Everything you need to know about ChatData Sync private, governed structured memory — how it connects to your AI assistant, what it costs, and how you control access.
Getting started
ChatData Sync gives AI assistants private, governed structured memory for real-world records. Describe what you need to track in plain English, and it creates an organized tracker your assistant can safely read, update, search, and summarize with access you control.
No. You describe what you want to track in plain English, and ChatData Sync builds the structure for you. Connecting your AI assistant is a copy-and-paste step that takes about thirty seconds. If you are a developer and prefer to interact with the underlying connection directly, that option is available — but it is never required.
ChatData Sync is best for real operational records you want an assistant to remember across chats. Popular examples include client appointments, tutoring sessions, pet-care visits, home repair jobs, freelance project milestones, fitness goals, and vehicle maintenance records.
A strong vertical proof point is auto repair: customers, vehicles, repair orders, inspections, parts, estimates, approvals, invoices, and technician notes managed through assistant conversations.
Yes. The Free plan gives you one tracking space and 200 assistant actions per month with no credit card required. It is a fully working version of the product — a good way to see ChatData Sync in action before choosing a paid plan.
Yes. The setup wizard has a microphone button that transcribes what you say (using OpenAI Whisper) straight into the description box, so you can talk through what you want to track. You still need a short written or spoken description before you import a CSV/Excel file or build a tracker by hand — that description is what teaches both the generator and your assistant what the tracker is for, and the wizard will remind you if it is missing.
Yes, and your existing records stay safe. In the tracker's layout editor you can add new sections (tables) and fields up to your plan limits, change a field's type, and add a link field that relates one section to another — for example, a Payments section that links to a Clients section.
Your connected assistant can make the same additive changes by conversation through the add_section and add_field tools (and read the current layout with describe_schema). These tools only add structure; they never delete sections, fields, or data. Removing or renaming structure stays an owner-only action in the app so nothing is lost by accident.
AI assistants and connections
ChatData Sync provides an MCP-backed capability prepared for plugin distribution across ChatGPT and Codex. Public Plugins Directory availability begins only after OpenAI approval and publisher release. Existing custom GPTs can continue using the supported GPT Action connection path. Both routes connect to the same private tracker and governed access controls.
A ChatGPT Action is the supported direct connection path for an existing custom GPT. It tells that GPT how to reach your private ChatData Sync tracker. OpenAI now uses plugins as the primary discovery format across ChatGPT and Codex, but Actions remain available for backward-compatible custom-GPT setups.
MCP stands for Model Context Protocol. It is an open standard that allows AI assistants — including Claude and other compatible tools — to connect to structured data sources in a consistent, secure way. ChatData Sync provides an MCP endpoint for every tracking space, so compatible AI tools can read and update your information without any custom code on your end.
ChatData Sync exposes each tracker two ways, so it works with almost any assistant:
https://chatdatasync.com/api/mcp over streamable HTTP, then authenticate with a named assistant secret or the legacy tracker key in the x-api-key header — or sign in with OAuth and let the client carry a bearer token. The server offers tools to create, list, search, summarize, fetch, update, and delete records, plus additive layout tools (describe_schema, add_section, add_field) whose access is governed by the connection policy.https://chatdatasync.com/openapi.actions.json (REST base https://chatdatasync.com/api/v1) and use a named assistant secret or the legacy tracker key.Named connections can be limited to selected operations, tables, fields, and an expiry date. They can be suspended, rotated, or revoked independently. Step-by-step guides: connect Claude via MCP and the MCP server reference.
For direct MCP setup, use a named assistant secret or the legacy tracker key in a custom header named x-api-key. This is the normal copy-and-paste setup for clients where you enter the MCP URL and headers yourself.
Use bearer tokens only when the host completes ChatData Sync OAuth for you, or when you are calling signed-in account APIs instead of the tracker-scoped MCP connection.
If the screen asks for a header name, enter x-api-key. If the screen asks you to sign in with ChatData Sync, use that OAuth flow and let the host send bearer automatically. A named assistant secret is shown only when its connection is created or rotated.
Yes. The same REST endpoints the dashboard uses are documented and available to scripts. Sign in once with the Firebase Web SDK and send your ID token as Authorization: Bearer.
The project-management surface covers the full lifecycle: POST /api/v1/projects creates a tracker (the response includes its apiKey), GET /api/v1/projects lists everything you own, GET /api/v1/projects/{projectId} returns the full schema, PATCH /api/v1/projects/{projectId} evolves it, POST /api/v1/projects/{projectId}/items:import seeds rows in batches of 500 without consuming assistant actions, and DELETE /api/v1/projects/{projectId} removes it.
Per-tracker record caps apply the same way they apply to chat-driven writes, and non-owners receive 404 to prevent enumeration. See the API documentation page for an end-to-end example script.
Yes. Using a named assistant secret or the legacy tracker-scoped x-api-key header, GET /api/v1/sync/{projectId}/{table}/{recordId} returns a single record, PATCH on the same path merges the supplied fields into the existing record (use PUT to replace them), and DELETE removes the record and decrements the tracker's record count. Named connections can limit which tables, fields, and operations are available.
To stop scripted writes from creating duplicates, you can declare an identity_key on any table in the schema — the field that uniquely identifies a row, such as sku, email, or external_id. When that key is set, POST /api/v1/sync/{projectId}/{table} performs an upsert whenever the incoming payload includes a non-empty value for it, updating the existing row instead of appending a copy. Without an identity_key, POST stays append-only.
Where-filter values are coerced to the field's declared type, so digit-only values such as HTS codes, SKUs, or ZIP codes work on text fields: ?where=htscode:eq:731010. Send the same Idempotency-Key when retrying a mutation, and send If-Match or expectedRevision for an update or delete when you need stale-write protection.
Comparison with other tools
Airtable requires you to design your own tables, columns, and field types before you can track anything. ChatData Sync generates the structure from your plain-English description, so you go from idea to working tracker in minutes. It is built specifically so your AI assistant can read and write your data with scoped access.
ChatData Sync is not trying to replace Airtable. It solves a different problem: private structured memory for assistants.
Notion is built for writing, notes, and wikis. It is not designed to be a live, structured memory layer that your AI assistant can update on command. ChatData Sync is purpose-built for that job: organized, queryable tracking with strict access controls per space.
ChatData Sync is not trying to replace Notion. It is the assistant-memory layer teams and solo operators add when chat workflows need reliable records.
Pricing and plans
ChatData Sync offers five plans:
Need more room without changing plans? Add-ons stack on any plan: Record Boost (+10,000 records per tracker, $5/month), Action top-up (+500 actions, $5 one-time), and Action add-on (+1,000 actions/month, $7/month).
All plans include CSV export. See the full pricing page for details.
If you reach your monthly assistant action limit, your AI assistant will no longer be able to read or update your tracker until the limit resets at the start of your next billing cycle. Your data is always safe and accessible through your dashboard. You can upgrade your plan, or stay on your plan and add more: a one-time Action top-up (+500 actions for $5) or the recurring Action add-on (+1,000 actions/month for $7). If a tracker runs out of record room, add a Record Boost (+10,000 records per tracker for $5/month).
Yes. Each tracker has a hard cap on records (rows): Free 500, Single Tracker 5,000, Side Hustle 15,000, Growing Business 30,000, Team 50,000. When a tracker reaches its cap, new writes from chat or the API return a clear record limit exceeded message. You can delete older entries to make room, upgrade to a higher plan, or add a Record Boost — a $5/month add-on that adds 10,000 records to every tracker. Stack as many as you need, up to a hard maximum of 50,000 records per tracker.
Yes. On the Create tracker screen, choose Upload a CSV or Excel file. ChatData Sync reads the column headers to design the tracker, and on the review step you can leave the “Also bring in the rows of data from your file” option checked to import the rows themselves. Excel workbooks with multiple sheets become one section per sheet.
We support up to 50,000 rows per sheet, subject to your plan’s per-tracker record limit. If you hit that limit during import, the tracker is still created with the rows that fit and we tell you exactly how many landed so you can decide whether to upgrade.
Yes. CSV export is included on every plan, including the free tier. Your data belongs to you, and you can download it at any time from your dashboard.
Privacy and security
Yes. Every tracking space is tenant-isolated, and access is limited to the owner, explicitly shared collaborators, and assistant credentials authorized for that tracker. Named connections can restrict operations, tables, fields, and expiry, and can be suspended or revoked independently. Your data is never shared between unrelated users, never used to train AI models, and never visible to other ChatData Sync accounts.
Named assistant secrets are shown only when a connection is created or rotated, and the server stores only a hash and a hash-derived fingerprint. Store the secret in the assistant setup immediately. The legacy tracker key is masked by default on the Connect screen. If any credential is exposed, suspend or revoke the named connection, or rotate the project key, and the old credential stops working.
Support and feedback
Use the built-in Issue Tracker in the app. Open it from your dashboard to post a bug report or feature request, follow its status as it moves from new to in-progress to resolved, and add comments on the thread. It is the fastest way to reach us and to see what other improvements are already underway.
Start with a free tracking space — describe what you need to track and your assistant handles the rest.
Start free — no credit card required