CPAT Group Discuss your challenge

How we work

Three ways to work with us, often together: experienced people for any IT role, AI applications and agents built for your team, and training so your own people can build them too.

How we staff IT and technology roles

Back to top

Experienced people on your team, full time or fractional – and the option to automate their work.

  1. 1

    Tell us the role

    The skills, experience, start date and how long you need them. One call is usually enough.

  2. 2

    We find the right person

    Experienced people who’ve done the work, many cross-trained to cover more than one role.

  3. 3

    You interview and decide

    Every candidate passes our rigorous interview process first. Then they interview with your team and their manager. Not the right fit? We keep looking.

  4. 4

    They join your team

    Full time or fractional, with a 6‑month minimum. We check in regularly, keep them trained on AI and can build apps and agents to automate their work.

  5. 5 · Optional

    Hire them on

    Want them as your own employee? We’ll help with the transition – no conversion fee. Entirely your call.

How we build AI applications and agents

Back to top

Fast because it’s iterative – not because we skip steps.

  1. 1

    We meet with you

    We learn your top 10–15 needs, problems or ideas, and who’s involved. Happy to sign an NDA first.

  2. 2

    Requirements and your environment

    We gather requirements with your users and learn your environment – architecture, security, data and access. About 8 working days.

  3. 3

    A free first version

    A working prototype in days, built to your standards in your environment. Not where you want it? You’ve spent nothing.

  4. 4

    Iterate with your users

    Weekly rounds: demo, feedback, refine. Each round sharpens the requirements and adds data connections and polish.

  5. 5

    Production-ready, then yours

    Security review, testing, documentation and training, then rollout with a review cadence. Typically 1–5 weeks.

Discoveryrequirements, systems, data, security
Every week, with your users
BuildDemoFeedbackRefine
Productiontested, approved, your team trained

What sets the pace

Getting to a working prototype is fast. The refinement sets the rest of the timeline, and three things matter most:

  • Requirements. We start from a shared idea and sharpen it every round.
  • Data readiness. The right data sources, in usable shape.
  • Access. Things move fast once your IT team grants permissions.

Everything we build follows your enterprise architecture and security standards, and nothing reaches production without your approvals.

What we’ll ask about each idea

No RFP and no long proposal – just a short conversation.

  • The problem or opportunity
  • Who’s involved and who decides
  • Where the requirements stand
  • The systems and data sources it touches, and how ready the data is
  • Who controls access and permissions
  • What success looks like
  • Your initial thoughts on a solution

What it runs on – and what it costs to run

Every solution runs in your environment, on providers you choose. After hand-off you pay them directly, at their prices – no markup from us. See what we charge to build it

Where it runs

In your own cloud – Azure, AWS or Google Cloud – behind your company sign-in (Entra ID, Okta or Google). Your data stays in your environment.

The AI behind the agents

Claude from Anthropic by default, through your cloud’s AI service (Azure AI Foundry, Amazon Bedrock or Google Vertex AI) or Anthropic directly – or a model you’ve already approved, like Azure OpenAI or Gemini.

Scheduled jobs

Agents that work on their own – checking a mailbox every 15 minutes, a nightly data refresh – run as scheduled functions (Azure Functions, AWS Lambda or Google Cloud Run), logged and monitored.

Ongoing costWho you payTypical range per solution
AI model usage – billed per use, so it grows with volumeYour cloud provider or AnthropicAbout $10–$200 a month
Hosting, storage and scheduled jobsYour cloud providerAbout $20–$150 a month
Sign-in, email, Teams and tools you already licenseYour existing licensesUsually $0 extra
Support after hand-off – optionalCPAT, or your own teamUp to $1,250 a week, or $0 if your team runs it

Ranges are estimates for a typical team tool. Before you commit, we give you a run-cost estimate for your volume and set up a budget alert in your cloud account.

Citizen developer training: From Idea to Live

Back to top

Hands-on training that turns your people into citizen developers – building real apps and AI agents, safely, inside your own environment. No coding experience required.

What your people learn

  1. Identify and prioritize – spot work worth automating, pick the right kind of solution, score ideas for a quick win and weigh them against your team’s capacity
  2. Design and build with AI – blueprint a tool in five minutes, then build it in short rounds with an AI assistant
  3. Publish it properly – source control, your cloud, storage, secure AI and company sign-in, step by step
  4. AI agents – features, agents and agentic workflows, built where people approve every change
  5. Test and maintain – test with one real user, then ship updates safely

Pick a format

  • Leadership briefing60–90 minutes · the method, the governance model and live examples
  • Two-session workshop $1,500 flat2 × 90 minutes · the full curriculum with a live build · up to 20 attendees
  • Hands-on bootcamp1 day · everyone leaves with a first working version of their own tool

On-site or virtual. See pricing. Tailored to your cloud, sign-in, AI tools and intake process.

Governance built in

  • Company accounts and a sandbox – never personal ones
  • Access earned through training, requested through a ticket
  • Read-only connections to systems of record
  • Every tool registered in one place, with a clear hand-off to IT when it outgrows citizen development

Includes the deck, a step-by-step guide, a prompt library and an opportunity-scoring worksheet.

Send us your list.

Contact us