HEXi

Agent Memory Connector

Connect your existing AI workflow — Claude Code, OpenAI Codex, Cursor, or any MCP-compatible agent — directly to HEXi's institutional memory.

• Search facts — query the knowledge graph from inside any agent session.
• Queue episodes — propose new knowledge for human review before it enters the graph.
• Retrieve nodes — look up entities, decisions, and actors by semantic similarity.

pHEX exposes a token-gated MCP endpoint. Add it to your claude_desktop_config.json or any MCP client and your agent gains long-term memory backed by HEXi's governed knowledge graph.

Your Bespoke Blueprint

Talk to HEXi directly. Describe your organisation, your team, your projects and your goals — and HEXi will build a bespoke configuration blueprint tailored to you.

• Profile — HEXi learns your domain, team structure, and work patterns.
• Blueprint — a ready-to-deploy ontology and agent configuration for your instance.
• Memory — every conversation persists, so HEXi always picks up where you left off.

Time · Notes · Status

Log work sessions directly against sprint items. Track time, capture notes, and move each session through a clear status lifecycle — all without leaving the board.

• Sessions — multiple work entries per item, each with date, time, and duration.
• Status — Started → Complete → Billed → Cancelled lifecycle per entry.
• Agents — project agents can log, query, and update work via tools.

Knowledge Archivist

The Librarian is HEXi's dedicated Knowledge Archivist — validating every episode before it enters the graph, enforcing schema standards, and keeping institutional memory honest.

• Stewardship — reviews queued episodes and enforces entity and relationship schemas.
• Integrity — detects knowledge gaps, sync errors, and orphaned graph nodes.
• OOK — flags schema violations and maintains a living audit log.

Governance Pipeline

The nerve centre of HEXi. The Orchestrator coordinates every governance decision — assembling the council, collecting votes, monitoring bias, and finalising consensus without ever casting a vote itself.

• Neutral — structurally prevented from voting in any decision it runs.
• Full council access — delegates to all 21 agents in parallel for domain reasoning.
• Memory — every decision is recorded with full rationale and vote breakdown.

Your Project's Agent

Every project in HEXi gets a dedicated conversational agent. Pre-loaded with milestones, sprints, team, and history — it knows your project before you say a word.

• Always oriented — full project snapshot injected on every conversation start.
• Action-capable — creates sprints, updates items, logs work, and manages triggers.
• Governed — raises project decisions through the full council pipeline.

Why HEXi wins

Single-agent AI gives you one perspective. Human committees give you multiple perspectives slowly. HEXi gives you adversarial, accountable, self-improving governance — at machine speed.

• Adversarial by design — 21 agents with opposing biases stress-test every decision.
• Track-record weighted — agents earn credibility through prediction accuracy, not configuration.
• Structural neutrality — the Orchestrator coordinates but never votes.

What you gain

HEXi is not a productivity tool you adopt. It is a governance layer that changes how your organisation makes decisions, retains knowledge, and improves over time.

• Faster decisions — 21 agents deliberate in parallel; sessions that take days complete in minutes.
• Documentation that writes itself — every session produces a structured record as a byproduct of normal operation.
• Bespoke from day one — chat with HEXi to build your organisation's blueprint before you sign up.