Build
I design and build the platforms where teams create AI agents — assistants that answer questions, search internal knowledge, and take real action in the systems organisations already run on: SharePoint, Salesforce, ServiceNow and more.
Applied AI · platforms & governance — Manchester, UK
Enterprise AI — from pilot to production.
I take enterprise AI out of pilot purgatory — from proof of concept to production that real people use and a board can trust. 300+ AI agents live, rollouts reaching 100,000+ people, and the security testing and oversight that keep every one of them safe.
I work with CTOs, CISOs and Heads of Data at universities, banks and enterprises.
Organisations & brands I’ve worked with
PMB Finance The gap
The demo works.
Production doesn’t.
At least 30% of enterprise AI projects are abandoned after the proof of concept — stranded in pilot purgatory, never reaching the people they were built for.Gartner, 2024
I make the jump the others can’t. I’ve taken AI from proof of concept to production again and again — across two major universities and tier-one banking — with the governance and security that let regulated organisations actually put it live. Safe and trusted first, used by real people second.
300+ AI agents in production on platforms I deployed
100k+ students & staff relying on AI I’ve shipped
~220 attacks every agent survives before it meets a user
4 AI clouds governed from a single place
How I get you there
Build it. Make it safe.
Get it used.
Govern
Before an AI agent reaches real people, I make sure it can’t leak data, be manipulated or go off-script — automated security testing, data protection, and live monitoring once it’s running. Governance I learned in tier-one banking.
Scale
Technology only counts if people use it. I’ve led AI rollouts to 65,000 and 36,000 staff and students — with the training, support and guardrails that let an organisation adopt AI with confidence, on cloud foundations I built myself.
Selected work
Platforms with real users,
governance with real teeth.
One platform, 300 AI agents
The problem: hundreds of staff wanted to build their own AI agents, with no safe, central way to do it. What changed: I built a single platform where every agent is created, governed and managed in one place — and the integration layer that lets those agents safely read from and act on the systems the organisation already runs on. Today it runs 300+ agents in daily use.
AI for 100,000 students and staff
The problem: two major UK universities needed AI that was safe to put in front of everyone — not a locked-down pilot a handful of people ever touch. What changed: I took early-stage platforms to production — single sign-on, audit trails, usage controls, guardrails and a secure sandbox where agents can run code without touching anything they shouldn’t. Together they now serve 100,000 students and staff — drafting job descriptions, supporting research and producing content every day.
Attacking our own AI — before anyone else can
The problem: an AI agent can leak data, be manipulated or go off-script — and you won’t know until a real user finds out. What changed: I built automated security testing that fires ~220 attacks at every agent before it ships, paired with live monitoring that raises the alarm the moment a deployed agent steps out of line. Problems get caught before people ever meet them.
Also: data-protection and personal-data controls · self-service research environments on demand · Microsoft 365 Copilot adoption for 65,000 people · automated compliance across 280 cloud accounts at a tier-one bank.
Contact
Let’s get your AI
out of the pilot.
Most engagements start with a Production Readiness Review — a focused look at where your AI is stuck and what it takes to get it safely into production. A 30-minute call, no pitch deck.
ahmed@sidekraft.co.ukBuilt and governed enterprise AI at the University of Manchester, the University of Exeter and NatWest.