LabChat · Per-lab AI, bespoke deploy
Ask your lab notebooks anything.
An AI assistant grounded in your lab’s notebooks, papers, and protocols — including the handwritten and scanned lab notebooks that were never searchable before. Every answer cites the exact passage it came from. Your corpus stays in your lab’s index. We don’t train on it, and answers never pull from the open web.
LabChat is deployed per lab — each lab gets its own index and URL. Setup usually takes about a day.
What it does
Grounded in your lab’s research.
LabChat indexes your lab’s notebooks — including the handwritten and scanned ones — alongside your published papers and internal protocols. Answers come back in plain English, with a numbered citation behind every claim.
Your lab notebooks, finally searchable
Handwritten and scanned lab notebooks were never searchable before. LabChat reads them and answers questions from them — with a citation back to the exact page.
Grounded in your papers
Every answer includes numbered citations to the specific paper or protocol PDF it came from. Answers only pull from your lab's index.
Lab onboarding
New lab members ask questions about protocols, methods, and past experiments — and get answers drawn from the lab's own work. Tribal knowledge becomes queryable.
Shows multiple sources
LabChat surfaces every retrieved passage behind an answer, not just one. When a protocol was updated mid-cohort or a finding was revised, both versions show up so you can read them side-by-side.
In the conversation
Cited answers, every time.
LabChat reads your lab’s corpus and writes grounded, referenced responses. Numbers in superscript link to the exact paragraph in a notebook, paper, or protocol PDF.
- Inline citations to notebooks, papers, and protocol PDFs
- Every source card links back to the original notebook page or PDF
- Regenerate, copy, thumbs-up — tuned by your team's feedback
- Full audit log every PI can review
How it works
Three steps to grounded answers.
No model training, no fine-tuning. LabChat builds a private retrieval index from your lab’s corpus and uses it at query time.
01
Upload your corpus
Add your lab's notebooks, papers, and protocol PDFs — including scanned and handwritten notebooks. We ingest and index them into your private corpus.
02
Ask a question
Type in plain English. LabChat searches your indexed corpus and retrieves the most relevant passages from your notebooks, papers, and protocols.
03
Get a cited answer
Responses include numbered citations linking to specific sources. Share threads with collaborators, cite them in lab meetings, reference them later.
Private by default
Your lab’s corpus
stays in your lab’s index.
Each lab gets its own encryption keys, its own URL, and its own retrieval boundary — nothing crosses between labs.
Encryption
AES-256 at rest, TLS 1.3 in transit
Per-tenant encryption keys managed in AWS KMS with automatic rotation.
Isolation
Private per-lab index
No cross-tenant retrieval. Your papers never appear in another lab's results.
Audit
Full query log
Every question, retrieved passage, and answer is logged for your PI's review.
Compliance
HIPAA-ready, IRB-aware
PHI-flagged documents stay inside your HIPAA boundary. Consent metadata surfaces at retrieval time.
Your lab’s knowledge, queryable, in about a day.
LabChat is deployed per lab, on dedicated infrastructure. You send us your corpus, we ingest it and stand up your URL, you start asking questions. Setup is usually a day. The team-onboarding payoff is forever.
LabChat isn’t a self-serve sign-up — each deploy is configured against your lab’s corpus. Every inquiry gets a response within 1 business day.