How to Build an AI Meeting Notetaker Across Zoom, Google Meet, Outlook, and Salesforce
March 10, 2026
AI meeting notetakers have quickly become a core feature in many B2B SaaS products. Sales teams, customer success teams, and account managers want automatic summaries of their meetings, action items extracted from conversations, and follow-ups written back to CRM systems.
Products like meeting intelligence tools, sales assistants, and AI productivity platforms provide these capabilities by connecting to their customers' meeting platforms. Instead of users manually taking notes, the application joins meetings, retrieves transcripts, generates summaries, and updates CRM records automatically.
The challenge is that meeting data lives across multiple systems. Meetings are scheduled in Google Calendar or Outlook, hosted on platforms like Zoom or Google Meet, and the resulting insights often need to be attached to CRM systems like Salesforce. Each system exposes different APIs, schemas, and webhook models.
Unified simplifies this by providing normalized APIs across calendar, meeting recording, messaging, and CRM platforms. Developers can retrieve meeting metadata, transcripts, and recordings through a consistent API and use that data to power AI summarization workflows.
Why SaaS Products Build AI Meeting Notetakers
Many SaaS platforms build AI meeting assistants for their customers.
Examples include:
- Meeting intelligence tools
- Sales meeting assistants
- AI note-taking applications
- Customer success platforms
- Revenue intelligence tools
These products help teams automate several tasks that normally require manual effort.
Common capabilities include:
- Automatic meeting summaries generated from transcripts
- Action item extraction from meeting conversations
- Follow-up reminders created automatically
- CRM updates based on meeting outcomes
- Searchable meeting knowledge bases
By connecting to customer calendars and meeting platforms, SaaS products can automate note-taking across all meetings without requiring users to manually record notes.
Unified Categories Used for AI Meeting Notetakers
An AI meeting assistant typically connects to several Unified API categories.
| Category | Purpose | Key Objects |
|---|---|---|
| Calendar & Meetings | Meeting schedules and recordings from providers like Zoom and Google Meet | calendar_event, calendar_recording |
| CRM | Customer and deal context stored in platforms like Salesforce | contact, company, deal |
| GenAI | Prompt generation and model responses | genai_prompt, genai_model |
Using normalized APIs across these categories allows developers to build AI assistants without implementing separate integrations for each meeting provider.
Unified Calendar and Meeting Objects
Unified normalizes meeting data across multiple providers through two primary objects.
Calendar Event
The calendar_event object represents a scheduled meeting.
Key fields include:
| Field | Description | Usage |
|---|---|---|
| id | Unique identifier for the meeting | Internal reference |
| title / subject | Meeting title | Used in meeting summaries |
| start_at | Meeting start time | Used to schedule the notetaker |
| end_at | Meeting end time | Used to stop recording |
| attendees | List of participants | Used to identify speakers |
| conference | Meeting provider and join URL | Used for meeting access |
| organizer | Meeting creator | Used for CRM ownership context |
| location | Meeting location or link | Display context |
| web_url | Provider link to the event | Reference to source meeting |
Events can be retrieved using:
GET /calendar/{connection_id}/event
GET /calendar/{connection_id}/event/{id}
These endpoints allow applications to identify upcoming meetings and retrieve meeting metadata.
Meeting Recording
The calendar_recording object represents a meeting recording and transcript.
Key fields include:
| Field | Description | Usage |
|---|---|---|
| id | Recording identifier | Internal reference |
| event_id | Associated meeting event | Links recording to event |
| start_at / end_at | Recording timestamps | Used for timeline analysis |
| provider | Recording provider (Zoom, Teams, etc.) | Source platform |
| duration_seconds | Recording duration | Meeting length analytics |
| media | Media segments including transcripts | Transcript extraction |
| transcript_download_url | URL for full transcript | Used for summarization |
| recording_download_url | URL for audio/video | Optional processing |
| language | Transcript language | Model selection |
| web_url | Recording view link | Reference to original recording |
Recordings can be retrieved using:
GET /calendar/{connection_id}/recording
GET /calendar/{connection_id}/recording/{id}
These objects provide everything needed to generate AI meeting notes.
Retrieving Meeting Transcripts
Meeting transcripts are stored within the media array of a recording object.
Each media segment may include transcript data.
Important transcript fields include:
| Field | Description |
|---|---|
| start_at | Timestamp when the speech segment begins |
| end_at | Timestamp when the segment ends |
| text | Transcript text |
| attendee | Speaker metadata including name and email |
| language | Language of the segment |
These transcript segments can be assembled into a full conversation transcript.
Applications can retrieve transcripts in two ways:
- iterate through transcript segments returned in the API response
- download the full transcript using
transcript_download_url
Once the transcript text is assembled, it can be passed to an AI model to generate summaries.
Connecting Customer Integrations
Customers authorize their meeting platforms and CRM systems through Unified Connect.
Typical process:
- Your application launches the Unified Connect authorization flow.
- The user selects integrations such as Zoom, Google Meet, Outlook, or Salesforce.
- The user authorizes access to their accounts.
- Unified redirects back to your application with a connection_id.
Your application stores the mapping:
user_id → connection_id
All subsequent API requests use this identifier to retrieve meeting data.
Generating AI Meeting Notes
Once transcripts are retrieved, the application can generate summaries using Unified's GenAI API.
A prompt is created using the genai_prompt object.
Example prompt request:
{
"messages": [
{ "role": "system", "content": "You are an AI meeting assistant." },
{ "role": "user", "content": "Summarize this meeting and extract action items." }
]
}
The request can include parameters such as:
model_idtemperaturemax_tokens
The response includes:
- generated summary text
- token usage
- model response metadata
This allows applications to generate:
- meeting summaries
- key discussion points
- follow-up tasks
- sentiment insights
Writing Meeting Insights Back to CRM
Many AI meeting assistants automatically update CRM records.
For example, after summarizing a meeting, the application can:
- attach the summary to a deal record
- create a note activity
- update deal stage or next steps
- assign follow-up tasks
Using CRM APIs, the meeting insights become part of the customer's activity history.
This ensures sales teams always have an accurate record of customer conversations.
Real-Time Meeting Automation with Webhooks
Unified provides webhook events that allow meeting assistants to react to calendar changes in real time.
Important events include:
| Event | Description |
|---|---|
| event.created | New meeting scheduled |
| event.updated | Meeting details changed |
| event.deleted | Meeting cancelled |
| event.started | Meeting begins |
| event.ended | Meeting finishes |
| recording.created | Recording becomes available |
| recording.completed | Transcript processing finished |
These events allow AI assistants to:
- join meetings automatically
- start recording at the correct time
- retrieve transcripts once recordings are processed
- trigger summarization workflows immediately
Supported Platforms
Unified supports a broad range of meeting and CRM integrations.
Calendar & Meetings (27 integrations)
Examples include:
- Zoom
- Google Meet
- Microsoft Outlook
- Microsoft Teams
- Webex
- Calendly
- Apple iCloud
CRM (47 integrations)
Examples include:
- Salesforce
- HubSpot
- Pipedrive
- Zoho CRM
- Microsoft Dynamics
This allows AI meeting assistants to support many customer environments without building separate integrations.
Why Product Teams Choose Unified
Building an AI meeting assistant typically requires multiple integrations across meeting platforms and CRM systems.
Unified simplifies this by providing:
- normalized calendar and meeting recording objects
- unified APIs across 27 meeting integrations
- CRM integrations across 47 platforms
- webhook events for real-time meeting workflows
- standardized transcript retrieval across providers
Instead of maintaining dozens of integrations, product teams can focus on building AI features that automatically summarize meetings, capture action items, and update CRM systems.
Start building AI meeting notetakers across Zoom, Google Meet, Outlook, Salesforce, and many other platforms today.