DeepIP

What it is

DeepIP is an AI patent platform that runs inside Microsoft Word instead of asking attorneys to move into a new editor. It drafts applications from invention disclosures, generates office action responses, runs prior art, patentability, freedom-to-operate, and invalidity searches, and produces patent drawings — all from an add-in beside the document the attorney already has open. An Outlook integration handles deadlines, and connectors to IP management and docketing systems plus a REST API sit underneath.

François-Xavier Leduc and Edouard d’Archimbaud, the CEO and CTO of the AI data company Kili Technology, founded it in 2024, with headquarters in Paris and New York. That placement is the product thesis. Solve Intelligence is a browser editor; Patlytics started at the litigation end with claim charts. DeepIP bets that the attorney who drafts in Word will not change editors, and that the software winning the prosecution desk is the one that meets that attorney where the work already happens.

Funding came fast: a $15M Series A on 27 March 2025 led by Resonance, then a $25M Series B on 3 March 2026 co-led by Korelya Capital and Serena, with Headline and Balderton participating — $40M total. On 12 June 2026 DeepIP acquired PatentMaker, the Munich drafting tool built by German and European patent attorney Dr Matthias Hofmann, which the companies say is used by nearly half of Germany’s top IP firms plus Infineon and Siemens. DeepIP opened its German office inside Boehmert & Boehmert’s premises, and existing PatentMaker contracts continue unchanged while the two products merge over time.

Why it shows up in Legal Ops stacks

• Zero editor migration. Drafting, review, and office action work happen in the Word document that eventually gets filed. For a firm whose templates, macros, and review habits all live in Word, that removes the adoption fight that kills most drafting pilots. The vendor claims adoption up to 20% higher and usage 40% higher than standalone tools — its own figures, not audited.
• Prosecution, not just first drafts. Office action responses come with rejection mapping and case-law citations, and the search modules cover prior art, patentability, FTO, and invalidity in the same place. That is the answer to “which platform covers prosecution after the application is filed?”
• The security file clears a European review. ISO 42001, ISO 27001, and SOC 2 Type II, GDPR, and compliance with §203 StGB, the German professional-secrecy statute that matters to any Munich or Düsseldorf firm. Models run on Microsoft Azure through a zero-retention arrangement, and the vendor states it retains no client data and does not train on it.
• Scale signals. 400+ law firms and corporate IP teams across 25 jurisdictions in the June 2026 announcement (the homepage now says 500+), 25,000+ applications drafted, more than half of the top 50 North American IP firms, and named customers including Greenberg Traurig, Morgan Lewis, Philips, Dexcom, and Mewburn Ellis. Vendor numbers throughout.

Pricing reality

No public price list. The site offers a demo and a free trial behind a form, and enterprise deals go through sales.

The most cited figure is $350–420 per user per month, from a January 2026 comparison published by Patentext, a competing patent tool — an anchor, not a quote. Solo and small-team estimates on review sites run lower per head, but they are unsourced and we would not budget on them. On the same Patentext comparison Solve Intelligence lands near $775 per user per month, so DeepIP is likely the cheaper full-lifecycle seat by a factor of roughly two. For ten drafters, that puts the drafting layer at about $42,000–50,000 a year before negotiation.

Put three things in the order form: which modules the seat includes (drafting, office actions, and search are listed as separate modules), whether the trial converts to an annual commitment, and — if you are a PatentMaker customer — the migration terms and the date your current contract rolls onto the merged platform.

Best for

Patent prosecutors at IP boutiques and AmLaw firms, plus in-house IP counsel at hardware and medtech companies, who draft and respond to office actions in Microsoft Word and need one tool across drafting, prosecution, and search — especially teams with a European or German docket after the PatentMaker deal.

Not for you if evidence-of-use claim charting for litigation is your recurring cost: Patlytics built that deliverable first. Not for you if your team already lives in a browser editor and wants figures, chemical structures, and review in one web workspace — Solve fits that better. Not for you if you need docketing, annuities, or IP spend management — Anaqua stays on the invoice. And not for you if you want to drive patent work from an MCP client: DeepIP publishes no MCP server.

Alternatives — and when to pick them instead

• Solve Intelligence — the best-funded drafting peer, $55M raised, 700+ IP teams, with EPO and UPC workflows from its Palito acquisition. Pick it when attorneys are willing to work in a browser and filing volume is the bottleneck.
• Patlytics — $40M Series B in April 2026 and a read-only MCP server. Pick it when claim charting and infringement detection drive the budget.
• PatSnap, Clarivate Derwent, and Questel — the incumbents by market share on patent data and analytics. Pick one when the job is landscape analysis and a defensible data license rather than drafted work product.
• Harvey or Legora — generalist legal AI. Pick it when patents are a small slice of a mixed docket and vendor consolidation beats specialist depth.

Watch-outs

• Two products are being merged. PatentMaker customers were promised continuity, but “brought together over time” has no published date, and a merged platform means changed workflows for someone. Guard: get the PatentMaker support end date and the data-migration path in writing before you renew either product.
• The vendor’s own numbers disagree. 400+ versus 500+ customers, and the search index described as both 120M+ and 150M+ patents on the same site. None of it is audited. Guard: measure attorney-hours per application across ten filings before the trial and ten during it; that is the only figure that survives renewal.
• Output quality varies by art unit. Independent review notes more revision is needed for biotech, pharma, and chemistry applications than for software and mechanical work, and that §101 Alice/Mayo rejections still need heavy attorney input. Guard: run the trial in your hardest art unit and on your hardest rejection type, not your easiest.
• “Connects to docketing” is not a named list. The site says leading IP management platforms without naming which ones. Guard: make integration with your specific IPMS a contractual acceptance criterion, tested during the trial, before signature.
• Model dependency on one provider. Generation runs on OpenAI models through Azure. Guard: confirm in the DPA which Azure region processes your data and what notice you get if the model provider changes.