“How do we pay for this?” is the question that stalls most AI projects. The good news: the answer in 2026 is more affordable than the “big software project” mindset assumes. Here’s a practical funding playbook for UK businesses, cited to official sources. (dgm implements osFoundry as an independent partner; we are not a grant-writing service.)

Start from reality: self-fund, then offset

The honest framing: most UK businesses fund AI adoption themselves and use public support to offset the cost — because, as our grants guide explains, there’s no UK scheme that simply buys you AI software (the one that discounted software, Help to Grow: Digital, closed in February 2023).

That sounds discouraging until you realise the up-front cost has fallen.

Why the funding maths changed

Older AI/software projects meant large up-front licences — exactly what grants were designed to offset. Modern platforms work differently. osFoundry, for example, is usage-priced with no per-seat fees and supports bring-your-own-key (you pay model providers directly), so a pilot can start small. The big costs today are usually scoping, data preparation and integration — work, not licences.

The funding stack: layer your sources

Think of AI funding as a stack you assemble, not a single grant:

  1. Self-funded pilot. Start with a small, well-scoped project you can fund from operating budget. Prove ROI.
  2. R&D tax relief. If the work involves genuine technological development, claim under the merged scheme (20% credit; ERIS ≈ £27 per £100 for loss-making R&D-intensive SMEs). Relief comes after the spend.
  3. Adoption grants (if eligible). SME manufacturers: Made Smarter (up to £20,000 at 50%). Northern Ireland: Invest NI DTFF (£5,000–£20,000 at up to 70%).
  4. Finance. Early-stage businesses: Start Up Loans (£500–£25,000). Growing SMEs: the Growth Guarantee Scheme (up to £2m, government-guaranteed to the lender, open to 31 March 2030).
  5. Free advice. Growth Hubs, Innovate UK Business Growth and BridgeAI de-risk the project at no cost.

Prove it small, then scale

Funding bodies, lenders and your own board all respond to evidence. A pilot that demonstrably saved hours or cut a cost is a far stronger basis for an R&D claim, a grant application or a finance request than a plan on paper. So the sequence matters: scope tightly, fund a pilot you can afford, measure, then scale with whatever support you’ve become eligible for.

Where dgm fits

dgm scopes and builds the AI project — defining the use case, preparing data, and implementing on osFoundry, with the option to self-host in your own AWS/GCP/Azure or use osFoundry’s EU region for UK data-sensitive work (it publishes US/EU/JP regions, not a UK one). A clean, measured pilot is what later makes any funding conversation credible.

dgm is an independent integration partner, not a grant-writing service, with zero integrations so far. For tax claims and applications, use your accountant or a grants specialist. To scope a fundable, affordable AI pilot, book a consultation with dgm. General information, not financial advice.