The LB—Tech showreel poster: cool teal line-work on the left and an ember forge on the right, converging to a single point of light over an ink field.

Showreel 2026 · 10 sec loop · Sound off

AI expert · Solo practice · Folio 2026

Lee Barry.

I train AI teams to get materially more from AI — and I design, build and ship production apps. One expert, end to end.

United KingdomAvailable now

Scene 02 — Two ways in

Two doors. One Lee.

No account managers, no hand-offs — whichever door you take, you work with me directly.

Rendered plate: scattered cool points gathering into ascending ordered rows with a rising leader line — a team's capability taking shape.

Train your AI team

Practical enablement for teams who have the tools but not yet the leverage — workflows, prompt craft and delivery practices, taught on your real work.

  • Workflows
  • Enablement
  • Playbooks
Rendered plate: blueprint arcs and construction lines converging into a glowing forged ingot — an idea becoming a product.

Build an app with Lee

Direct senior delivery from discovery to launch — the person you brief is the person who ships.

  • Discovery
  • Prototype
  • Engineering
  • Launch
Training scene poster: a field of scattered teal points settling into ordered rows — a team's capability taking shape.

Scene 03 — Train your AI team

Your team,
fluent in AI.

AI tools are everywhere. Leverage isn't. I turn scattered AI usage into shared, practical capability — taught on your real work, not slideware.

Who it’s for

Engineering, product and operations teams adopting AI — and the leaders accountable for turning AI spend into delivery.

What I teach

Agentic workflows

Multi-step AI work designed to hold up in production, not just in demos.

Prompt & context engineering

Reliable output on purpose — structuring context so the model has what it needs.

AI-assisted delivery

Code review, testing, documentation and refactoring with AI in the loop, safely.

Evaluation habits

Knowing where AI earns its place, where it doesn't, and what it costs.

How an engagement works

  1. Diagnose

    Where AI actually helps your work — and where it's theatre. A written assessment, yours to keep.

  2. Workshops

    Focused, hands-on sessions on your codebase and your documents — never synthetic examples.

  3. Embed

    Pairing with your team on live work until the new workflows stick.

  4. Playbook

    Your team's own documented practice — prompts, workflows, review habits. No dependency on me.

Outcomes — What teams pursue: shorter delivery cycles with AI in the loop; repeatable workflows instead of individual tricks; the judgment to use AI where it earns its place. Measures are agreed up front — no magic, no theatre.

App build scene poster: a bright pulse travelling a pipeline of five stations toward a glowing delivery point.

Scene 04 — Build an app with Lee

Brief me.
I ship it.

Direct senior delivery, from first conversation to live product. You work with me — the person who designs it is the person who ships it.

  1. 01

    Discovery

    Goals, constraints and the sharpest version of the idea. You get a written plan with real numbers, not a sales deck.

  2. 02

    Prototype

    A clickable version in days. Decisions get made on evidence, not imagination.

  3. 03

    Engineering

    Production build: typed, tested, observable. The unglamorous work that makes software survive.

  4. 04

    Launch

    Deployment, monitoring and handover. You own everything — code, infrastructure, knowledge.

The proof — Healthcare, legal, education and data — the selected works below are the proof.

Intermission reel poster: a warm structural lattice crossed by a sweeping ember scan line.

Intermission

From first commit to live deployment — software that survives production.

Scene 05 — Selected Works

01 / 06

Healthcare · CRM — 2025

Kinrae

A UK care CRM spanning domiciliary care, supported living, children's homes and complex care — one record system for the whole organisation.

  • Next.js
  • TypeScript
  • PostgreSQL
Rendered plate: a network of glowing care-record nodes linked above a pulse trace, ember light on an ink field.
Fig. 01 — Care network

Legal-Tech · Platform — 2026

Matilda

Operations platform for Texas plaintiff personal-injury firms — medical chronologies, real-time AI deposition transcription and case intelligence.

  • TypeScript
  • Electron
  • AI Pipelines
Rendered plate: columns of blurred legal transcript text with one deposition line held in sharp amber focus.
Fig. 02 — Transcript streams

Education · MIS — 2026

AP System

Management information system for UK Alternative Provision — safeguarding, compliance-gated scheduling, billing and gross-profit in one place.

  • Next.js
  • Neon
  • Clerk
Rendered plate: a regimented lattice of timetable blocks, a few lit teal with flagged status dots.
Fig. 03 — Timetable lattice

Data · Conversational BI — 2024

ChatBI / Elev8

Conversational business-intelligence — natural-language questions over live operational data, answered with AI-generated insight and charts.

  • Python
  • TypeScript
  • LLMs
Rendered plate: electric-blue signal bars rising from a dialogue baseline with a soft reflection.
Fig. 04 — Conversational signal

Desktop · AI — 2025

FlowScribe

Voice dictation desktop app with fully local Whisper transcription — private, fast and offline by design.

  • TypeScript
  • Whisper
  • Electron
Rendered plate: layered violet spectrogram ridgelines crossed by a single transcription caret.
Fig. 05 — Spectrogram ridges

Security · Tooling — 2025

Vibeguard

A security scanner purpose-built for AI-generated applications — catching the vulnerabilities vibe-coding leaves behind.

  • Python
  • Static Analysis
Rendered plate: a radar display mid-sweep, threat pings flaring inside the scanned arc in signal red.
Fig. 06 — Radar sweep

Scene 06 — The Practice

/01

Web Platforms

Multi-tenant SaaS, CRMs and management systems — Next.js, React and TypeScript on Vercel with Postgres underneath.

/02

Desktop & Native

Electron, Tauri + Rust, Kotlin Multiplatform and Swift — real apps for real desks, on-prem or connected.

/03

AI & Automation

LLM integrations, transcription pipelines, conversational BI and scraping systems that do the work of teams.

/04

Data & Analytics

BI platforms, Snowflake pipelines, reporting and dashboards — operational data turned into decisions.

/05

Security & Tooling

Security scanning, developer tooling, proxies and CLIs — the sharp edges that keep everything else safe.

90+

Projects in the vault

6

Sectors shipped to

7

Languages in production

3

Platforms — web, desktop, mobile

Final Scene — Start here

Start here.

Path 01

Train your AI team

Tell me about your team, your stack, and where AI should be earning its keep.

Email about training

lee@lb-tech.co.uk · subject pre-filled

Path 02

Build an app with Lee

Tell me what you're building, why now, and when it needs to exist.

Email about an app build

lee@lb-tech.co.uk · subject pre-filled

Currently available for training engagements and app contracts — UK & remote.

github.com/korallis ↗

Also showing — the archive

  • Fortis Secure — Healthcare transport dispatch
  • Leo Analytics — Snowflake operational intelligence
  • On Your Bike — Motorcycle instructor platform
  • Workflow Kit — Tauri + Rust desktop tooling
  • WoWPro — Kotlin Multiplatform addon manager
  • Ez-Prox — macOS AI proxy installer