Query & Agent Architecture
Users interact with the meeting knowledge base through natural language queries and autonomous agents. Queries retrieve semantically relevant context and synthesize answers with citations; agents operate as configurable jobs that run intelligence workflows on a schedule or on demand.
Query Flow
Every query follows the same retrieval-augmented generation (RAG) path. The embedding step uses the same model that was used at ingest time, ensuring vector space consistency between documents and queries.
Agent Framework
Agents are autonomous or semi-autonomous jobs that operate on the meeting knowledge base. Each agent targets a specific intelligence layer and is triggered by a schedule or event rather than a user query.
Generates post-meeting summaries with action items, decisions, and open questions.
Detects topic and entity overlap across streams within configurable time windows.
Identifies emerging patterns and recurring signals over longer time horizons.
Proactively notifies stakeholders when configurable trigger conditions are met.
Produces periodic briefing documents and executive summaries on demand.
Each agent is a configurable job definition composed of four elements. Definitions are stored in the relational DB and editable through the admin interface. The Job Runner handles all scheduling and execution.
Schedule (cron), event (meeting ended), or manual invocation via admin UI.
Which meeting streams, time windows, and metadata filters the agent operates against.
The instruction set passed to the reasoning model, parameterized by scope results.
Where results are written — notification channel, report store, or webhook.
Reasoning Model Strategy
The reasoning model is configurable per agent and per query type. This allows cost optimization: simpler L1 summaries can use a faster, cheaper model while complex L3 synthesis and executive reports reserve the highest-capability model. The model selection strategy is a first-class configuration concern, not a hard-coded choice.
| Model | Strengths | Use Case |
|---|---|---|
| Claude Sonnet 4 | Strong reasoning, good cost/performance balance | Default query and agent model |
| Claude Opus 4 | Highest reasoning capability | Complex L3 synthesis, executive reports |
| GPT-4o | Strong general purpose | Alternative / fallback |
| Local models (Llama, Mistral) | No API costs, full data privacy | Future option if API cost or data sensitivity becomes a concern |