BridgeAI Deployment Catalyst 2026: what it funds and how to apply

The BridgeAI Deployment Catalyst (AI for Operational Systems) gives a UK AI SME and an organisation that will use its system £100k to £200k in grant to prepare that system for live operations. It is a £5m Innovate UK competition, open now and closing at 11am on 8 December 2026.

It is the first phase of a programme that could run to three phases, so the work you do now also sets up a route to larger funding later.

What is the BridgeAI Deployment Catalyst?

It is an Innovate UK competition for AI systems that are built but not yet running for real. The Innovation Funding Service page describes the Operational Systems strand as funding AI whose main purpose is to perform, control or coordinate operational activity in workflows, processes, software, equipment or infrastructure.

There are two sister strands: AI for Decision Critical Systems, for AI that supports consequential judgements, and AI for Assurance and Compliance Systems, for AI that monitors, tests or evidences compliance. You must apply to the strand that matches your system's primary purpose. An application in the wrong strand is ineligible and cannot be moved.

The whole programme is worth up to £52m across three phases. Phase 1, this competition, has up to £5m.

	Phase 1 (this competition)
Grant per project	£100,000 to £200,000 in total
Project length	3 to 5 months
Opens	21 October 2026
Closes	8 December 2026, 11am UK time
Outcome	1 February 2027
Start and end	Start by 1 April 2027, end by 31 August 2027

Phase 2 is indicated at up to £1m over about six months, and Phase 3 at up to £2m over about 22 months, both with co-investment expected. These figures are indicative, and each phase has a stage-gate review.

Who can apply?

Projects must be collaborative, with two or three organisations.

The lead must be a UK registered SME with an existing AI or machine learning solution at technology readiness level 5 or 6. It needs a working minimum viable product. It may have been piloted, but it must not already be commercially deployed for the proposed use case. The lead must carry 40% to 70% of total eligible costs, and can lead only one application across the three strands, and cannot be a partner on any other.

The deployer must be a UK registered medium or large business, or a UK public sector organisation. It must provide a real operational setting, with access to users, data, systems and workflows, and carry 30% to 50% of total eligible costs.

An optional partner can be any UK business, university, research organisation, charity or public body, up to 30% of total eligible costs.

Subcontractors are not allowed.

Which sectors does it cover?

The deployer must operate in one of seven sectors:

• Health and life sciences
• Financial services
• Advanced manufacturing
• Clean energy industries
• Defence
• Professional and business services
• Creative industries

Transport, logistics, construction and agriculture are not on the list. If your deployer sits in one of those sectors, this competition will not fit unless its business also falls within one of the seven.

How much of the cost does the grant cover?

The competition funds feasibility studies, with rates set by the size of each organisation:

• up to 70% for micro and small businesses
• up to 60% for medium businesses
• up to 50% for large businesses

Research organisations doing non-economic work can take part at higher rates, up to 30% of total eligible costs between them. Universities claim 80% of full economic cost.

What kind of project fits?

The scope asks you to show that AI is central to the deployment, that the use is innovative, and that there are real technical, operational, regulatory or assurance challenges to solve. Routine configuration or integration will not count as novel.

Illustrative themes include agentic and multi-agent AI, multimodal AI, robotics, autonomous systems, computer vision, reinforcement learning, and AI-enabled optimisation and control. Digital twins that use AI for operational decisions can fit too, and our guide to funding a digital twin project covers the other routes.

Out of scope:

• early-stage research or concept work
• generic productivity tools, back-office automation or routine process improvement
• AI that is peripheral to the deployer's core operations

Phase 1 must end with a demonstration using representative deployer data and systems. During Phase 1 you must also secure letters of support from potential deployers outside the consortium who could join later phases.

What do assessors look for?

Three independent assessors score ten questions, each out of 10, with answers of up to 400 words. The assessor guidance sets out what a strong answer contains:

• Market size: total addressable and serviceable obtainable market, justified with data.
• Real-world benefit: quantified benefits, including an annual savings estimate for a single site.
• Solution readiness: what has been built, how it works, clear IP ownership and a legible architecture diagram (a PDF of up to two A4 pages).
• Route to deployment: the regulatory, assurance and standards requirements before go-live.
• Skills: a team with relevant track record and clear roles.
• Deployer readiness: a detailed description of the people, data, systems and workflows available.
• Learning: what evidence you will collect and how it informs future deployment.
• Risks: the top three risks to wider deployment, each with a tailored mitigation.
• Milestones and delivery: feasible milestones tied to what the grant pays for, with time and cost control.

Separately, the scope answer is checked before assessment. If it is out of scope or not detailed enough, the application is not sent to assessors. Innovate UK may also balance its portfolio, so a high score does not guarantee funding.

Where do applications fall down?

The first four come from the assessor guidance and the last from the competition page. The one in between is general advice.

• No single-site savings figure. The guidance asks for one. Work it out with your deployer, using their numbers.
• IP claimed without evidence. Show who owns what and on what terms.
• A missing or unreadable diagram. It is a scored part of the readiness question.
• Generic risks. A bare risk such as data quality, with no mitigation, scores poorly. Name the risk for this deployer and say what you will do.
• A thin deployer. If the deployer's role reads as a letter of support, the deployer readiness answer will be weak.
• Leaving it late. Scope queries need to reach BridgeAI@iuk.ukri.org at least 10 working days before the deadline.

How does a design and engineering partner fit?

Because subcontracting is not allowed, an outside design or engineering firm can only be in the project as the one optional partner, within the 30% cost limit. That is a real constraint, so decide early whether you need it.

Where it helps is the work that decides whether a deployment sticks: the interface people use to trust and act on the AI's output, the integration with the deployer's systems, and the data pipelines that feed the demonstration. It also helps before you apply, when you need a clear technical architecture and a scoped plan.

TEO builds AI tooling and data platforms, from document ingestion and assistants to data pipelines, dashboards, evaluation and guardrails. We start with the workflow and design the interface so people trust what the system tells them.

Next steps

1. Check your deployer's sector against the seven on the list.
2. Pick the strand that matches your system's primary purpose.
3. Watch the briefing recording and slides on the competition page (the briefing was held on 27 October 2026).
4. Agree the single-site savings case and data access with your deployer now.

If BridgeAI does not fit, see what Innovate UK funds now that Smart grants are paused. If you want help scoping the build or shaping the technical side of an application, talk to us.