Ammon Morris, frontend lead and engineering manager who ships AI features in document-heavy products.

The problem

Every lawsuit runs on discovery: one side drafts interrogatories and requests for production, and the other answers them from its own case files. Briefpoint used AI to write the first draft on both sides. My job was making those drafts something an attorney would trust enough to check, edit, and sign.

My role

I managed a team of four frontend and test engineers, planned features with designers, product, AI researchers, and attorneys, and wrote code alongside the team.

What I built

• Interrogatory Answers and Find Documents. Attorneys upload their case files, and the AI drafts an answer to each interrogatory and pulls the responsive documents for each request for production, with Bates citations. Every answer links to its source, so the attorney can check it before it goes out.

• A faster way to draft requests. A searchable library of past requests sits one click from the set, and defined terms like YOU and PLAINTIFF are inline pills that show and edit their definitions, so attorneys start from proven language instead of a blank page.

• A new response editor. I built a working prototype of a new TipTap editor for answering discovery, which the team used as a starting point for the version that shipped.

• Discovery Playbooks. A firm's past objections and responses surface beside the editor while drafting, and one click inserts them, so the whole firm answers the same way.

AI-first development

I led the move to AI-first development, first on my team and then across the company. Engineers assigned tickets to Claude Code and ran several at once in git worktrees. We wrote our conventions down as shared skills, so every agent followed the same rules, and I taught the workflow to the whole engineering team.