Memory as infrastructure for AI
AI that remembers — and can prove it.
CrawlQ is your AI research and content workspace. GraQle is the memory layer underneath it. Together, every insight your team builds is stored, traceable, and replayable — not just saved, but verifiably saved.
The problem
AI tools forget. That is expensive.
Context decays between prompts, people, and model versions — and when it does, no one can reconstruct why an output was produced or whether it was on-brand.
Prompt amnesia
Essential context gets retyped from scratch every session, thinner each time.
Document fragmentation
Knowledge exists, but the relationships and authority behind it are unclear.
Decision opacity
Teams cannot explain which evidence or policy shaped an output.
Model drift
Swap the model and results change — with no stable organisational anchor.
The solution
One substrate. One application on top.
GraQle keeps the memory; CrawlQ puts it to work. The intelligence is never retyped, and it stays portable across models.
GraQle — the substrate
A governed memory and reasoning layer.
Holds concepts, evidence, versions, policies, decisions, and outcomes as a live graph — so context is portable across models and never has to be retyped. Open-core; a developer can drive it directly over MCP.
CrawlQ — the application
The workspace your team uses every day.
Audience research, brand memory, strategy, and content — the first mature domain application on the substrate. What your team builds is replayable: trace any output back to the research and the approver that produced it.
The proof
Not a claim — a public record.
The moat isn't a bigger model. It's that the memory is verifiable by anyone, without taking our word for it.
When a human approves AI content in CrawlQ, the root of that record is written to a public transparency log CrawlQ cannot edit or delete from — the same kind of log that secures the world’s open-source software. You export the trail from your own CrawlQ workspace and hand it to your auditor; they then confirm it against the public record — and that verification step needs no access to CrawlQ at all. Only a fingerprint is ever public — never your content, your reviewers’ names, or your workspace.
Where to go next
Two ways in.
Use the workspace, or build directly on the substrate.
Use CrawlQ
The full workspace — Brand Memory, Research, Governance, Canvas — with the audit trail built in. For teams that want the product.
Start with CrawlQIntegrate GraQle
The open-core substrate on its own — an MCP-compatible memory and reasoning layer for your own stack. For developers.
See the engineering pageCommon questions
The questions people ask first.
- Is GraQle the same thing as CrawlQ?
- No. GraQle is the memory and reasoning layer — the substrate. CrawlQ is the application your team uses on top of it. You can use CrawlQ without ever touching GraQle directly; developers can also use GraQle on its own.
- What does “replayable” actually mean?
- Every output can be traced back to what shaped it — the source entities, the brand context, the approving human. It is an evidence chain, not a time-travel feature.
- How is the audit trail verifiable?
- When a human approves AI content, the root of that record is written to a public transparency log CrawlQ cannot edit or delete from. An auditor confirms your exported trail against that public log without needing any access to CrawlQ. Only a fingerprint is ever public — never your content or your people.
- Can I use GraQle on its own?
- Yes. GraQle is an Apache-2.0 open-core SDK that exposes its tools over MCP, so a developer can drive it directly from any compatible IDE or agent. See the engineering page for the install steps.