Subsumio is not a licensed case-law database and not a Word-native drafting tool. It's the knowledge layer over your firm's OWN files — with drafting, redlining and public-source legal research built in. Here is exactly where the market leaders are better — and where we are.
Snapshot: June 2026. Third-party data from public sources (linked below). Prices marked “individual” are not published by the vendor.
If you need any of the following as your primary tool, buy it from the companies below — they are genuinely good at it: legal research against licensed case-law databases (CoCounsel/Westlaw, Lexis+ AI, vLex, Beck-Noxtua/beck-online), M&A due-diligence workflows (Luminance, Harvey). Subsumio does have contract drafting, redlining and public-source legal research — but our core is a different question: what does YOUR firm already know — across matters, emails, memos and meetings — with citations and explicit gaps.
Harvey, Legora, CoCounsel (Thomson Reuters), Beck-Noxtua, Luminance — against the knowledge layer.
| Feature | Subsumio | Harvey | Legora | CoCounsel | Beck-Noxtua | Luminance |
|---|---|---|---|---|---|---|
| Legal research w/ case-law database | RIS-OGD (AT), openlegaldata (DE), OpenCaseLaw (CH) — public sources, no licensed DB | (LexisNexis partnership) | Ja | Westlaw — category leader | beck-online, German law | contract focus |
| Drafting / contract redlining | contract drafts, redlining, pleadings, playbooks | Ja | Ja | Ja | Ja | (contracts) |
| Legal-specific model / training | ZeroEntropy Legal v1 (specialized legal-domain model) + general LLMs via API | building custom legal foundation models (Harvey Labs, June 2026) | model-agnostic (OpenAI + Anthropic) | Ja | own German legal AI trained on beck-online | proprietary legal LLM (150M+ documents) |
| Public legal benchmark (Vals VLAIR) | not tested — retrieval benchmarks only | top scores; also open-sourced Legal Agent Benchmark (LAB, May 2026) | k. A. | top scores | k. A. | k. A. |
| Q&A over your own documents, cited | page-level citations | Ja | Ja | Ja | Ja | Ja |
| Cross-matter contradiction surfacing | Dream Cycle, automatic | k. A. | k. A. | k. A. | k. A. | in contract review |
| Typed knowledge graph over your data | extracted on every write | k. A. | k. A. | Nein | k. A. | Nein |
| Gap analysis (“what's NOT in the file”) | in every answer | k. A. | k. A. | k. A. | k. A. | k. A. |
| Beyond legal (deals, clients, firm memory) | cross-domain by design | legal-focused | Nein | Nein | Nein | Nein |
| Compounding firm brain (learns from every matter) | your brain knows every case you've run | k. A. | k. A. | k. A. | k. A. | k. A. |
| Self-hosting / on-premise option | full engine on your hardware | cloud | cloud | cloud | sovereign EU cloud, no self-host | on-premise offered |
| No training on your data | never trains shared models | never trains on customer data by default | zero AI training on your data | k. A. | does not use customer data for model training | no third-party data sharing |
| Published entry price | €299/seat/mo (Starter, monthly) · €690/seat/mo (Professional, annual) | individual (reports: ~$1,200/seat/mo, 20-seat minimum; $300M ARR, $11B valuation) | individual (reports: ~$3,000/user/yr, 10-seat min = $30k/yr floor; $100M ARR, $5.6B valuation) | $104–$639/user/mo depending on plan (public configurator, ≤10 attorneys); bundled with Westlaw — no standalone CoCounsel | €350/user/mo self-service (3 licenses = €1,050/mo); €410/user/mo after 12 months | individual (enterprise-only; reports: $50k+/yr; $30M ARR) |
| Seat minimum | from 1 (Starter) | ~20 (reports) | 10 seats (reports) | solo plans exist (online up to 10 attorneys) | 3 licenses | k. A. |
| Agentic AI / multi-step workflows | agent system with skills + eval | 25,000+ custom agents, Agent Builder (Mar 2026) | Agent, Workflows builder | agentic workflows, Deep Research | function combination across research/analysis/drafting | multi-agent architecture (Jan 2026 update) |
| Word / Outlook integration | Word Add-in | Word, deeper MS 365 integration | Word + Outlook add-ins | Microsoft 365 integration | Microsoft integration | Word, Outlook, Salesforce, VDRs |
| Mobile app | native mobile app | Harvey Mobile | k. A. | k. A. | k. A. | k. A. |
Legal research w/ case-law database
Drafting / contract redlining
Legal-specific model / training
Public legal benchmark (Vals VLAIR)
Q&A over your own documents, cited
Cross-matter contradiction surfacing
Typed knowledge graph over your data
Gap analysis (“what's NOT in the file”)
Beyond legal (deals, clients, firm memory)
Compounding firm brain (learns from every matter)
Self-hosting / on-premise option
No training on your data
Published entry price
Seat minimum
Agentic AI / multi-step workflows
Word / Outlook integration
Mobile app
The questions a DACH/EU firm's IT and data-protection officer ask first — and where Subsumio already has a shipped answer.
| Feature | Subsumio | Harvey | CoCounsel | Beck-Noxtua | Glean |
|---|---|---|---|---|---|
| EU AI Act Art. 50 — AI output labeled (visible + machine-readable) | badge + frontmatter + X-AI-Generated header | k. A. | k. A. | k. A. | k. A. |
| Four-eyes / human approval gate for agent actions | approval queue, documented reason | k. A. | k. A. | k. A. | k. A. |
| Source-coverage / hallucination-caution badge on answers | grounded / partial / unsupported | k. A. | k. A. | k. A. | k. A. |
| GoBD building blocks (10-yr retention stamp, tamper-evidence, Verfahrensdoku) | for law firms — building blocks, not a “audit-proof” claim | Nein | Nein | Nein | Nein |
| DACH legal interfaces (beA drafts, RVG, conflict check, RIS/openlegaldata) | native | Nein | Nein | German law focus, § 203 StGB / § 43a BRAO compliant | Nein |
| MFA / 2FA (TOTP) for the dashboard | built in | via SSO/IdP | via SSO/IdP | k. A. | via SSO/IdP |
| Team / org model with roles & invites | org entity, member/admin/owner | enterprise admin | k. A. | k. A. | enterprise admin console |
| GDPR Art. 20 data export (account + brain, JSON) | one click | k. A. | k. A. | k. A. | k. A. |
| SSO (SAML/OIDC) + SCIM | on the roadmap (WorkOS path) | SAML SSO, audit logs, IP allow-listing | SSO integration | k. A. | category standard |
| SOC 2 / ISO 27001 certification | not yet — self-host compensates | SOC 2 Type II + ISO 27001 | (Thomson Reuters) | k. A. | Ja |
EU AI Act Art. 50 — AI output labeled (visible + machine-readable)
Four-eyes / human approval gate for agent actions
Source-coverage / hallucination-caution badge on answers
GoBD building blocks (10-yr retention stamp, tamper-evidence, Verfahrensdoku)
DACH legal interfaces (beA drafts, RVG, conflict check, RIS/openlegaldata)
MFA / 2FA (TOTP) for the dashboard
Team / org model with roles & invites
GDPR Art. 20 data export (account + brain, JSON)
SSO (SAML/OIDC) + SCIM
SOC 2 / ISO 27001 certification
Glean and Notion AI organize company knowledge — the closest neighbors to what Subsumio actually does.
| Feature | Subsumio | Glean | Notion AI |
|---|---|---|---|
| Q&A over company knowledge, cited | + explicit gap analysis | Ja | within Notion content |
| Knowledge graph | typed edges (invested_in, works_at, …), queryable | enterprise graph (people, docs, activity) | Nein |
| SaaS connectors (Slack, Jira, Drive, …) | folders, email, MCP, API — no 100+ connector catalog | 100+ connectors — category leader | Notion ecosystem |
| Document ingestion (PDF, DOCX, EML, XLSX, audio) | native, incl. scanned-PDF OCR fallback | via connectors | imports |
| Self-hosting / on-premise | full engine on your hardware | SaaS-only | SaaS-only |
| No training on your data | never trains shared models | k. A. | k. A. |
| Entry price & minimum | €299/seat/mo Starter · €690 Professional, from 1 seat | $50/user/mo base + ~$15/user/mo Work AI add-on (reports: ~$65/user/mo all-in, ~100-seat minimum, ~$60k/yr entry) | $20/user/mo (Business plan, AI included; retired standalone $8 add-on May 2025) |
Q&A over company knowledge, cited
Knowledge graph
SaaS connectors (Slack, Jira, Drive, …)
Document ingestion (PDF, DOCX, EML, XLSX, audio)
Self-hosting / on-premise
No training on your data
Entry price & minimum
Because we publish the rows we lose: no licensed case-law database, no legal benchmark participation, fewer connectors than Glean. Every third-party claim links to a public source, unknowns are marked “k. A.” instead of guessed, and we publish corrections.
Only if what you actually need is the knowledge layer, not the licensed-research workspace. Many firms will run both: Harvey/Noxtua for licensed research, Subsumio for drafting, redlining and queryable memory over their own matters. We replace neither Westlaw nor a lawyer.
Because they run on someone else's cloud and someone else's model, and they index licensed case law. Subsumio indexes YOUR matters and runs on infrastructure you control — self-hosted or EU cloud — with a WhatsApp copilot your lawyers use daily. Same calibre of synthesis, your jurisdiction, your data. The honest catch: we search public legal sources (RIS, openlegaldata), not licensed databases like Westlaw or beck-online.
Correct — and we say so in the table. The difference: with Subsumio it's your own matter files made answerable, on your own hardware, with deterministic page-level citations your IT can verify.
All third-party information from public sources as of June 2026, linked below; no guarantee of completeness. Vendors change pricing and features — corrections welcome at hello@subsum.eu and will be published.