Skip to main content

Core Concepts

Understanding the key building blocks of the Lira platform.

Organizations​

An Organization is the top-level entity in Lira. Everything — meetings, integrations, knowledge base, team members, usage quotas — is scoped to an organization.

  • Users create or join organizations after signing in
  • An organization has a profile (name, industry, description) that Lira uses as context during meetings
  • Each org can have its own custom email domain for customer support
  • Usage and billing are tracked per organization

Meetings​

A Meeting represents a single session where Lira was deployed. Each meeting stores:

  • Transcript — Speaker-attributed messages with timestamps
  • Summary — AI-generated short and detailed summaries
  • Tasks — Extracted action items with assignees and status
  • Recording status — Audio capture metadata
  • Participants — Who was in the meeting

Bot​

The Bot is the headless Chromium browser instance running on the backend that joins Google Meet as "Lira AI." Each bot deployment:

  1. Launches a Playwright browser
  2. Navigates to the Google Meet URL
  3. Joins as a participant
  4. Captures audio bidirectionally
  5. Streams to Amazon Nova Sonic for real-time AI

Only one bot can be active per meeting at a time.

Knowledge Base​

The Knowledge Base is your organization's context that Lira uses to give informed responses:

  • Website crawl — Lira can crawl your company website and index the content
  • Document uploads — Upload PDFs, DOCX files, and other documents
  • Vector search — Content is embedded via OpenAI and stored in Qdrant for semantic retrieval
  • RAG pipeline — During meetings, relevant knowledge base content is injected into Lira's context

Tasks​

Tasks are action items extracted from meetings by AI. Each task has:

  • Title and description
  • Assignee (mapped to org members)
  • Priority and status
  • lira_review_status — Lira autonomously validates tasks before execution
  • Integration sync — Tasks can be pushed to Linear, GitHub Issues, etc.

Integrations​

Integrations connect Lira to your existing tools. Each integration goes through:

  1. OAuth flow — User grants access
  2. Member mapping — Lira maps org members to external accounts
  3. Bidirectional sync — Data flows both ways (e.g., tasks to Linear, Slack messages to Lira)

Personality Modes​

Lira has 4 configurable personality modes that change how she participates in meetings:

ModeBehavior
SupportiveEncouraging, affirming, builds on ideas
ChallengerDevil's advocate, questions assumptions
FacilitatorKeeps discussion on track, ensures everyone is heard
AnalystData-driven, focuses on metrics and evidence