Internal at Neomanex — commercialisation deferred
AI that
remembers.
NeoRouter stores the experiences, paths, and context your AI sessions generate — then serves them back by semantic query in milliseconds. The next session starts with everything the last one learned.
At a glance
Institutional Memory for AI Operations
- Status
- Internal — commercialisation gated on Gnosari $5K MRR
- Built for
- Platform engineers and AI-ops leads running many-session agent workloads
Why it matters
Positioning pillars
Context cards, not logs
Retrievable by meaning.
Stored experiences are indexed as context cards: paths, notes, decisions, and gotchas tied to semantic queries. Unlike logs, they are retrievable by meaning — not timestamp. Unlike vector DBs, they carry operational context (which file, which decision, which fix), not just embeddings.
Instant retrieval, not exploration
Cold-start tax, eliminated.
A new AI session queries in natural language. NeoRouter returns relevant files, paths, and notes in milliseconds — replacing the exploratory "read 40 files to understand the codebase" tax that every cold session otherwise pays. The AI starts productive, not lost.
Compounding knowledge
The system gets smarter as it is used.
Every exploration that creates a route makes the next session faster. Knowledge does not die when a session ends — it accumulates. The organisation learns, the agents benefit, and the cold-start problem shrinks with every query. Institutional knowledge compounds instead of evaporating.
The mechanics
How it works
Step 1
Query by natural language
A new session asks "where does auth live?" or "how did we fix the stale cache issue?". NeoRouter returns the paths, notes, and decisions captured by previous sessions — no re-reading the codebase.
Step 2
Create routes on-demand
When a session discovers useful context, it stores a route. No build step, no nightly job, no staleness window — the index grows organically as work happens.
Step 3
Serve via MCP
NeoRouter exposes an MCP server as its primary interface. Any MCP-compatible client — Claude Desktop, Claude Code, Gnosari agents, ConvOps workflows — can query it directly. One memory layer, every tool.
Neomanex internal AI operations
NeoRouter is the institutional memory every AI session at Neomanex queries at startup. Every file read, every decision, every fix gets captured as a route — and every future session benefits. Internal since late 2025; load-bearing for the entire AI-operations stack.
The ecosystem
Fits into the portfolio
Related products
Uses under the hood
Questions, answered