Skip to main content

​
The honest answer to each “but we already have…”

These tools manage work assignment and status tracking. They’re good at answering “who’s doing what?” They’re not designed to answer “what did we decide and why?” or “what does an agent need to know before touching this part of the system?” Decisions live in ticket comments. Knowledge lives in a separate wiki. There’s no enforced link between them.Memex AI doesn’t necessarily replace your project tracker. It’s the layer underneath it, the decision and knowledge substrate that gives every ticket its context.
Wikis are where knowledge goes to die. They’re write-once, read-never, maintained-by-nobody. They have no concept of staleness, no drift detection, no link to the decisions that produced them, and no awareness of whether the code still matches what they describe.Memex AI blueprints are not pages. They’re living documents with provenance, scope, and automated freshness guarantees. See Drift Detection.
These are the closest precursors to what Memex AI provides, and they validate the need. But they’re per-repo, per-tool, and manually maintained. They don’t connect to decisions. They don’t track dependencies. They don’t detect drift. They’re a prototype of the blueprints layer without the decision layer or the coordination layer.Memex AI is the managed, multi-agent, cross-repository evolution of these files.
Many teams try this. Strategy docs in /docs, ADRs in /decisions, conventions in CONTRIBUTING.md. It works for small teams with one repo. It breaks when:
  • You have multiple repositories that share architectural decisions
  • Multiple agents need to coordinate across repos
  • Documents drift from reality and nobody notices
  • A decision in one repo affects work in another

​
What changes

Before Memex AI

  • Decisions scattered across Slack, tickets, meeting notes, and memory
  • AI agents pick up tickets, read stale wiki pages, and build the wrong thing
  • Two agents work on related features with no awareness of each other
  • New team members spend two weeks absorbing tribal knowledge
  • Re-prioritisation is chaotic, nobody can trace the decision graph
  • Documentation is always wrong, everyone knows it, nobody fixes it

After Memex AI

  • Every initiative has a Strategy that articulates the objective
  • Every decision has an ID, status, rationale, and explicit impact links
  • Agents read the strategy and blueprints before writing code, and stop if something is unresolved
  • Agents coordinate through a shared context layer, not through hope
  • New team members (human or AI) load the strategy and contribute immediately
  • Re-prioritisation is a graph operation, reverse D4 and see the cascade
  • Institutional knowledge stays current because the system detects drift

Join the waitlist

Memex AI is in early access. Come build with us, request access at memex.ai.