From your first idea to your first saved entry. No coding needed.
A tracker is an organized place for something you want to keep up with, like recipes, car maintenance, or client payments.
Follow the screenshot walkthrough to see each step.
Keep it simple: “Track car maintenance with the date, mileage, service, and cost.” You can type or use the microphone when available.
Starting from a spreadsheet? Choose Import from CSV or Excel; no description is required. Put column headers in the first row and use a file up to 10 MB. Each usable Excel sheet becomes a section. Review the fields and import notes, then choose whether to bring in the rows too. Renaming draft sections or fields keeps the source data connected; removing a field excludes its column. If an import stops partway through, your tracker stays saved. Check the Data tab before retrying so you do not duplicate rows.
A tracker contains sections. Sections group related entries, and fields hold each entry's details. Check that the draft includes what you need. You can rename or remove draft fields before creating the tracker. After creation, you can add sections and fields; renaming or removing saved fields requires a separate data migration.
Open your tracker's Connect tab and choose an assistant. Follow that platform's guide; setup differs between hosted sign-in and manual connections. Give each additional system its own named connection. They use the same records without copying them.
Use the ChatData Sync app link, or set up Actions in your own custom GPT.
Add a connector and follow the sign-in or manual setup offered by your client.
Other compatible agents and systems connect through MCP or REST. Keep connection keys private, like passwords.
Try a small first entry: “Log an oil change today at 62,400 miles, $58.” Then ask: “Show me the entry you just saved.”
You can ask your assistant to find past entries, update a status, or total costs. Check its confirmation, especially for names, dates, and amounts. AI can make mistakes.
Use the Data tab to view or edit entries yourself and download a CSV. Owners can inspect earlier versions, recover entries deleted within 30 days, review proposed changes, and download or import a JSON backup. That backup lists attached files but does not contain their bytes; multi-page exports are live, not a point-in-time snapshot.
Choose review-before-saving for a named connection when you want to approve record changes first. Pending means not saved. You can record a due date and ask what is coming up; that does not automatically schedule a reminder.
A successful verification returns the identical server answer. If the records, layout, query, or access changed, the assistant should stop and refresh the shared summary, not silently report another answer. Ordinary questions are not automatically verified.
Missing amounts and partial coverage must be disclosed. Summaries use live records, not a frozen snapshot. Assistants can still add different wording outside the verified answer. See the optional developer reference for connection details.
Your tracker is private by default. You choose who to invite and what each named assistant connection can access. You can pause or revoke a connection from your tracker.
A tracker-specific connection accesses that tracker only. Some hosted sign-in connections can request broader access to your own trackers; review what you authorize before connecting.
The Free plan includes one tracker, 200 assistant actions per month, and 1 GB shared by entries and files. Metered actions cover record changes; reading and browsing do not use that allowance. Bulk changes count per entry.
See pricing for paid plans and add-ons, and privacy and security for processing and deletion details. Billing changes are managed through the account's subscription settings. Canceling a subscription is not the same as deleting a tracker or account.
Use the Actions-compatible OpenAPI specification for tracker-scoped calls. Send a tracker key or named assistant secret in x-api-key. Signed-in account endpoints use an account ID token; managed connections use OAuth bearer tokens.
For request examples, limits, pagination, and supported operations, open the API reference.
Endpoint: https://chatdatasync.com/api/mcp, using streamable HTTP. Capability information is at /api/mcp/info. Direct clients send x-api-key; managed clients complete OAuth sign-in. Do not paste a tracker key as an OAuth bearer token.
Record tools support reads, searches, summaries, and changes. Owner-authorized layout tools add sections and fields without deleting existing structure. Tracker-specific credentials cannot create or access other trackers. Listing across trackers and creating trackers require explicit account-wide owner access.
Use structured summarize_records over MCP or POST /api/v1/sync/summary with x-api-key. Reuse the returned query and add its answerId as expectedAnswerId. Repeat answer verbatim. A mismatch returns answer_changed (HTTP 409 for REST). Check result.complete; verification does not make a partial result complete. Heuristic intent summaries cannot verify answers.
Named assistant connections can limit operations, sections, fields, and expiry. Their secrets are shown only when created or rotated. Suspend or revoke an exposed connection, or rotate its key.
For mutation retries, reuse Idempotency-Key with the same request. For updates and deletes, send the saved revision using If-Match or expectedRevision. A stale revision returns a conflict rather than overwriting a newer edit.
Start with one thing you want to track. No credit card required.
Create your first tracker