Deep-Law is a Turkish-language legal work and production system that brings research, drafting, case management, and configurable AI agents into a single workspace for lawyers practicing under Turkish law.
Key features and functions include:
Petition Studio
Combines file context, sourced legal research, and drafting assistance in an A4-format document editor. Work can begin from a blank document, a template, an existing case file, a selected UYAP filing, or a lawyer-selected research package. Output supports Word, PDF, UDF, and plain text formats, with comment, permissioned sharing, and version tracking retained throughout.
Legal Research and Case Notebooks
Supports research into case law and legislation with visible source and citation trails. Case notebooks organize documents, chronology, analysis, and working notes within a shared file memory, and can generate outputs such as source summaries, timelines, and evidence maps. A Podcast Studio function produces a text transcript and audio summary based on selected sources.
Contract Studio
Introduces a read-only preview of planned contract review and redlining functionality. New drafting, commenting, sharing, and permission changes are not yet enabled in this module.
UYAP and UETS Context
A user-controlled browser extension attached to the lawyer's active portal session brings UYAP filing and document context, and PTT UETS notification records, into the workspace for review, including surfacing possible statutory deadlines. The vendor states this is not an official government partnership; official portal actions, deadline entries, and filings remain with the lawyer.
Agent Studio
Allows a lawyer to configure an AI agent's professional role, approved knowledge sources, skill sets, model routing policy, communication channels, and authority boundaries. Agents can coordinate tasks and produce auditable outputs but do not independently file documents, apply electronic signatures, or make final legal determinations; consequential actions require explicit lawyer approval.
Model Routing
Draws on models from seven providers (including OpenAI, Anthropic, Google, and DeepSeek) along with research-oriented tools from Perplexity, Tavily, and Exa. Four routing modes are offered — an automatic optimized mode, single-model mode, role-based routing, and a multi-step orchestration mode — with availability varying by task, plan, and workspace surface.
Loading...