JTJoe Taylor
Product design engineering leader · Innovation lab manager · Venture-backed founder

I turn ideas into working products in days.

For 18+ years I’ve turned ambiguous ideas into products people use, at startups and big tech in San Francisco, New York and London. I’ve led teams of engineers, product managers and designers, but I still get hands-on, designing and building the work myself. Today that means working AI-first, from first prototype to production.

Design case study: LaserOwl

A detailed walkthrough of my thinking behind designing LaserOwl to be AI-native, and how I built it all myself in a couple of weeks (skip to 2:00 mark to avoid my rambling introduction).

Work
Featured product · 2026

LaserOwl#

Client-ready candidate documents for recruitment firms, in seconds.

What it is
An AI-native platform that turns raw inputs such as candidate CVs into polished, firm-branded documents ready to send to clients. Recruiters refine each one in conversation with an agent that edits the document directly, saves every change as a new version, and remembers what they've said about the candidate.
Who it's for
Recruitment firms whose consultants reformat every candidate CV into the firm's own style before it goes to a client, and who want that work done in seconds rather than by hand.
What I did
Everything, solo, in a few weeks: the product model and UX, its own design system of composed shadcn/ui components, the agent and document pipeline, production infrastructure and billing.
How it's built
A public design system, with principles for AI that is built in, not bolted on. Read the design system
Where it is
Live, and drawing interest from recruitment companies.
Next.jsTailwindshadcn/uiPostgresRailwayOpenAIStripe
Featured engagement · Jul 2024 – Feb 2025

Harness.io#

Reimagining a developer platform's Code Repository for the AI era.

The project
A multi-year reimagining of the Harness.io Code Repository: remapped developer flows and the first AI experience in the product, since previewed publicly as Agent-Ready Code Repository with AI Code Review.
The team
Brought in by Harness.io's SVP of Engineering to unblock a tiger team of product and frontend engineers across several time zones, working remotely from the UK and on-site in Mountain View.
What I did
When the team couldn't get close enough to the external agency's Figma designs, I built 50 screens in real code to match them. I remapped the core flows using product telemetry, prototyped how AI could surface organically instead of being bolted on, and set up the experience design (XD) team to own the design system, components and accessibility. I built the initial design system and shadcn/ui components the product was then built on.
Where it is
Initially a three-month contract, extended by a further four months. The work has since been previewed publicly.
ProductUXDesign systemsDeveloper toolsAI
Watch the preview
Product · 2026

Thesara#

Evidence and confidence for any proposition, tracked over time.

What it is
A thesis and proposition evaluation product. It breaks a proposition into assumptions, searches for evidence that supports and contradicts each one, and tracks how that evidence changes confidence over time.
Who it's for
Anyone testing a thesis or proposition who needs to see how the evidence, and their confidence in it, changes over time.
What I did
I worked directly across the product model, interaction design, frontend and the LLM research pipeline.
Where it is
A working product, live at app.thesara.io.
Next.jsshadcn/uiPostgresRailwayClaudeOpenAI
Try Thesara
Product · 2025–2026

Waypoint#

The reasoning behind AI-written code, in the pull request.

What it is
A reviewer can't ask an agent why it did what it did. Waypoint streams a coding session's decisions and actions as they happen and makes them available as rationale in the pull request.
Who it's for
Teams building with AI coding agents, and the reviewers who have to approve what those agents write.
How it works
Hooks and MCP tools capture each session as it happens. It works with Claude, VS Code and Codex.
What I did
I designed and built it hands-on, as a working product rather than a slide-deck concept.
Where it is
A working product you can try at app.getwaypoint.dev.
Next.jsshadcn/uiPostgresDrizzleRailwayClaudeMCPHooks
Try Waypoint
ExperienceFull CV
Pug UX
Ran a studio building responsive web apps for technology and entertainment clients.
About

I work across product, design and code. For 18+ years I’ve turned ambiguous ideas into shipped products at startups and big tech in San Francisco, New York and London, designing the experience and writing the production code myself.

I like to get something real in front of people early. At Cisco’s innovation lab we put working pilots in front of internal teams and customers and iterated on usage and interviews. At Harness.io I used product telemetry to rethink core flows. Today I build working apps as high-fidelity prototypes, so teams learn from real behaviour rather than static mockups.

I’m a builder who also leads. I’ve hired and managed teams of managers, engineers, product managers and designers, and I embed in the teams I work with, joining stand-ups and planning and helping product and engineering make decisions and keep moving.

AI has changed the speed at which I can do all of this. I build with Next.js, Tailwind, shadcn/ui and Postgres, accelerated with Claude Code, Codex and MCP, and bring the design and engineering judgement to make sure what gets built is the right thing, done well.

How I think about AI-native design

Principles from the LaserOwl design system, and the same thinking I brought to Harness.io.

Conversation left, artefact right

The chat sits beside the work, never over it or away from it.

Just in time, not on demand

AI appears where the next step is, grounded in the page you are on.

The person stays in charge

Show what it is working on, mark what it changed, and keep every change reviewable and undoable. It proposes; they decide.

Quiet when there is nothing to say

No unprompted chatter and no badge on every control.

Built in, not bolted on

No assistant tab. The conversation stands beside the artefact, arrives just in time, and offers the next step.

Work with me

Alongside lead roles, I take on selected projects.

Focused product sprint

Bring me a product idea, new feature or difficult flow. I’ll take it from rough problem through product thinking, UX and working software.

2–5 daysProduct, design & frontend by meNo handoffs

Product experience review

A focused review for software that works but isn’t good enough yet. I use the product deeply, identify where the experience breaks down, and return concrete recommendations and, where useful, working improvements.

2–3 days£3,000 fixed
I work with
Founders

Making an idea or hypothesis real enough to test, without building a team around it first.

Product teams

Exploring new products and features, or fixing workflows that aren't working well enough.

Engineering teams

Bridging product, UX and implementation, either building directly or working alongside your frontend engineers.

Innovation & AI teams

Rapidly prototyping and testing new ideas to understand what's actually possible before making a larger investment.

Contact

Have something you want to make real?

If you have an idea you’re not sure is worth building yet, or something that needs to become real quickly, I’d like to hear about it.