DeepIP is an AI-driven patent drafting and prosecution assistant. $15M Series A (Mar 2025), $25M Series B announced (total $40M per LegalTech Talk). NYC and Paris. Ex-Kill Technology founders. IPWatchdog: ‘the Harvey of Patent Law.’ Integrates with MS Word. ISO 27001 + SOC 2 Type II. Supports software, mechanical, electronic, pharma, and biotech patents. Law360, TechFundingNews, Business Insider, Serena Capital coverage. Drafts patents and suggests office action responses.
Company Info
- Founded: 2023
- Team size: 1-10 employees
- Funding: $15M
- HQ: United States
- Sector: Legal Technology, Intellectual Property
What We Haven’t Verified
This page was assembled from publicly available information. Feature claims and workflow mappings are based on what the vendor and third-party listings publish — not hands-on testing or practitioner feedback.
Workflows
Based on practitioner evidence, Deepip is used in these workflows:
What practitioners struggle with
Real frustrations from legal professionals — the problems Deepip addresses (or should address). Sourced from practitioner reviews, Reddit threads, and case studies.
Patent attorney conducting a prior art search for a client's invention spends 2-3 days manually searching USPTO, EPO, and non-patent literature databases — reading hundreds of abstracts, mapping claims to prior art references, and still worrying they missed something in a Chinese or Japanese patent that wasn't translated. The search costs the client $5,000-15,000 and the attorney still can't guarantee completeness
Where it fits in your workflow
Before Deepip
Patent attorney receives invention disclosure → uses DeepIP to draft patent application → AI assists with claims, specification, figures descriptions
After Deepip
Patent drafted → filed with USPTO/EPO → office actions received → DeepIP suggests responses → prosecution completed
Integrations & hand-offs
DeepIP ↔ Microsoft Word (integrated). DeepIP → USPTO/EPO (filing). DeepIP ← office actions.
Also used by similar teams
Community Data
Loading practitioner-sourced data…