aiTudu Book a call

● AI agent & Power Platform architect · Warsaw · remote

AI agents that stay governed, stay in budget and reach production.

I'm Abdulrahman Ibrahim Tudu. I help organisations running Microsoft 365 put AI agents into production, whether they're built in Copilot Studio, Claude or ChatGPT, so they pass security review, don't surprise finance and actually get used.

Book a 30-minute call See the work
Works with → Copilot StudioClaude & Claude CodeChatGPT & CodexMCPPower AutomatePower AppsDataversePower BIAzure AIPAC CLI

Services

Three things I do well.

01

AI agent governance & rollout

Copilot Studio, Claude or ChatGPT: decide who gets access, what data goes where and how agents reach production, before the first incident.

  • Governance policy and approval flow
  • Access, SSO groups and data rules per tool
  • Environment strategy, DLP and Copilot Author groups
  • Claude Code and Codex rollouts for dev teams
02

AI cost control

Credits, seats and API keys are open-ended by default. I set the limits that actually stop spend, and show who used what.

  • Copilot Studio billing plans and per-agent limits
  • Seat licences and API spend caps across vendors
  • Budget alerts that reach the right people
  • Chargeback per business unit in Power BI
03

AI agent & automation builds

Production agents and flows, delivered like software, with tests, runbooks and a clean handover.

  • Copilot Studio and Claude-based agents
  • Connected to Dataverse and Microsoft 365 via connectors and MCP
  • Document AI: PDFs in, structured data out
  • Managed solutions, UAT guides and runbooks

Start with a fixed-price "which AI tool for which job" assessment: Copilot, Claude and ChatGPT compared against your use cases, licences and data rules. Then: build project · advisory retainer · workshop

Selected work

Real projects, described without client names.

Finance · AI automationPower Automate · Dataverse · Azure AI

Vendor invoices: from shared mailbox to SAP, without retyping

Problem
Supplier invoices arrived as PDFs in a shared mailbox and had to reach the SAP invoice-management process as structured e-invoices.
Built
A queued pipeline: each PDF becomes a tracked record, Azure AI extracts the fields, the flow generates and validates PEPPOL UBL XML, then hands it to the SAP/VIM mailbox.
Delivered
A versioned managed solution, a UAT test guide and an import runbook that matches the running system.
Multi-country · governanceCopilot Studio · PPAC · DLP

A Copilot Studio governance pack people actually read

Problem
Business units wanted to build agents; central and country admins needed clear rules for who builds, who approves and who owns what.
Built
An authoritative governance guide plus short role guides for makers, admins and business owners, and a recommendation for publishing them to each audience.
Delivered
A client-ready pack and workshop material for the rollout.
Central IT · FinOpsPPAC · Azure Cost Mgmt · Power BI

Pay-as-you-go spend limits and chargeback per business unit

Problem
A central admin team bills each business unit for its Power Platform and Copilot Studio usage, and needed limits that really stop spend.
Built
A four-stage control model (Azure, billing plan, environment overage, per-agent limits) and a Power BI chargeback model fed by Cost Management exports and the Power Platform licensing API.
Delivered
An illustrated explainer, and the report generated as code, deployed to the Power BI service and refreshed in a test tenant.
My own systems

I run LLM agents every day

Several agents on Claude and OpenAI models handle my own work. One administers Power Platform through the PAC CLI, Dataverse and MCP servers, under strict identity and audit rules. I use what I recommend.

Also built

Akadaa

A school-management SaaS for Nigerian schools that I founded and built: admissions, results, fees, parent access and CBT.

How I work

Small, clear steps. Nothing hidden in a slide.

  1. Assess

    Read the tenant, the licences and the bill. Write down what's true today.

  2. Design

    One document: the decision, the options I rejected and why, the cost.

  3. Build

    In your environments, as managed solutions, tested against real cases.

  4. Hand over

    Runbooks, UAT guide and a walkthrough, so your team owns it.

Notes & videos

What I learn on projects, written down.

Note · Oct 2026

Copilot Studio on pay-as-you-go: where you can actually cap spend

Azure budgets only alert. The hard stops live in the admin center. A stage-by-stage map.
YouTube · coming soon

@aitudu

Screen-recorded walkthroughs from a demo tenant: cost control, governance, real builds, and the same agent built in Copilot Studio and in Claude.

Next note

Chargeback for Power Platform in Power BI

Cost per business unit from Cost Management exports and the licensing API.

About

Hi, I'm Abdul.

The "AI" in AI Tudu is my initials: Abdulrahman Ibrahim Tudu. It just happens to be what I work on, too.

I'm a Power Platform solution architect and administrator, and I administer and use Claude, ChatGPT, Claude Code and Codex every day. I grew up in Gusau, in northern Nigeria, trained as a scientist (MSc Applied Biotechnology) and have lived in Warsaw since 2020, working with several clients at a time.

Outside client work I build things I care about: Akadaa, a school-management platform for Nigerian schools, and the Northspire Education Foundation, which supports students in northern Nigeria. Both run on the same habits I bring to clients: templates, automation, and nothing left undocumented.

  • Based inWarsaw, Poland (CET)
  • WorksRemote, in English
  • FocusCopilot Studio, Claude, Power Platform, Azure
  • BackgroundMSc Applied Biotechnology

Contact

Rolling out agents or automation this quarter?

Tell me what you're building and what worries you about it. I'll reply within two working days with a few questions, or a short call if that's quicker.

Email to book a call