Why SMEs Should Be Central to Government AI Plans
AI success in government depends on who you work with as much as what you buy. SMEs bring speed, sharp expertise, and closer collaboration - if we make it easier for them to engage. That means rethinking procurement, building trust, and strengthening in-house capability.
The SME entry problem is fixable
Smaller, homegrown firms face high barriers to entry because of risk-averse procurement. As Chad Bond, director of strategy and innovation at Zaizi, put it: "We can work faster, with deeper expertise. The challenge is how to create an entry bridge so that government can actually take advantage of that."
That bridge looks like shorter routes to pilot, outcome-focused contracts, and space to prove tools in days and weeks - not months of paperwork.
Trust is the non-negotiable
Trust runs through every AI decision - trust in the model, the data, and the partner. "Trust is non-negotiable in national security," said Bond, calling for explainability and transparency in deployments. Clear audit trails, model cards, and open testing standards lower risk for both buyer and supplier.
What Ukraine taught us about pace
Ukraine showed what's possible when procurement enables speed. Emergency processes let teams make decisions at pace and deliver capability when it mattered. The UK can move faster too.
New powers under the Procurement Act allow competitive, flexible procedures and dynamic markets, shifting away from document-heavy processes to hands-on evaluation. See official guidance on The Procurement Act.
Procurement 2.0: prove it, don't pitch it
Departments can run short, dynamic competitions built around live demos, sandbox access, and prototype testing. Let providers show what they've built, how it explains decisions, and how it integrates with your data and controls.
This mindset pairs well with an open-source ethos: share code, patterns, and tools to increase transparency and reuse across government. The Service Manual guidance on using open source software is a good anchor for this approach.
Frontline reality: the NCA's pivot to SMEs
Paul Edmonds, CDO at the National Crime Agency, made the case plain. Early reliance on major consultancies "was an utter disaster. They had less knowledge of AI than we did, and they were implementing outmoded models based on generic consultancy approaches."
His takeaway: build a supply chain with more SMEs. "Our more traditional supplier base couldn't do that kind of agile, collaborative working. And the specialist knowledge in the SME marketplace is far deeper than in our existing supplier space."
Become a smarter buyer
The shift is from transactional compliance to intelligent collaboration. Shared goals, shared language, and shared delivery rhythms - not just shared contracts.
- Run outcome-based competitions and pay on milestones.
- Use small, timeboxed discovery and alpha sprints with SMEs to prove value early.
- Mandate explainability, audit, and secure-by-design as acceptance criteria.
- Default to open standards and publish reusable patterns where possible.
- Adopt lightweight contracts that enable iteration instead of locking into fixed specs.
Close the skills gap across three levels
- AI literacy for all: Everyday use of AI tools, prompt quality, data hygiene, and knowing when to escalate risks.
- Leadership capability: Clear outcomes, risk appetite, procurement choices, and model accountability.
- Deep technical expertise: Teams who can evaluate models, build prototypes, and adapt solutions in-house.
If your team needs a fast lift in AI literacy or role-based upskilling, explore focused options here: AI courses by job.
Practical next steps for departments
- Stand up a cross-functional unit (policy, data, security, commercial) to own AI pilots from day one.
- Publish a short "problem statement" and invite SMEs to demo working prototypes against real but safe data.
- Use dynamic purchasing or flexible procedures to contract in weeks, not quarters.
- Require model cards, data lineage, and human-in-the-loop controls for high-risk use cases.
- Open-source non-sensitive components to speed reuse across government and markets.
- Measure value with clear operational metrics: time saved, risk reduced, cases progressed.
The takeaway
If the UK wants meaningful AI outcomes in public service, SMEs can't sit at the fringes. Treat them as partners, remove needless friction, and strengthen your internal capability. That combination delivers pace, trust, and real results for citizens.
Your membership also unlocks:
