What AskMyDocs is
AskMyDocs is a self-hostable AI hub for enterprise knowledge. It fuses hybrid retrieval-augmented generation (pgvector + full-text search + a reranker), a typed canonical knowledge graph with human-gated promotion, a streaming chat surface built on the Vercel AI SDK, and a full admin operations cockpit into a single Laravel platform. It is the open-source, on-prem alternative to Glean / Notion AI / ChatGPT Enterprise — without the per-seat lock-in or the six-figure on-prem contract.Most “RAG over docs” tools treat your knowledge base as a pile of
interchangeable chunks: they re-discover the answer from zero on every query,
never persist what your team has already decided, and re-propose options that
were explicitly dismissed three quarters ago. AskMyDocs is built around the
opposite premise — that an enterprise KB is institutional memory, not a
vector index.
How it fits together
The six moats
These are the differentiators no other public RAG platform — open-source or SaaS — currently ships. Each links to the page that explains the theory, the design, and the decision rationale in depth.Human-gated canonical promotion
A three-stage pipeline holds the LLM at “draft”; only humans and operators
commit canonical storage. Immutable editorial audit trail.
Institutional memory + anti-repetition
A retrieval-time knowledge graph folds in neighbours, and a ⚠ firewall stops
the LLM re-proposing approaches your team already rejected.
Self-compiling Auto-Wiki
A second-class
auto tier the system builds and maintains itself —
behind an anti-hallucination firewall that always ranks human > auto > raw.Field-level PII redaction
GDPR-grade redaction wired at every persistence boundary, default-off and
granular per touch-point — not just data residency.
MIT, self-hostable, on-prem
Runs on any Laravel + PostgreSQL + pgvector host. Zero vendor lock-in; the
entire sister-package stack is MIT and independently reusable.
Eval-harness CI gate
A RAG regression gate on every PR plus nightly LLM-as-judge and adversarial
cohorts — an out-of-the-box eval surface nobody else publicly ships.
Who it is for
- Enterprise teams ingesting architectural decisions, runbooks, standards, incidents and domain concepts into a navigable KB.
- Regulated-industry operators (GDPR, EU AI Act) needing field-level PII redaction and an immutable audit trail at every persistence boundary.
- Engineering orgs that want LLMs to stop re-proposing rejected approaches.
- Anyone allergic to vendor lock-in who wants Glean-class capability on infrastructure they own.
Where to go next
Quickstart
Ask your first grounded question in five minutes.
Core concepts
The canonical layer, the graph, evidence tiers, and the firewall.
Architecture
The full system design, request lifecycle, and decision rationale.