What it is
LegalFly is a Ghent-based legal AI platform for in-house teams, built around one architectural decision: every document is anonymised before a model sees it. A fine-tuned model strips names, parties and other identifiers first — on the strictest tier inside the customer’s own environment — and only the redacted text reaches the LLM. The platform is LLM-agnostic and selects the foundation model per legal scenario instead of owning one.
Six work surfaces sit on that layer: Discovery for research and advice across 110+ jurisdictions, Review for playbook-driven contract review and negotiation, Multi Review for due diligence over large file sets, Draft for document generation, Legal Radar for regulatory monitoring, and Agent Studio for building agents against the team’s own playbooks. Word and Outlook add-ins put Review and Draft where the lawyer already is.
The company was founded in April 2023 by four former Tinder product people — Ruben Miessen, Kasper Verbeeck, Dennis Montégnies and Gregory Vekemans — and has raised about €17M: a €2M seed led by redalpine, then €15M+ in Series A announced 2024-07-15, led by Notion Capital with redalpine and Fortino Ventures participating. No Series B has been announced. Offices are in Ghent, London and Dubai.
Why it shows up in Legal Ops stacks
- The data-protection objection gets an architectural answer instead of a DPA. Most legal AI answers “what leaves our building?” with contract terms and a hosting region. LegalFly answers it by removing the sensitive text before the model call, with anonymisation levels you adjust and specific terms you whitelist when redaction would destroy the context the model needs. That is the argument that gets a Belgian or German privacy officer to sign.
- It is built for corporate legal departments, not firms. The CEO puts the client mix at 96% corporates and public sector against 4% law firms — the inverse of Harvey’s centre of gravity. Named users include the European Commission’s HR Directorate-General, under a contract signed in 2026 for HR legal decision workflows, plus SAP, Lufthansa, Allianz and Agristo.
- Deployment goes past managed SaaS. Three tiers: SaaS, single tenant in an Azure region you pick, and single tenant with anonymisation running on-premise. ISO 27001 and SOC 2 Type II are both in place, with annual independent penetration testing, SAML 2.0 / OIDC / OAuth against Okta, Azure AD or Google Workspace, and attribute-based access control.
Pricing
Nothing is published — there is no pricing page, and every route ends at a demo booking. Pricing is enterprise-only and quoted per deal, and three levers move the number: which deployment tier you take, how many seats, and how many of the six modules are in scope. On-premise anonymisation is the expensive end of that first lever.
For a band to negotiate against, the published anchors in this segment are GC AI at $500 per seat per month and Beck-Noxtua at €350 per user per month behind a 3-seat floor. Those are other vendors’ prices, not LegalFly’s, and a multi-module enterprise deployment with a platform fee lands above a single-product seat price rather than below it. Treat a quote under that band as a signal to ask which modules were left out.
Best for
In-house legal, compliance and claims teams at European corporates — banking, insurance and regulated industry above all — where the blocking question is what personal or client data reaches a US-operated model, and the answer has to be “none of it” rather than “it is contractually protected.” The fit sharpens when the team runs contract review, regulatory monitoring and due diligence out of one platform; a team that only needs redlining is buying more surface than it uses.
Alternatives and when to pick them
- Harvey — pick it for firm-side, multi-jurisdiction practice work, and when US-operated model infrastructure is acceptable. Its LexisNexis alliance puts Shepard’s inside the product; LegalFly has no primary-law equivalent.
- Legora — the fastest-growing entrant in the segment, past $100M ARR in April 2026 across 800 law firms and legal teams, with EU data residency in 16 countries. Pick it for breadth and platform maturity, accepting residency-in-EU rather than redaction-before-inference.
- Wordsmith — LegalFly’s CEO names it as the closest competitor, and it is the one to run head-to-head when the primary job is the legal front door: routing and resolving inbound requests rather than reviewing documents.
- Noxtua — pick it when sovereignty means owning the models and the infrastructure, and when beck-online is already your research standard. LegalFly wins on breadth of workflow; Noxtua wins on licensed German publisher content.
Watch-outs
- Anonymisation costs the model context, and the cost is not uniform. Stripping party names weakens exactly the reasoning that depends on who the parties are — change-of-control language, affiliate definitions, group-company carve-outs. Guard: during the pilot, run five clauses where identity is load-bearing through Review twice, once with the parties whitelisted and once without, and make the delta a written acceptance criterion before signing.
- No published price, no self-serve tier, and a reported high seat minimum. Guard: make the first quote itemise the platform fee, the seat floor and the per-module cost separately, and price the SaaS tier alongside single-tenant so the sovereignty premium is a visible line rather than a bundled one.
- No MCP server and no documented public API. LegalFly integrates into SharePoint, Google Drive, Teams, Slack, Outlook and Box, but that is inbound content access, not a developer interface your own agents can call. Guard: if the plan is your assistant querying the legal layer, keep iManage or another MCP-shipping system in the design and treat LegalFly as an end-user seat.
- The company is small relative to what it is being trusted with. About €17M raised against enterprise and public-sector deployments, with a Series B discussed publicly but not closed. Guard: take a 12-month initial term rather than three years, and get source-availability or transition-assistance terms into the contract before the pilot converts.