What does Day AI integrate with? The connections behind your GTM context graph
Updated
The short answer. Day AI integrates with the places your customer conversations already happen: Gmail and Google Calendar, Day AI's notetaker on Zoom, Google Meet and Microsoft Teams, Slack, Gong, Granola and Fireflies. Every one of them feeds a single GTM context graph, the foundation of AI for GTM: one record of every person, company, deal and conversation, linked to each other. Day AI's agents read and act on that graph, and reach Salesforce, HubSpot and your other tools through MCP connectors, whether you run Day AI in place of your CRM or on top of Salesforce or HubSpot. Claude, Cursor or ChatGPT can work from the same graph through Day AI's own MCP server.
Say you run sales at a B2B marketplace with about five people, and you've just moved the team off Salesforce onto Day AI. It's 3:40 on a Wednesday and a demo has just ended. The prospect's RFP has been sitting in your inbox since Monday. A month ago, your mornings started with a pipeline export you downloaded to feed your own workflow. Now Gong is connected to Day AI, and thousands of past calls sit in the graph, each linked to its contacts, company and deal. You open Claude, where the Day AI MCP is connected, and ask it to work on the RFP and draft the follow-up from everything this buyer has said across every call. Both drafts come back citing the calls they drew on. You fix one line in the follow-up, send it, and ask Claude to set the deal's next step to the RFP review. The deal in Day AI now carries that next step. Nobody downloaded a spreadsheet today.
Where the usual answer goes wrong. Most integration pages hand you a list of logos, as if every connection did the same job. They don't. Some connections are where the record comes from, and some are where agents do work, and if you connect the second kind before the first, your agents have nothing true to read. Day AI provides the context graph and agent control plane to quickly and securely upgrade your company brain. The graph comes first: feed it, then point agents and tools at it. You may already have Gong, a CRM and a Claude project wired together by hand. That was the right start; the connections below turn it into one record your whole team can use.
The connection order: feed, record, act, reach, with Day AI
Think of every integration as one of four jobs. Connect them in this order, and each layer has something real to work with.
- Feed: the conversations each person has. Email, calendar and recorded meetings are the bulk of what happened with every customer. In Day AI, each person connects their own Google account and turns on the notetaker, and the sharing rules they set decide what teammates and their agents see.
- Record: the conversations your team already keeps elsewhere. Slack channels about deals, your call recorder's history and your meeting notes. In Day AI, a workspace Owner connects Slack, Gong, Granola or Fireflies once, and they land on the same people, companies and deals as the email and meetings.
- Act: the systems agents read and write. Your CRM, project tools and internal apps. In Day AI, these are MCP connectors: Salesforce and HubSpot, tools like Linear and Notion, and any MCP server a workspace Owner registers, each tool set to Always allow, Needs approval or Off.
- Reach: the places your team already works. Claude, Cursor, ChatGPT and your own code. In Day AI, the MCP server and the TypeScript SDK let those tools read and write the same customer memory under each person's own access.
How to connect Day AI, step by step
Step 1: Connect Gmail and Google Calendar to Day AI, person by person
Each rep connects their own Google account; nobody can connect someone else's mail. Before anything is shared, each person sets their email sharing rules: share everything or nothing by default, with exceptions by domain, address or Gmail label. A new person's first email and calendar sync takes one to six hours, so ask everyone to connect on day one. From then on, customer email and meeting invites show up in Day AI on the people, companies and deals they're about, without anyone typing them.
Step 2: Turn on the Day AI notetaker for Zoom, Google Meet and Microsoft Teams
The notetaker joins your Zoom, Google Meet and Microsoft Teams calls, and each meeting gets a summary, key points, action items and attendees. Each person picks a recording default per calendar account: all meetings, external meetings only, internal meetings only, or none unless asked. Meeting sharing is Invite only or Workspace, set separately for internal and external one-on-ones and group meetings, and when attendees' settings conflict, the most private one wins. Then ask your Day AI agent to create a follow-up skill with the "Meeting summary" trigger, so a follow-up draft is waiting for you after each recorded call; this is the connection that gives agents something to do in week one.
Step 3: Add Slack, Gong, Granola, Fireflies or Zapier in Day AI's workspace settings
A workspace Owner connects these once for the whole workspace, under Workspace Settings > Integrations:
- Slack. Connect Slack, then add Day AI to the public channels where deals and accounts are discussed from the channel list, or invite it into a private channel from Slack. Day AI can also deliver skill results to you in Slack.
- Gong. Connect Gong and your call history and transcripts come into Day AI, shown in Meetings next to native recordings and linked to contacts and organizations. Day AI reads Gong and never changes anything in it.
- Granola. A workspace Owner adds a Granola API key, and shared notes, transcripts and attendees come in. Skills set to run when a meeting recording is ready run on Granola notes too.
- Fireflies. A workspace Owner adds a Fireflies API key, and your team-shared Fireflies meetings come in.
- Zapier. Push people and companies into Day AI from any app Zapier connects, such as a form or a billing tool, and let that data trigger skills.
Step 4: Run Day AI as your CRM, or connect Salesforce or HubSpot
If you'd rather not keep a separate CRM, run Day AI as your CRM, with pipeline, contacts and companies kept current from your team's conversations. If you keep your CRM, connect it as an MCP connector. A workspace Owner saves your company's OAuth app for Salesforce or HubSpot once, and each person then connects with their own login. Agents read your CRM from day one, and when you turn on the MCP write tools they write back to it too, updating stages, notes and fields after a call. Either way, the graph from Steps 1 to 3 is what the agents read.
Step 5: Add the connectors your agents act in
Under Connectors in your Day AI settings, each person connects the tools they want their agent to use in chat and in skills. Common picks in Day AI for a go-to-market team are Linear, Notion, Asana, Intercom and Docusign, and Day AI also offers Outlook 365 Mail, Outlook 365 Calendar, Microsoft Teams and Atlassian connectors in beta. Each connector's tools can be set to Always allow, Needs approval (Day AI asks in the conversation before each use) or Off. If a tool you rely on isn't listed, a workspace Owner can register it as a custom MCP connector with a name, an endpoint and its authentication, and it then works like the built-in ones.
Step 6: Connect Claude, Cursor or ChatGPT to Day AI over MCP
Day AI's MCP server is at https://day.ai/api/mcp, and day.ai/mcp walks through setup for Claude (web, desktop and mobile), Claude Code and Cursor; ChatGPT and other clients that support remote MCP servers with OAuth sign-in connect the same way. You authorize the connection by choosing one of your Day AI agents, and from then on, asking Claude about a deal pulls from the same customer memory your agents use, with citations back to the call or email. For your own apps, the Day AI SDK is an MIT-licensed TypeScript client for the same tools.
Where a hand-wired stack breaks
- Each connection is its own island. A Gong export in one folder, a CRM export in another and a Slack bot in a third means every question starts with stitching. In Day AI, the Gong call, the Slack thread and the email about one deal all show up on that deal, so one question gets one answer.
- Nobody knows whose email the bot can see. A shared script with one person's mailbox token works until the second rep joins. In Day AI, each person's own sharing rules decide what teammates and their agents see, and an agent works with the permissions of the person it works for.
- The CRM drifts from the calls. Copying call notes into fields by hand falls behind within a week. In Day AI, an agent proposes the field and stage changes from the conversation, and you decide which ones it makes on its own.
- Every new tool needs new glue code. Adding a project tracker or an internal app means another integration to maintain. In Day AI, a new tool is one MCP connector a workspace Owner registers, and each person connects it and sets its tools to Always allow, Needs approval or Off, like any built-in connector.