Onboarding Buddyqbrin assistant
›How do I get my dev environment set up?
Reading eng-handbook.pdf
Checking #new-hires
Run ./setup from the infra repo, access comes with your Okta group. Full guide linked below.eng-handbook.pdf
The universal trust layer for enterprise answers: never confidently wrong, always cited, at a fraction of the cost. Across any source, in any language.
qbrin connects conversations, files, emails, and decisions across your tools, so teams can ask in any language and get clear answers backed by the original source.
Release v2.4 ships March 14, signed off by Maya Chen in the launch review.
A question fans out to your live systems and knowledge, the evidence comes back, and qbrin verifies every claim against it. You get a decision you can act on — or an honest “not enough evidence,” never a confident guess.
Measured on a 1,000-question HotpotQA A/B against a tuned RAG baseline — the honest floor, not a cherry-pick. See the full benchmarks ↓
Release v2.4 ships March 14, signed off by Maya Chen in the launch review.
One home view shows what's been brought in, who's involved, and what's been decided, so leadership gets an instant pulse on the organization without chasing status updates. People, decisions, and commitments are pulled out of everyday conversation for you, automatically.
How much it reads for one answer
qbrin reads ~687 tokens per answer, a short briefing, not a pile of documents
vs a heavy production RAG pipeline (multi-query ×3 / 20-chunk + reranking) · ~8× vs even a lean 10-chunk RAG call
Ask a closed-book LLM about a company’s filings and it invents financial figures about a third of the time. qbrin reads the actual document and answers from it, so it’s right, or it tells you it isn’t sure.
Closed-book LLM vs qbrin’s retrieval-grounded answers on FinanceBench, real public-company 10-K filings, each graded against the source passage. Preliminary sample (n=16).
500 real company questions (470 answerable + 30 unanswerable), the most product-representative test we run. The flagship result is never confidently wrong plus citations you can trust, not a blanket accuracy crown.
Honest read: the real win here is risk / safety and citation-trust, not blanket accuracy. On this 94%-answerable workload a naive RAG that always answers scores ~83% on the answerable subset, qbrin’s edge is that it’s never confidently wrong and its citations hold.
One story holds across all of it: qbrin is the most trustworthy (lowest rate of confident wrong answers, highest citation-trust) and the cheapest per answer, and its retrieval is competitive-to-leading once tuned to a corpus, with one architectural exception: relational-graph traversal (a rival’s home turf), where it trails. On raw all-answerable public Q&A it runs a calibration posture (abstains rather than guess); the 2026 embedder upgrade lifted that retrieval substantially (FinanceBench gold-in-context +18pp same-harness). Every cell below is a real run.
| Dataset | Safetyfewer made-up answers | Citation-trustcited source actually holds | Current-factretires stale / contested | Retrievalfinds the right source | Answer accuracycorrect when it answers | Cost / tokenstokens read per answer | Multilingualparity across languages |
|---|---|---|---|---|---|---|---|
| CTP-Bench (temporal/contested)n=12 | Leads | ||||||
| Temporal vs mem0 (open-source)n=8 | Leads | ||||||
| Temporal vs Zep / Graphitin=6 | Leads | ||||||
| PrecisionMemBench (memory-precision)n=77 | Leads | Leads | |||||
| BrainBench (relational retrieval)n=145 | Trades for calibration | ||||||
| Multi-hop vs HippoRAG 2n=200 | Leads | ||||||
| MuSiQue (hard multi-hop)n=300 | Competitive | ||||||
| multihop-rag (retrieval)n=2255 | Leads | Leads | |||||
| multihop-rag (answer / unanswerable)n=120 / 301 | Leads | Trades for calibration | |||||
| ragbenchn=100 | Trades for calibration | Competitive | |||||
| financebenchn=150 | Leads | Trades for calibration | |||||
| Enterprise RAG Benchmark (ERB)n=500 | Leads | Leads | Leads | Trades for calibration | Leads | ||
| G3 contested enterprise Q&An=60 | Leads | Competitive | Leads | ||||
| Multilingual (en / hi / te / ta)n=188 | Leads | Competitive | Leads | ||||
| Knowledge-map compression (SOTA)n=80 | Competitive | Leads | |||||
| Hallucination / fake-entity setn=60 | Leads | ||||||
| Broad RAG general Q&An=120 | Competitive | Competitive | |||||
| Internal enterprise benchmarkn=40 | Competitive | Competitive | Competitive | ||||
| False-premise safety (real / fake)n=55 | Competitive |
Measured, not modeled, each row is a benchmark we ran on qbrin’s own pipeline (competitor figures are their own published numbers or faithful re-runs). Several wins are corpus-specific and several safety A/Bs are honest ties; the per-suite tabs below carry the exact figures and caveats. The honest headline: no competitor matches qbrin on trust, safety-under-abstention and cost, and its retrieval is competitive-to-leading once tuned to a corpus (except relational-graph traversal, where it trails), on raw answerable-Q&A accuracy it runs a calibration posture (trades coverage for caution), and the 2026 embedder upgrade lifted that retrieval substantially.
Numbers come from real benchmark runs on qbrin’s own pipeline, the ERB set of 500 real company questions (470 with a known correct source), plus trap questions built to bait a wrong answer. Each figure below is labelled MEASURED (we ran it), MODELED (computed from measured tokens × published prices), or PUBLISHED (a competitor’s own number). No per-suite second-by-second timings are shown unless we measured them.
Spin up an Onboarding Buddy, a Policy Helper, a Meeting-Prep assistant, or an Account Brief, each one answering only from what your company actually knows, always with sources. Every team gets an expert on tap, and nobody writes a line of code.
Gmail, Drive, Slack, Notion, Jira, and more every week. Connect each one in a few clicks, and from then on new information flows in on its own: no uploads, no copy-paste, always current. Missing a tool? We build the connector for you, on demand.
Bring one real question your team keeps re-asking. We'll connect a source, read-only, and show you the answer, sourced, in seconds, in a 20-minute walkthrough. Nothing changes in your tools.
Every tool is linked over a secured, encrypted connection, your information is protected in transit and at rest.
Answers respect each person's access level, so sensitive information stays with the people meant to see it.
Your company's knowledge is never used to train anyone else's tools. It works for you, and only you.
A search box hands you a list of links and leaves the reading to you. qbrin reads everything for you and gives back a clear, plain-English answer, with the exact email, file, or message it came from linked right underneath. You get the answer, not the homework.
Every answer is built only from your company's own emails, chats, files, and decisions, and every claim links straight back to the original source. You can open it and read it for yourself in one click. If qbrin can't find solid support for something, it keeps looking rather than guessing, so you get a complete, sourced answer or an honest “here's what we have”, never an invented one.
qbrin notices when your own records keep mentioning a tool you haven't connected yet, for example, a system referenced 47 times this month that it can't see. It surfaces that as a recommendation, so the gaps in your company's memory get spotted for you, before they cost you an answer.
Every answer respects each person's existing access level. qbrin mirrors the permissions already set in your tools, so sensitive information only ever reaches the people meant to see it, in answers and in search alike. People can't reach anything through qbrin that they couldn't reach directly.
No. Your company's knowledge is never used to train anyone else's tools. Connections are encrypted end to end and our controls are aligned with GDPR. Your data works for you, and only you.
In a 20-minute walkthrough we connect one source, read-only, and show qbrin answering a real question from your own knowledge, nothing changes in your tools. Connecting Gmail, Drive, Calendar, Slack, or WhatsApp Business takes a few clicks, and from then on new information is read in continuously, so it stays current on its own.