Section 6

Intelligence Layers

The core architectural differentiator: meeting intelligence is structured around GH Systems' organizational hierarchy rather than treated as a flat searchable archive.

Layer Hierarchy

Each layer broadens the scope of analysis — from a single meeting, to patterns across streams, to organization-wide signals.

L1Meeting Stream Consolidation
L2Cross-Stream Aggregation
L3Organizational Pattern Recognition
increasing scope →

Layer Details

L1
Meeting Stream Consolidation
Per-meeting
Audience

Direct participants in each meeting stream

Trigger

Automated, post-meeting

Output

Per-meeting summary, action items, decisions, open questions

This is the baseline that COTS tools provide. The custom implementation allows control over summarization prompts, output format, and routing.

L2
Cross-Stream Aggregation
Cross-stream
Audience

Managers and directors with cross-functional visibility needs

Trigger

Scheduled (daily/weekly) or on-demand

Output

Cross-stream synthesis highlighting where topics, clients, or risks surface across multiple meeting types

Example

A client name mentioned in a solution architecture review, a PMO delivery standup, and a sales interlock within the same week gets flagged as a converging signal.

Implementation

Scheduled agent job that queries the vector DB across meeting streams for overlapping entities/topics within a configurable time window, then uses an LLM to synthesize findings.

L3
Organizational Pattern Recognition
Organization-wide
Audience

Executive leadership

Trigger

Scheduled (weekly/monthly) or on-demand

Output

Strategic-level patterns, trends, and anomalies across the entire meeting corpus

Example

Identifying that infrastructure scaling concerns are appearing across three different client engagements simultaneously, suggesting a systemic capability gap.

Implementation

Agent with broader query scope and longer time horizons, using the reasoning model to identify patterns invisible at individual meeting or stream level.