A groundbreaking initiative at GitHub sees its legal team leveraging AI to build custom tools, demonstrating that sophisticated AI development is no longer exclusive to seasoned engineers. This approach empowers legal professionals without traditional coding backgrounds to automate complex workflows, significantly boosting productivity and offering a model for other non-technical departments.
Automating Repetitive Legal Tasks
The core challenge addressed by GitHub's legal team was the automation of repetitive tasks inherent in legal work, such as intricate contract reviews and responding to frequently asked legal questions. Historically, these processes demanded considerable manual effort and reliance on pre-existing templates. By utilizing GitHub Copilot CLI, which enables users to articulate desired functionalities in natural language, the team successfully developed tailored solutions without extensive coding. This fostered a culture of innovation, encouraging previously hesitant individuals to explore and build AI-powered solutions.
terms-ai: Streamlining Contract Drafting
Ngandu Kasuku, Principal Product Counsel, spearheaded the development of "terms-ai," a tool designed to revolutionize contract drafting. Kasuku, whose work involves complex data and infrastructure deals, faced the challenge of bespoke partnership agreements. Initially using Copilot CLI for isolated tasks, he recognized the potential for a more integrated system. By establishing a project scaffold that stored essential documents, instructions, and workflows in a repository, Kasuku created a version-controlled environment for his AI prompts. This structured approach not only enhanced consistency but also significantly reduced manual effort.
A key feature of terms-ai is its emphasis on plain language, moving away from archaic legal jargon. Kasuku incorporated principles from a broader movement advocating for clarity in legal drafting. The tool also draws upon a library of previously executed agreements, allowing it to reference existing, approved work when processing new addendums or agreements. Crucially, sensitive information is maintained within a secure, access-controlled internal environment, separate from the open-source workflow. Kasuku reported that terms-ai has effectively halved his review and drafting time, leading to more uniform provisions and a preferred plain-language style.
Automating DMCA Notices and Beyond
Jesse Geraci, Online Safety Counsel, focused on the critical need for rapid analysis of source code in response to Digital Millennium Copyright Act (DMCA) notices. The initial project aimed to transform ad-hoc, individual prompt work into a dependable, repeatable process for fact-gathering and consistent data analysis related to DMCA triage, code comparison, license checks, and circumvention reviews. Geraci expressed surprise at the extent to which he could build functionality without direct engineering support.
Geraci's workflow was built around plain-language files containing instruction sets, policy references, and report templates. By applying his legal drafting acumen, he embedded structured legal judgment directly into the workflow. This evolved to accommodate various analysis modes for different stakeholders and integrate with external data sources. While the transition to a desktop application with a user-friendly interface for predefined legal workflows required substantial coding, the core workflow customization remained accessible via plain language in Markdown. This architectural flexibility has allowed the tool to expand its capabilities beyond DMCA analysis to include contract review, NDA triage, risk assessment, compliance checks, and response drafting. The underlying framework utilizes reusable skills and agents, but the pivotal innovation lies in empowering legal teams to dictate AI behavior through readable Markdown, ensuring human review remains central to a more consistent, transparent, and scalable legal analysis process. This mirrors the principles seen in understanding the github copilot core workflow.
Democratizing Tool-Building
The successes achieved by GitHub's legal and program management teams underscore a significant trend: AI assistants are evolving into powerful instruments not merely for code generation, but for empowering domain experts to construct and customize their own AI-driven solutions. This capability is poised to accelerate digital transformation across diverse departments, shifting development from IT-centric models to a more distributed approach to innovation. The ability for the github legal team builds tools serves as a powerful example of this paradigm shift.
Top comments (0)