An IT consultancy for AI automation and AI agents.

PN & CPS Technology Ltd is an actively trading UK IT consultancy specialising in AI automation, custom AI agents and business process optimisation. We provide consultancy and technical implementation services to UK and European businesses, working on client projects and professional services engagements.

What we do.

We help UK and European businesses decide what to automate, build it and keep it working. See the services page for the detail.

AI consultancy and automation strategy
Where AI and automation will pay off in your business, and where they will not.
Custom AI agent development
Agents built around your tools, your data and your approval steps.
Business process and workflow automation
Repeated manual work turned into workflows that run on their own.
Software integration and optimisation
Your existing systems connected, and the slow parts made faster.
Technical consultancy and support
Hands-on technical help during a project and after it goes live.

Our own tool.

We also make Command Center, our own tool for running AI coding agents. It is available to other teams in early access.

Command Center

Windows 10 and 11.

Early access

Run up to 16 Claude Code sessions in one window and see the moment one needs you. A coordinator splits the work, and an audit gate reviews every deliverable before it reaches you.

Why we built Command Center.

09:10

Coordinator plans the releaseTeam Hub, redaction, one re-send. Four child sessions.

plan

13:40

Children report backEach with the tests it ran and their output.

4 reports

14:05

Audit gate: FAIL, then PASSA token in a nested URL slipped past redaction. Fixed, re-run.

PASS

14:30

Checkpoint, waiting for usPublish the release?

asks

Fig. 1An example of how a release goes through Command Center. The last step waits for a person.

We ran many agents ourselves, every day, and kept losing track of which one needed us.

It started with three Claude Code sessions in three terminals. Then six, then a dozen, spread over two PCs. The agents were good at their work. We were the bottleneck: a session would ask a question and sit there for forty minutes while we were reading a different one. Another would finish, report a pass, and nobody would check whether the thing it built actually ran.

So we built the tool we wanted. One window with every session side by side. An amber cell and a sound the moment an agent asks something. A coordinator that splits a job into child sessions and reads their reports. An auditor that reviews every deliverable before it reaches us.

We use it on our own work every day, and Command Center is built with Command Center. See what is in the first build.

What we believe.

Four rules shape the product.

Local first.

Your code, your transcripts and your reports live in folders on your own PC. The app talks to Anthropic with your key, from your machine. We do not keep a cloud copy of your work, and we built the product so we never need one.

In the app: sessions run locally, the phone remote stays on your private network.

The human approves what sends, spends or deploys.

Agents can plan, write, test and report on their own. In a coordinated job, before one sends to an audience, starts a paid generation or deploys, it stops and asks you. Autonomy is useful up to the point where a mistake leaves your machine.

In the app: checkpoints in coordinated jobs, amber cells, collision warnings.

Evidence before done.

"Done" means someone ran the thing and saw it work. Every deliverable gets an independent review that returns PASS or FAIL with ranked findings, and the report shows what was run. A confident summary is where checking starts.

In the app: the audit gate and the reports child sessions send back.

Your key, your bill.

You bring your own Anthropic API key and Anthropic bills you for model usage directly. We charge for Command Center itself, so we have no reason to make your agents use more tokens than the job needs.

In the app: spend per session, per day and per account.

How we work.

A short path from a report to a fix, and the same tool we offer to others.

We will ship in small, dated releases
Each release will list what changed and what we fixed, with a date, on the changelog.
The people who build it answer support
A message to support through the contact form is read by someone who works on the code. Bug reports with steps to reproduce go straight to the people who write the code.
We use our own product all day
Our coordinator plans our work and our audit gate reviews this website. When something annoys us, it gets fixed.
We say what is not there yet
Command Center runs on Windows 10 and 11 today. macOS is on the roadmap, with a waitlist on the download page. We will not announce a date until we have a build we trust.
We do not sell data
Our income comes from client projects and professional services. We do not sell or rent personal data and we do not run advertising trackers. The Privacy policy lists everything we do collect.

Company facts.

The basics about the company and its tool, Command Center.

Careers

We have no open roles right now. If you run agents for a living and want to help build this, use the contact form and tell us what you would fix first.

Press kit

Logos, product screenshots and these company facts are available on request: use the contact form and choose Press.

Legal name: PN & CPS Technology Ltd
Address: 20 Swan Street, Manchester, M4 5JW, United Kingdom
SIC code: 62020, Information technology consultancy activities
Trading status: Actively trading
Email: info@pncpstechnology.com
Telephone: +44 7399 886811
Website: www.pncpstechnology.com
Product: Command Center for Windows 10 and 11
Built on: The Claude Agent SDK. PN & CPS Technology Ltd is not affiliated with or endorsed by Anthropic.

Tell us what you want to automate.

Write to us about a project, or request early access to Command Center for Windows and we will email you when you can download it. Solo, Pro and Team will start with a 14-day free trial.