.com

How we handle personal data is described in our privacy policy.