Resources
Everything we have measured, written down.
Reports, docs, comparisons and engineering notes. Nothing here is behind a form or an email wall.
- Sample sizes shown
- Limits named
- No form, no email wall
All resources
Find the page you need.
Product pages, reports, docs, comparisons and engineering notes. Filter by kind.
- New
Give your AI the task, not your company
Every AI tool asks for everything. The math, what goes wrong, a fix in three verbs, and what to do this week with no product.
Read the essay - Early access
Know every identity that can act
One inventory of the people, AI agents, API keys and connections that can act, each scored for exposure.
See Know - Early access
Scope each AI to its job
Limit an agent, or an outside tool’s key, to the sources its job needs, and preview what that would have cut.
See Scope - Live
Check an action before it happens
Send a proposed action to the API and get go, hold or ask a person. A public sandbox is open.
See Check - Solutions
Solutions for the AI you already use
Code review, support agents and finance approvals: what each one needs to see, and where a person signs off.
See solutions - Report
Agent authorization proof
A real LangGraph agent, run with and without qbrin: four unsafe merges without it, zero with it, and the valid one still ran.
Read the proof - Report
Benchmark scorecard
Head-to-head results with the sample size and the caveat on every row.
Open the scorecard - Report
AI in the SOC: security benchmark
A measured SOC comparison: qbrin against RAG, an MCP agent and a naive LLM analyst, on live security data.
Read the benchmark - Docs
API reference
The REST API: authenticate, ingest, ask, and get a cited answer or a clear abstain.
Open the docs - Docs
Action authorization API
The pre-execution endpoint that answers ALLOW, HOLD or ESCALATE for a proposed agent action.
See the endpoint - Comparison
qbrin vs Glean
An architecture-level comparison, including where Glean is stronger.
Read the comparison - Comparison
Glean alternatives
qbrin, Microsoft 365 Copilot, Coveo, Elastic and Guru: pricing models and who each fits.
See the options - Note
How qbrin works: evidence before answers
A visual guide from connected sources to a cited answer or a deliberate abstention.
Read the note - Note
Evidence vs verification vs authorization in AI
Why source retrieval, claim support and permission to act are three separate gates.
Read the note - Note
qbrin benchmark results: an honest guide
The published numbers, the sample sizes, and the cases where qbrin does not lead.
Read the note - Note
When an absence becomes a finding
The quietest failure in AI systems: a failed read that gets reported as a confident negative.
Read the note - Note
The most dangerous prompt injection is stopped by retrieval, not detection
Credential-scoped boundaries at the retrieval plane.
Read the note - Note
A safety gate that blocks 100% of attacks and 100% of real work
Why substring grounding checks look perfect on the scoreboard and reject valid commands.
Read the note - Note
A 400-token cap on our verifier was silently dropping correct answers
How a fixed limit turned parser errors into silent refusals.
Read the note - Note
AI abstention: why knowing when to stop matters
The business case and measurement discipline behind saying “not enough evidence”.
Read the note
Ask your company anything.
See qbrin answer from your own documents, with sources, in a live demo.