Edge Studio

The Problem

Leadership teams in technology companies are notoriously hard to assess. Most diagnostic tools are surveys. They’re shallow, easy to game, and produce generic outputs that sit in a PDF and go nowhere.

The brief was different. Build a tool that meets leaders where they are — in a conversation — and turns what they say into a concrete, scored picture of where the business is strong and where it’s exposed. Then turn that picture into a 90-day plan they can actually execute.

How It Was Built

Edge Studio is a React and Node.js monorepo built entirely through Claude Code. There are four packages — frontend, backend, shared types, and an AI service — wired together through npm workspaces and deployed to Vercel as a serverless application backed by a Neon PostgreSQL database.

The assessment covers 12 strategic zones across four pillars: Unified Team, Operating Strength, Competitive Advantage, and Scalable Growth. Leaders can complete it three ways: a written form, a text-based conversation with an AI interviewer, or a fully voice-driven session where the system speaks questions aloud and listens to responses.

Voice sessions use OpenAI’s TTS API to read questions out with a configurable voice persona, and OpenAI Whisper to transcribe spoken answers in real time. There’s silence detection, a media recorder hook that handles device selection, and a browser-side speech synthesis fallback — the whole pipeline runs without storing audio files on disk, which was a hard requirement for Vercel’s serverless environment.

Once all team members have completed the assessment, GPT-4 generates the report. Fourteen separate API calls run in sequence — twelve zone-by-zone analyses, a priorities synthesis, and a roadmap. Each zone produces a scored insight with evidence drawn directly from what respondents actually said. The priorities surface the top cross-zone actions with owners, expected impact, and first steps. The roadmap builds a 13-week Gantt-style execution plan with initiative tracking and decision logging.

The platform includes a full admin layer, magic-link authentication, a registration approval gate for access control, and a real-time report generation flow with polling and animated feedback while GPT-4 works.

What Makes It Interesting

Most AI-assessed tools score against keywords. Edge Studio doesn’t. GPT-4 reads the full context of what each respondent said and scores against three weighted dimensions — clarity, consistency, and capability — with stage-specific weightings that adjust based on whether the company is a startup, a scaleup, or an established business. A startup’s leadership alignment matters more than its org-chart systems. The scoring reflects that.

The voice experience required more precision than expected. Getting reliable transcription meant tuning recording timeslices, handling MIME types across browsers, setting a domain prompt for Whisper so it understood the industry context, and adding minimum audio size checks on both ends to reject near-silent recordings. Small details, but they’re the difference between a voice feature and a voice feature that actually works.

The Result

A production-ready leadership assessment platform, live at edgestudio.uk, with no external development team. Startups and scaleups bring their C-suite through the assessment, get a scored diagnostic across all 12 strategic zones, and leave with a prioritised 90-day roadmap generated from what their own leaders actually said — not from generic templates.