Outside Context Protocol

Outside Context Protocol is a legal training simulation platform for trainee solicitors. It is intended to give junior lawyers a more realistic experience of working on matters, without the risk and time associated in this supervision. Traditionally trainees learn through osmosis, working on elements of matters alongside more senior lawyers and learning by virtue of the red pen.

As technology is enabling senior lawyers to get a lot of this work product without the use of juniors, there is a risk that this causes a skills gap in the future, where junior lawyers are not learning the skills they need to the same level of robustness as their predecessors.

In order to solve this, training lawyers through simulated (but grounded in reality) exercises gives both a chance to experience and learn, but also to supervise and grade the abilities of lawyers coming through. In an era where technology could drive the amount of work goign to juniors down, we may end up in a position where a trainee is expected to clock a certain number of hours in simualted training as well as on real life matters - similar to how pilots are trained.

Supervisors can upload real matter documents, such as anonymised source materials and ideal outputs based on previous matters they have worked on or delivered. The platform uses AI to generate structured, multi-step training exercises that mirror actual legal work. Trainees work through each exercise with AI-powered grading, a virtual supervisor to ask questions, and detailed performance reports on completion.

This isn't intended to replace supervision and training, but adding to the resources a junior lawyer can call upon.

• AI-generated exercises: Upload matter documents and the platform automatically creates structured training exercises with steps, grading criteria, and a briefing narrative
• Multiple step types: Read, draft, email, review, identify, and advise - simulating the range of tasks a trainee handles on a real matter
• AI grading with detailed feedback: Trainee submissions are scored against ideal outputs with breakdowns of strengths, areas for improvement, and critical issues
• Virtual supervisor chat: Trainees can ask questions during the exercise; the AI stays in character as a busy but supportive supervisor, guiding without giving away answers
• Question quality assessment: The platform evaluates whether trainee questions are useful and relevant, encouraging good professional habits
• Step navigation and resubmission: Trainees can revisit earlier steps, review their previous grades, and resubmit for re-grading
• Final performance report: On completion, a narrative assessment summarises the trainee's overall performance across all steps
• Role-based access: Supervisors create and manage exercises; trainees only see published exercises ready to attempt
• Document parsing: Supports PDF, Word (.docx), and plain text uploads with automatic text extraction
• Prebuilt seed exercises: A selection of ready-made exercises included out of the box so trainees can start immediately