For engineering teams

Fast on the code you know. Coached on the stack you're learning.

TAOS is a coaching layer for Claude Code, Cursor and ChatGPT. It gets out of the way on the code each engineer already knows, and asks for their approach first on the parts they are still learning, so juniors ship with AI and still grow into seniors.

Start your team's 7-day trial → Install the Claude Code plugin

Founding teams: £120 a seat for your first year on the annual plan, with an onboarding call with us. 10 of 10 places left.

Why now

Your juniors ship more code than ever. Do they understand it?

The work that used to teach engineers their craft (reading the docs, debugging the odd failure, writing the boring version first) is now the part AI does.

50% vs 67%

Less understanding, no real speed gain

In a randomised trial of 52 mostly junior engineers learning a new library, the AI group scored 50% on a comprehension quiz against 67% for those coding by hand, and was not significantly faster.

Anthropic research, January 2026.

How, not whether

The way they use AI decides what they keep

The same trial found the engineers who asked follow-up and conceptual questions kept their learning. TAOS makes that the default on the parts of the stack someone is still learning.

Anthropic research, January 2026.

53%

Leaders see juniors developing slower

Of 70 senior leaders surveyed, 53% named the slower development of junior talent as a sign of de-skilling, and only one in ten companies has a plan for it.

BCG, June 2026.

Built by professionals with academic backgrounds at LSE and the University of Oxford. See who is behind it →

How it works for an engineering team

Two commands. Then it runs inside the work.

1

Each engineer maps their skills

A five minute check: the languages, systems and practices they are strong in, the ones they are growing into, and what they want to keep sharp. Private to them.

2

Install where they already work

The Claude Code plugin installs with two commands and coaches from the first message. Cursor and ChatGPT connect as MCP servers. We set the team up together on the onboarding call.

3

Ship as normal

Boilerplate and the stack they know: it just does it. A new library, an unfamiliar system, a design decision: it asks for their approach first, then reviews the reasoning. You see the team's picture, never anyone's sessions.

See it work

From a five-minute chat to a sharper you

Tell it about your work, get a profile, then coaching that builds the skill instead of doing it for you.

Assessment
What do you want to get genuinely better at, not just faster?
Writing the client recommendation. I lean on AI for it and it shows.
And where would you want me to move fast and automate?
Gathering sources and the routine stuff.
Your profile
HD
Harriet DoyleStrategy consultant
7.4Readiness
RecommendationsCoach
Data analysisAugment
SourcingAutomate
FacilitationProtect
Coaching, not autocomplete
Draft the client recommendation from my notes.
Lead with your call. In one line, what should they do and why now?
Just tell them to go with the cheapest vendor.
You're framing it on cost. But this board is judged on delivery risk, and the cheapest vendor is the least proven. Recommend the reliable one and sell it on value, not price.
Actually, you're right. Let's recommend the proven vendor, because reliable delivery protects the timeline and the budget. Can you write up the recommendation?
Yes, on it. Give me a second.
The result
Skill Output quality

Your skill and your output quality climb together.

For engineering leads

What engineering managers ask first

How is this different from Claude Code's Learning output style?

Claude Code's Learning style is good, and free: one setting that makes the whole session more explanatory. TAOS decides per skill, from each engineer's own profile, so it stays fast on what they know and coaches only where they are growing. It also remembers progress over time and follows them across Claude Code, Cursor and ChatGPT. You can use both.

Will it slow the team down?

It coaches the parts someone is still learning, not everything. When they are shipping against a deadline, they can tell it to just do the task this time, and it will.

Can I see my engineers' sessions or code?

No. The team view is aggregate only: how the team is growing and where skills are at risk. Never anyone's conversations, code or profile.

What does it cost?

£200 a seat a year, or £20 a seat a month. Founding teams pay £120 a seat for the first year on the annual plan. Every team starts with a 7-day free trial; we email you two to three days before the first charge, and cancelling is one click.

Who builds TAOS?

Professionals with academic backgrounds at LSE and the University of Oxford, who ran a pilot of TAOS with knowledge workers in 2026. See who is behind it.

What happens on the onboarding call?

Thirty minutes with us. We install the plugin or connectors with the team, check each engineer's profile makes sense, and everyone leaves with prompts written for their own stack.

Start your team's 7-day trial →
Get started

Install in your tool

Same hosted endpoint, sign in with Google or any email, one portable profile. Pick the client you already use.

Install the plugin. You get the coaching layer from turn one, the TAOS skills, and your profile. Two commands inside Claude Code.

/plugin marketplace add angelo-leone/taos-connector
/plugin install taos@taos

Sign in when Claude asks (Google, or a code sent to any email), then say Run my TAOS assessment.

Want the tools on their own, without the coaching layer? claude mcp add taos --transport http https://taoshq.com/mcp connects just the server.

Add the TAOS plugin to Cowork from its public marketplace, then start coaching. No org admin needed.

  1. In Claude, open Customize, choose Add marketplace, then Add from a repository, and paste this owner/repo:
angelo-leone/taos-connector
  1. Click Sync to load the marketplace.
  2. Install the TAOS plugin from it. Sign in when your client asks (Google, or a code sent to any email), so the connector can reach your profile.
  3. Say Run my TAOS assessment, then bring a real task and let it coach you.

Your profile and the coaching method stay on the TAOS server; the plugin only connects you to them. Rolling out to a team? A Team or Enterprise owner can deploy it for everyone: talk to us.

  1. Claude desktop or claude.ai → Customize → Connectors → + → Add custom connector.
  2. Name it TAOS, paste the endpoint, click Add, then sign in with Google or an emailed code.
  3. Ask Claude Run my TAOS assessment.
https://taoshq.com/mcp

Prefer a local install for Claude Desktop? Download the desktop extension (.mcpb) and double-click it. Your profile stays on your machine.

  1. Turn on Settings → Security and login → Developer mode. Leave it on: custom connectors stop working the moment it goes off.
  2. Open Plugins in the left nav and click +. The button only appears once Developer mode is on.
  3. Name it TAOS, paste the endpoint, set Authentication to OAuth, and sign in with Google or an emailed code.
  4. Ask ChatGPT Run my TAOS assessment.
https://taoshq.com/mcp

Gemini does not support custom MCP connectors yet, so TAOS works as a paste-in Gem. Build your profile on the web first:

https://taoshq.com/assess
  1. On your dashboard, click Download → Gemini.
  2. Paste the bundle into a new Gemini Gem called TAOS Coach.

Any MCP-aware client (Cursor, Windsurf, Codex, Gemini CLI) can connect over HTTP with OAuth:

https://taoshq.com/mcp

See the install guide for per-client specifics.

Prefer to self-host? Talk to us →