Read every page, cite every claim
Legal work punishes uncited assertions. Bee grounds answers in your documents, holds entire agreement sets in a 1M-token window, and extracts terms as schema-valid JSON.
Built around how legal teams actually read
A legal answer is only useful with its source. Bee's retrieval pipeline cites the passages behind every answer, so review means checking a reference, not re-reading the corpus. Upload the data room once; the index serves the whole team.
For cross-document reasoning — inconsistent definitions across a contract stack, obligations that conflict between master and local agreements — the 1M-token long-context mode on Comb and Buzz holds the full set in one request instead of retrieving fragments.
From prose to structured terms
- →JSON mode extracts parties, dates, renewal terms, liability caps, and governing law into your schema — validated at generation time.
- →Vision input on Comb, Buzz, and Hive reads scanned agreements and exhibits alongside born-digital documents.
- →Tool calling connects extraction output to matter-management or CLM systems under your control.
- →Document capacity scales by plan, up to six-figure document counts on the top tier.
Confidentiality as a contract term, not a footnote
Customer data is not used to train the Bee base model. Prompts, documents, and outputs stay yours; processing is limited to serving, securing, and operating the service.
Retrieval indexes are per-tenant; encryption is TLS 1.3 in transit and AES-256-GCM at rest under per-tenant keys. The DPA — with EU Standard Contractual Clauses — auto-incorporates into paid plans, and counter-signed copies are available for procurement.
Firms with stricter requirements move down the deployment ladder: single-tenant Enclave Private Cloud from $7,500/mo, up to air-gapped sovereign deployment.
The honest boundary
Bee is drafting and analysis infrastructure for legal professionals — it does not give legal advice. Legal advice to end clients is a restricted (Tier-3) domain gated behind explicit acknowledgement, which is exactly where a bar-regulated profession should want that line drawn.
Frequently asked questions
- Can Bee review an entire data room?
- Retrieval scales to your plan's document capacity with cited answers across the set, and the 1M-token long-context mode on Comb and Buzz holds large agreement sets in a single request for cross-document reasoning.
- Are answers cited to the source documents?
- Yes — retrieval answers carry cite-backs to the exact passages, so every assertion is verifiable against the underlying document.
- Is client material kept confidential?
- Customer data is not used to train the Bee base model, indexes are per-tenant, and encryption applies in transit and at rest. The DPA and sub-processor list are published, with counter-signed copies on request.
- Can Bee handle scanned contracts?
- Yes — Comb, Buzz, and Hive accept image input natively, so scanned agreements and exhibits can be analysed alongside digital text.
- Does Bee provide legal advice?
- No. Bee is professional tooling for legal teams; providing legal advice is a restricted domain under the Acceptable Use Policy, and professional responsibility stays with the lawyers using it.
Related
Start on the free tier
Bee Cell is free — no card. Scale to paid tiers, the API, or sovereign deployment when you are ready.