Privacy Policy — Connie

Conigma, Inc., operating as Connie ("Connie," "we," "us," or "our") provides the Service. This Privacy Policy explains what information we collect, how we use and share it, how long we keep it, and the choices and rights you have.

By using the Service, including by installing or connecting the Connie app for Slack or Zoom, or by connecting the Connie MCP server to an AI assistant, you agree to the practices described in this policy.

Conigma, Inc.
1111B S Governors Ave # 43331
Dover, DE 19904, USA

This policy applies to:

• Users who create a Connie account and use the web application.
• Workspaces and their members who install or connect the Connie app for Slack or Zoom.
• Users who connect the Connie MCP server to an AI assistant or agent, such as Claude, whether by adding it as a custom connector or by installing it from a connector directory.
• Visitors to our website.

If you use Connie through an organization, your organization's administrator may control certain settings and data, and your use may also be subject to your organization's own policies.

Account and profile information. When you create an account, we collect information such as your name, email address, and authentication details.

Using Connie without an account. When you ask for a sign-in email while using Connie without an account, we keep a one-way hash of that email address and of a one-time code with your guest work, so that only the account with that address, signed in from the email's link, can add the work in another browser. The hash is removed when the work is added or a later email replaces it.

Connie MCP server and AI assistant connectors. The Connie MCP server lets an AI assistant you choose — for example Claude — read from and act on your Connie workspace on your behalf. Connecting it is always something you initiate, and it operates only within the workspace and permissions attached to your own Connie account.

• Authorization data. Connecting the MCP server runs an OAuth 2.0 authorization flow against Connie. Connie issues the access and refresh tokens that represent your authorization. We do not store the plaintext tokens themselves — we store a hash of each token together with its metadata: the associated workspace and user identifiers, the granted scopes, expiry, and the identity of the client that requested access. This metadata is retained on the schedule in the retention table in Section 6.
• Tool call requests. When the AI assistant invokes a tool, we receive the arguments needed to perform that operation — for example the identifier of a record to read, the fields to write, or the file being imported. We record request metadata, including the tool invoked, timestamps, the workspace and client involved, and the outcome, so that the operation can be executed, audited, debugged, and rate-limited. We do not log the arguments themselves or the data a tool returns. Where a tool call consumes credits, the corresponding credit usage record (the tool, the time, the user who authorized the connector, and the credits used) is kept as part of your workspace's billing history for the life of the workspace.
• Workspace data returned to the assistant. Tools return data from your own Connie workspace, such as records, list contents, message threads, or flow results. This data leaves Connie and enters your conversation with the AI assistant, and is then handled by that assistant's provider under their own privacy policy and terms. Connie does not control the retention or use of data once it has been returned into a conversation on a third-party AI platform. Only the data a tool is asked for is returned; the connector does not push your workspace contents to any assistant on its own initiative.
• Data you send in through the connector. Where a tool accepts data — importing a CSV of contacts, creating or updating records, drafting or sending a message — the content you supply through that tool is stored in your workspace in the same way as data entered through the web application, and is covered by the same retention rules.

What the connector does not do. The connector does not receive, request, or store the contents of your conversation with the AI assistant beyond the arguments explicitly passed to a tool. It does not access the assistant's memory, chat history, conversation summaries, or files you have uploaded to that assistant. It does not read data belonging to Connie workspaces other than those your account is authorized for.

Slack data (when you connect Slack). The Connie app for Slack lets authorized users work with Slack through Workflows, Connie AI Chat, Agents, and the Slack-native Connie teammate. With your authorization through Slack's OAuth flow, and only within the permissions granted at install time, we process Slack data as follows:

• Connection, workspace, and directory data. We receive identifiers for the Slack workspace, app, connected account, channels, messages, and users, plus member profile fields such as names and work email addresses where Slack makes them available. We use this information to route events to the correct Connie workspace, enforce the selected account's permissions, resolve mentions and recipients, and attribute actions. Slack credentials are held by our integration connection provider; Connie stores the connection reference and the account and workspace metadata needed to operate the integration.
• Workflow events and Slack-triggered automation. Connie subscribes to configured Slack events, including messages in channels the app belongs to and messages addressed to Connie, to power triggers such as "new message in a channel" and "app mentioned." A stored event record can contain the event type and timestamps; Slack team, channel, message, thread, file, and user identifiers; and the event payload, which can include message text and file metadata. Connie uses these records to evaluate and run configured Workflows and Agents, test triggers, prevent duplicate delivery, and diagnose failures.
• The Slack-native Connie teammate. When someone mentions Connie in a channel or thread, continues a thread in which Connie is active, or sends a message to the Connie app, we process that message, the relevant thread context, permitted attachments, and participant profile information so Connie can understand and answer. The invoking message, Connie's response, and the tool activity needed to produce it can be saved in a Connie conversation so the user can continue the exchange and authorized workspace members can review the Agent's work.
• User-initiated and task-specific reads and search. At a user's request, or when a configured Workflow or Agent needs the data for its current task, Connie can read permitted Slack messages, threads, channel details, reactions, profiles, and files. User-initiated Slack search uses Slack's Real-Time Search API for the current query. Connie does not scrape Slack in the background or create a separate persistent copy, archive, or search index from those results. The relevant result can appear in the AI Chat, Agent conversation, Workflow result, or Record where the user asked Connie to use it.
• Files and attachments. Connie can receive file metadata in an event and, when a user, Workflow, or Agent asks to use an accessible file, temporarily download its contents from Slack to complete that task. The downloaded binary is not retained as a separate Slack file archive. Extracted text or a result derived from the file can be included in the saved Connie conversation or workspace output for which it was requested.
• Slack actions. When an authorized user, Workflow, or Agent asks Connie to act in Slack, we process the destination, message or file content, thread and user identifiers, and the result needed to perform and record actions such as posting or updating a message, sending a direct message, adding a reaction, or managing a permitted channel or user group.
• AI processing. When a Slack request uses an AI feature, the Slack content needed for that request is sent through OpenRouter to the selected AI model provider to produce a reply, summary, classification, draft, or tool decision. As described in Section 3, neither Connie nor its AI providers use Slack content to train AI models.
• Fields received but not used as a product feature. Slack event and API responses can include identifiers or other fields that the requested feature does not display. We do not use those fields to contact Slack users, build advertising profiles, train models, or benefit another customer. If they are part of a stored event payload, they are deleted with that event; temporary response fields are discarded after the request.

Access and tenancy. Slack reads and actions are limited by the selected Slack account, Slack's channel and workspace permissions, the connection's access setting in Connie, and the person's Connie workspace permissions. Private-channel content is not exposed to someone who cannot access that channel in Slack. We do not use one customer's Slack data to benefit another customer or make it searchable across customers.

What Connie does not provide. Connie is not a general Slack inbox, Slack client replacement, bulk export or backup service, or persistent Slack search index. Connie does provide user-initiated, task-specific Slack search and message or thread reads in AI Chat and Agents, and shows Slack content that a user or automation deliberately includes in a saved Connie workspace item.

Retention, disconnect, and uninstall. Stored Slack automation event records are kept while the integration account is connected and are deleted when that account is disconnected or removed from Connie. Disconnecting or uninstalling stops new Slack access, but it does not by itself delete Slack content that was already included in a saved AI Chat, Agent conversation, Workflow result, or Record; that content follows the retention of the saved workspace item. Removing the app in Slack revokes its Slack access. To remove the Connie connection and its account-scoped stored data, also disconnect the account in Connie or contact support. The retention table in Section 6 and the controls and request process in Section 7 explain how to remove saved items, a connection, a workspace, or personal information.

Storage location and international processing. Stored Connie workspace data follows the storage location described in Section 5. Slack, our integration connection provider, OpenRouter, and the AI model provider used for a request may process data in other regions, as described in Section 9.

Zoom data (when you connect Zoom). With your authorization through Zoom's OAuth flow, and only within the permissions you grant, we process Zoom data to provide meeting-management features in your workflows: creating, updating, deleting, and viewing Zoom meetings, and reading your Zoom user profile to associate meetings with your account. OAuth tokens for your Zoom connection are held by our integration provider and are not stored in our own database; we store only the references needed to operate the integration, such as meeting identifiers and your Zoom user ID.

Meeting recording and call intelligence. If you enable call recording, Connie uses a third-party meeting bot to join your scheduled or live meetings on supported platforms (including Zoom, Google Meet, and Microsoft Teams), record audio/video, and generate a transcript. The recording and transcript are processed to produce summaries, action items, and CRM updates. Meeting recordings are retained per the schedule in the retention table below (configurable per workspace) and are deleted when you delete them or close your workspace.

Other connected integrations. If you connect other third-party accounts, we process the data from those services that is necessary to provide the connected features. Where a feature enriches your records using an external data provider, we send that provider only the fields required to perform the lookup and store the result against your record.

Information about other people (lead search and contact enrichment). Some features let a Connie user find business contacts, or add details to the people and companies already in their workspace. Depending on the feature, the result can include a person's name, job title, employer, work or personal email address, phone number, location, and professional profile information, such as a LinkedIn profile and its public posts.

• Where it comes from. Third-party business contact and company data providers that we query on the user's request, and publicly available sources such as company websites and public professional profiles.
• When it is collected. Only when a user, or an automation or AI assistant acting for them, runs a lead search or an enrichment. We send the provider only the fields needed for that lookup, for example a name and company domain, or a profile URL.
• Where it is kept. Only in the workspace of the user who requested it, where it is covered by the retention rules in Section 6. Connie does not compile or keep a separate database of people across its customers, and we do not sell or rent this information.
• Roles and legal basis. The Connie customer who looks up or stores this information decides why and how it is used and is its controller; Connie processes it on the customer's behalf to provide the Service. Customers must have a lawful basis for their own use, which for business-to-business outreach in the EU and UK is typically legitimate interests (Article 6(1)(f) GDPR and UK GDPR), together with the notice and objection rights that come with it, and must follow applicable direct-marketing rules. The data providers we query are independent controllers of the data they compile, under their own privacy policies.

If this describes information about you and you are not a Connie user, Section 7 explains how to ask for it to be removed.

Billing information. Payments are processed by our payment provider (Stripe). We receive billing-related event records but do not store full payment card numbers ourselves.

Usage and technical data. We collect log and usage data such as actions taken in the Service, device and browser information, and similar technical information.

We use the information we collect to:

• Provide, operate, and maintain the Service and its features.
• Evaluate and run the automations and flows you configure.
• Operate the Connie MCP server, authenticate connector sessions, and execute the tool calls an AI assistant makes on your behalf.
• Enforce the scope of a connector authorization, so that an AI assistant can only reach the workspaces and data your account is entitled to.
• Maintain an audit record of connector activity for security, debugging, abuse prevention, and rate limiting.
• Create and manage Zoom meetings on your behalf as part of your workflows.
• Record, transcribe, and analyze meetings you choose to record, and generate summaries, action items, and CRM updates from them.
• Power AI-assisted features such as drafting, summarization, and research.
• Process payments and manage subscriptions.
• Communicate with you about the Service, including support and important notices.
• Measure how people find Connie and how well our marketing works, including which posts, links and ads lead to sign-ups, trials and purchases.
• Monitor, secure, debug, and improve the Service.
• Comply with legal obligations.

We do not use your data, including your Slack content and data accessed through the MCP server, to train AI models. Our own AI features are powered by third-party providers accessed through OpenRouter, under provider terms that prohibit training on the content we send. Where you connect the MCP server to a third-party AI assistant, that assistant's provider determines how the data returned into your conversation is handled, under their own policies.

We do not sell your personal information. We share limited information with advertising partners to measure and target our ads; a Global Privacy Control signal opts you out. We share information only as follows:

• Service providers who process data on our behalf to operate the Service, including our database and hosting provider (Supabase), our integration connection provider, our meeting-recording provider, our payment provider (Stripe), and our AI model routing provider and the AI model providers it routes to. We engage these providers as vendors under our own contractual, security, and data-protection controls.
• Analytics and advertising measurement. PostHog, our product analytics provider, processes usage and account events (such as sign-ups, trial starts and payments) on our behalf. When you sign up, start a trial, are offered a trial or subscribe, we send LinkedIn a hashed (scrambled) copy of your email address and, if you came from a LinkedIn ad, LinkedIn's click ID, so we can measure our LinkedIn ads. When you sign up, start a trial, see a trial offer or make your first payment, we also tell Meta and Google about it from our servers so we can measure our ads. Both get a hashed (SHA-256) copy of your email address, the ad click ID that brought you (if any), and your IP address and browser user agent from that visit. Meta also gets a hashed copy of your account ID and your country, and Meta's _fbp and _fbc cookies from that visit. For a trial or your first payment, both also get the plan's price (and, for a trial, Meta gets the plan's name and billing period). We keep that IP address and browser information for 30 days after your last sign-up, trial, trial offer or payment step, only to match these conversions. We don't do any of this if your browser sends a Global Privacy Control signal.
• AI assistants and agents you connect. When you connect the Connie MCP server to an AI assistant, the workspace data returned by a tool is disclosed to the provider of that assistant as part of your conversation with it. This happens only in response to a tool call, and only for the data that tool returns. These providers are not our vendors and do not process that data on our behalf; their handling of it is governed by their own privacy policy and terms, which you should review before connecting.
• Contact and company data providers. When a user runs a lead search or enrichment, we send the data provider only the fields needed for that lookup, as described in Section 2.
• At your direction, with third-party services you choose to connect.
• For legal reasons, where required by law or to protect rights, safety, and the integrity of the Service.
• In a business transfer, such as a merger, acquisition, or sale of assets, subject to this policy.

We retain information for as long as needed to provide the Service and for the periods described below, after which it is deleted or purged automatically:

Depending on your location, you may have rights to access, correct, export, or delete your personal information, and to object to or restrict certain processing. To exercise these rights, contact us at the address below. We will respond consistent with applicable law.

If you are not a Connie user. If you believe a Connie customer found or stored information about you using lead search or contact enrichment, you can ask for it to be removed. Email support@connie.ai with the email address, phone number, or profile URL concerned. We will identify the workspaces that hold information matching it, pass your request to each customer concerned, and help them delete it. Because that information is held for our customers, they decide on and answer requests about their own use of it, including objections to marketing. You can also contact the business that reached out to you directly, and the data providers that supply such information, which handle requests about their own records.

Disconnecting the Connie connector. You can disconnect the MCP server from your AI assistant at any time, and you can revoke the authorization in Connie under Tools > My accounts. Revocation is enforced on every request the connector makes, so the assistant's next call is refused and it can no longer read from or act on your workspace; a call already in progress at the moment of revocation completes. The stored tokens are then purged as described in the retention table above. Disconnecting the connector does not delete data already returned into a past conversation on the assistant's platform — to remove that, use the controls your AI assistant provider offers.

Disconnecting Slack. You can remove the Connie app from your Slack workspace at any time from your Slack workspace's app management settings, and disconnect or remove its account under Plugins > Connectors in Connie. Removing the app in Slack revokes its Slack access; disconnecting the account in Connie also deletes the connection's account-scoped event records. Slack content already included in a saved AI Chat, Agent conversation, Workflow result, or Record remains with that item until you delete the item or workspace, or ask us to delete personal information as described above.

Disconnecting integrations. You can disconnect Zoom or other connected accounts at any time from Connie's Tools page, and the Zoom app from Zoom's App Marketplace under Manage > Installed Apps. When an integration is disconnected, the associated stored data is deleted as described in the retention table above. Synced inbox data for a disconnected email or messaging account is kept for 7 days so that reconnecting restores your history, and is then permanently purged.

Deleting a workspace. A workspace owner can delete a workspace from its settings. Everyone loses access to it immediately, its subscription is cancelled, and all of its data is permanently purged 7 days later. Until then, the workspace can be restored on request to support@connie.ai.

Closing your account. To close your personal Connie account, contact us at support@connie.ai.

We may process and store information in countries other than the one in which you reside. Where required, we use appropriate safeguards for such transfers. Slack content processed by our integration connection provider, OpenRouter, or an AI model provider may be processed in the regions those providers operate in. Where you connect the MCP server to an AI assistant, data returned into your conversation may likewise be processed in the regions that assistant's provider operates in.

The Service is not directed to children under the age of 13 (or the equivalent minimum age in your jurisdiction), and we do not knowingly collect personal information from them.

We may update this Privacy Policy from time to time. When we make material changes, we will update the "Last updated" date above and, where appropriate, provide additional notice.

If you have questions about this Privacy Policy or our data practices, contact us at:

Email: support@connie.ai
Conigma, Inc., 1111B S Governors Ave # 43331, Dover, DE 19904, USA