Meetric / Technology

The technology behind
the conversation layer.

From audio capture to structured context. Explore the engines, access controls and interfaces that connect conversations to your company’s systems and AI.

Explore the technology

CaptureVoice · meetings · calendars

StructureTranscription · classification

GovernPermissions · retention

ConnectAPI · MCP · integrations

01 / API & MCP

Structured context. Scoped access.

Connect your applications through APIs and your AI assistants through MCP. Retrieve conversation records, topics and supporting detail within the signed-in user’s access.

Authentication: OAuth authorization with PKCE. Requested scopes define the data an external client can read.
Retrieval: Start with search or a topic. Retrieve the relevant conversations, then request the transcript detail you need.
Access: Account and partner enablement, user entitlements and conversation permissions apply to MCP reads.

API & MCPREAD FLOW

Your application or AI assistantA request for conversation context

Authorize the userOAuth · PKCE · scopes

SearchFind the context

TopicsNarrow the evidence

ConversationsRead the source

Structured records, source references and permitted transcript detail

02 / AI chat & search

Retrieval before generation.

A question becomes a search across conversation content. Relevant passages are retrieved with their conversation references, filtered by access, and used as context for the answer.

Semantic retrieval: Search transcript passages, insights and entities by meaning, with filters to narrow the conversation set.
Source grounding: Retrieved passages carry conversation IDs, speaker information and timestamps so the evidence can be inspected.
Permission filtering: Account scope, conversation visibility and applicable license restrictions shape the available source set.

Retrieval pipelineSEARCH EXAMPLE

Semantic queryNorthstar renewal conditions

Account + visibility filters

9 Oct · 14:22 · Anna

“We can renew if implementation starts by Friday.”

Answer context includes the passage and its source reference.

03 / AI categorization

Classify the conversation. Select the workflow.

Conversation types connect incoming content to the right processing. Type rules, conversation context and explicit calendar overrides help determine which configured summaries and actions should run.

Your taxonomy: Define conversation types with descriptions, allowed channels, positive rules and exclusions.
Context and overrides: Use the conversation’s content and department context. A saved calendar-event type can override automatic classification.
Workflow resolution: The selected type resolves to configured action nodes, including summary templates and structured extraction.

Call with Northstar
Meeting with Helios
Call from Vantage
Interview, Maria L.
Email from Meridian
Demo with Orbital
Site visit, Malmö

How it knows

Who was in it
The customer in your CRM
The meeting title
What was said, read by AI

The calendar invitation
Who was invited and who joined
Internal or external
The deal it belongs to
The team of the person who took it
Earlier conversations with the same people
The channel it came in on
The language spoken
Your own rules for each type
How long it ran
The email thread it belongs to
The time of day

Sales
Customer call
Summary
Next step, Deal risk, Decision
Then
Deal updated in your CRM
Who sees it
Sales team
Support
Support call
HR
Interview
Product
Customer feedback

Board meeting · Onboarding call · Supplier call · + your own

04 / Governance engine

Control who can read it. And how long it stays.

Access and retention are separate controls. Apply conversation visibility to people and teams, then set distinct retention periods for recordings, transcripts and conversation data.

Visibility: Account, department, team and user permissions determine the conversation content available to each person.
Separate lifetimes: Expiring a recording or transcript can leave the conversation summary available under its own data-retention period.
Lifecycle enforcement: Retention evaluation removes access to expired content. Permanent deletion follows the configured grace period.

Retention policiesPOLICY EXAMPLE

Time since the conversation

Recording90 days

Access removed after evaluation

Transcript365 days

Available within retention

Conversation data730 days

Available within retention

Conversation data includes summaries. Each policy is evaluated independently.

05 / AI transcription

Audio becomes a timed, searchable transcript.

Speech recognition converts the recording into text. Language detection and speaker diarization add structure, while utterance timestamps retain the connection to the original audio.

Speech processing: Transcription jobs request automatic language detection and diarization to separate speakers.
Utterance structure: Text, speaker labels and start/end timestamps become the building blocks for source playback and retrieval.
Asynchronous processing: Queued jobs and completion events connect recording availability to transcription and downstream analysis.

Timed transcriptAUDIO → TEXT

Speaker 1 Speaker 2

00:1200:1600:2100:24

Start00:16.1

End00:20.7

SpeakerSpeaker 2

Utterance text and timing remain linked to the conversation record.

06 / AI notetakers

Meeting capture, connected to the processing pipeline.

A notetaker connects a scheduled meeting to its recording and transcript. Capture events move the conversation through processing without requiring a manual recording upload.

Scheduling: Calendar events or meeting links provide the meeting URL. Joining follows the configured capture rules.
Event-driven ingestion: Recording-ready and transcription-completed webhooks attach the available assets to the conversation.
Downstream processing: Once the transcript is available, configured classification, summaries and extraction can run.

Capture lifecycle

01Scheduled
02Capture
03Recording
04Transcript

Meeting queuedMeeting URL + conversation owner

Event sequence

bot.recording_readyPending

transcription.completedPending

07 / Calendar integrations

Calendar events become capture instructions.

Connect Google or Microsoft calendars through OAuth. Synced events supply meeting links and scheduling context; your join settings determine which meetings the notetaker should attend.

Event context: Meeting URL, time, organizer and participants connect the calendar event to its conversation.
Join modes: Choose automatic joining or ask first. Configure internal and external meetings separately.
Rule filters: Use organizer restrictions, business hours, titles, domains and participants to refine meeting eligibility.

Calendar integrationJOIN-RULE PREVIEW

Google CalendarMicrosoft Calendar OAuth

Join external meetingsAutomatic · meeting link requiredAlso join internal meetings

09:00

Northstar renewalExternal · video meeting

Join

11:30

Product team syncInternal · video meeting

Skip

14:00

Helios onboardingExternal · video meeting

Join

2 meetings match this join rule.

08 / Phone capture · SIPREC

A recording stream alongside the live call.

SIPREC connects a recording-capable telephony platform to a recording service. The call continues between its participants while a separate recording session carries media and metadata for capture.

Telephony connection: The provider’s PBX or session border controller supplies the recording stream. Setup depends on the provider and its SIPREC configuration.
Protocol separation: SIP establishes the recording session; media streams and session metadata describe the conversation being recorded.
Processing after capture: Once the recording is available, transcription and configured analysis turn the call into a searchable conversation record.

SIPREC architectureTWO SESSIONS

Communication session · live call

Customer

PBX / SBCTelephony provider

Employee

SIPRECMedia + metadata

Recording serviceSeparate recording session

After the recording is available

Transcript + structured contextClassification · summaries · extraction

SIPREC protocol · IETF RFC 7866

09 / Integrations

Context in.
Structured updates out.

Connect capture sources, CRM context and output destinations around the conversation record. Each integration defines the connection, supported actions and data it exchanges with Meetric.

Context enrichment: Match conversations to CRM contacts, accounts and deals so processing has the relevant business context.
Configured outputs: Send summaries and structured fields through configured CRM actions, integration workflows, APIs and webhooks.
Connector capabilities: Authentication, supported operations and field mappings depend on the connected system. Configure the actions each workflow should execute.