Esgenix

Esgenix is an AI-native patent drafting platform that uses multiple large language model agents operating as a collaborative "council" to generate and stress-test patent claims and full applications.

Key features and functions include:

Multi-LLM Claim Composer
Generates independent claim variants simultaneously using domain-tuned large language models, each approaching the invention from a different drafting perspective. Users compare, merge, and refine claims across variants rather than relying on a single AI output.

Scope Slider
A visual control that broadens or narrows claim breadth in real time, rewriting claim language and generating multiple coverage variations with inline diff review before changes are accepted.

Prior Art Radar
Surfaces prior art references ranked by similarity score during the drafting process.

Examiner's Lens
Checks claims against USPTO and EPO drafting requirements, including antecedent basis, claim syntax, enablement, unity, clarity, and inventive-step weaknesses.

Additional Analysis Tools
Supports infringement analysis, patentability assessment, redundancy detection, and claim strategy recommendations, including adversarial persona-based reviews intended to simulate examiner objections before filing.

Zero-Prompt Application Builder
Generates a full patent application package — specification, claims, abstract, figures list, and drawings guidance — from an uploaded invention disclosure without requiring users to write prompts.

Patent Counsel Digital Twin
Offers a selection of counsel-style personas (described by the vendor as 20+), including prosecution-focused, litigation-oriented, examiner-friendly, and broad-coverage approaches, to tailor drafting strategy.

Patent Drawings
Generates patent figures — flowcharts, block diagrams, and system architectures — directly from drafted claims and specifications, exported as patent-office-formatted PowerPoint files.

Legal Expert Agent
An agentic assistant that proposes concrete drafting actions, such as adjusting claim scope, running additional analyses, or switching claim types.

Collaboration and Review Tools
Supports inline comments, shared workspaces, version control, and a "Resolve and Amend" function that collates reviewer comments and rewrites claims to address them, subject to diff review before acceptance.

Explainability and Audit Trail
Each drafting decision includes source provenance and a redline audit trail, exportable as a PDF documenting actions taken during drafting.

Security Architecture
Uses a local-first model: invention data is stored in an AES-GCM encrypted vault (.esg file) on the user's machine by default. Optional encrypted cloud storage adds cross-device access; disabling it deletes stored cloud data. The vendor states that customer prompts and drafts are not used to train AI models.

Output Formats
Applications export as patent-office-formatted DOCX files; patent drawings export as PowerPoint files formatted to patent office margin standards.