SafePlate Funding map · checked 13 Oct 2026

Where SafePlate can apply, and why we qualify

download as pdf

Every open or upcoming UK grant that fits SafePlate today, what makes us eligible, what could block us, and the dates. Most of these are only reachable if we stop describing ourselves as a restaurant tool and start describing what we are: a safety layer that sits between any kitchen and the people it feeds.

The short answer

  1. Main bet: BridgeAI Deployment Catalyst (closes 8 Dec). We qualify as the lead today. The blocker is the deployer partner: restaurants are not an eligible sector, but a hospital, care group or defence caterer is. Start partner outreach this week.
  2. Rolling, big ticket: EIC Accelerator (EU, up to €2.5m grant). UK companies are eligible for the grant. Our stage (a working product tested in simulated service, ready for its first live kitchen) is eligible, but without a live customer we would be weaker. Best submitted after a first deployment.
  3. Optional research bets (11 and 18 Nov): two Innovate UK AI research calls fit only if we pitch a genuinely new technique, such as Mia's engine-verified safety layer. Long odds; one person, two weeks.
  4. Free money we should claim regardless: R&D tax relief on this year's engineering, and SEIS/EIS advance assurance before the next raise.

Dates

Days left update every time this page is opened.

Opportunities

Apply

BridgeAI Deployment Catalyst: AI for Operational Systems

Innovate UK · £100k–£200k grant per project · 3–5 months · £5m pot · invitation to Phase 2 (up to £1m) and Phase 3 (up to £2m)

Strong fitCloses 8 Dec
Why we qualify
  • UK-registered SME with a working AI/ML product: allergen detection model (99.77% recall), Mia, till and kitchen display.
  • Validated in simulated service (369,262 requests across ten simulated restaurants), which is TRL 5. Not in any live kitchen yet, so it is clearly "not in widespread deployment".
  • This grant would fund exactly the step we are missing: the first real deployment.
  • The strand is "AI that performs, controls or coordinates operational activity". SafePlate physically stops an unsafe order. That is the definition.
The catch
  • The deployer must be in health, financial services, advanced manufacturing, clean energy, defence, professional & business services or creative industries. Restaurants are not eligible.
  • The AI must be central. Lead with the model and Mia doing the work; the rules engine is the safety rail.
  • We can lead only one application across all three BridgeAI strands.
Next step
  • Approach an NHS trust or hospital catering team: allergen errors in patient meals are a patient-safety problem, so it is core to their operations.
  • Ask BridgeAI@iuk.ukri.org to confirm hospital catering counts as "health and life sciences".
  • Attend the 26 Oct collaboration event.
Competition page ↗

BridgeAI Deployment Catalyst: AI for Assurance and Compliance

Innovate UK · same rules and money as above · the alternative strand, not an extra one

PossibleCloses 8 Dec
Why we qualify
  • "AI that monitors, verifies or evidences compliance": Mia's spoken food-safety diary, the compliance PDF, the override audit trail and near-miss log.
The catch
  • Choosing the wrong strand makes the application ineligible, and it cannot be moved. We pick one: Operational (the refusal) or Compliance (the evidence).
Next step
  • Put both framings to Innovate UK in the same email and go with their steer.
Competition page ↗

EIC Accelerator

European Innovation Council · grant up to €2.5m (UK companies get the grant, not the equity part) · short proposal any time, then full-proposal cut-offs

PossibleCut-off ~3 Feb
Why we qualify
  • Built for a working product heading to market (TRL 5–8) with a European-scale story: the 14 EU allergens are one regulation across the continent.
  • The EIC expects traction evidence; a live deployment first (BridgeAI) makes this much stronger.
  • No sector restriction like BridgeAI's.
The catch
  • Competitive: about 13% of full applications reached interview in one recent round.
  • The 2027 cut-off dates are not official yet.
Next step
  • Write the short proposal in November so it is approved in time for the first 2027 cut-off (allow about 10 weeks).
Programme overview ↗
Stretch: research bets

Frontier AI: SME Champions, Phase 1

Innovate UK · £150k–£250k project costs, up to 70% funded · 1–3 months · single SME · route to Phase 2 (up to £1m)

StretchCloses 11 Nov
Why we could qualify
  • Theme 4, "AI safety and assurance": runtime control of AI agents and runtime verification. Mia already forces a voice agent's safety answers through a deterministic engine. Generalising that into a new runtime-control method is in scope.
The catch
  • Must start at TRL 1–3: new research, not our product.
  • Needs a named baseline and numeric targets.
  • The page itself estimates about a 2% success rate.
Next step
  • Only if someone can own a two-week write-up without slowing the main BridgeAI application.
Competition page ↗

Efficient, Structured and Controllable AI Systems

Innovate UK · £50k–£150k project costs, up to 70% funded · 1–3 months · single SME

StretchCloses 18 Nov
Why we could qualify
  • "Causal and intervention-grounded models": a model of how a supplier's recipe change ripples through ingredients, dishes and seats is a causal problem we have real data for (17,067 ingredients).
The catch
  • TRL 1–3 research with ablations and a falsifiable hypothesis.
  • Prompt engineering and agent orchestration are excluded.
Next step
  • Pick at most one of the two research calls.
  • The briefing on 16 Oct will show whether this framing lands.
Competition page ↗
Only with an overseas partner

International bilaterals: Germany, Singapore, Japan

Innovate UK with partner agencies · UK side up to £2.4m / £3m / £2m pots

PossibleFirst closes 1 Dec
Why we could qualify
  • Germany runs on the same EU allergen rules.
  • Singapore is a dense food-service market.
  • Japan has its own mandatory allergen list, a natural test of Passport travelling between countries.
The catch
  • Each needs a separate, unconnected business in that country as a partner.
  • Germany: a German SME applying under its ZIM scheme.
  • Japan: a Japanese SME.
Next step
  • Pursue only if a partner already exists.
  • Japan (20 Jan) leaves the most time.
All competitions ↗
Watch

Knowledge Transfer Partnership, SMART: SCOTLAND, SBRI Healthcare, FSA

Calls that fit but are not open to us right now

Watch
What they are
  • KTP: a university embeds a graduate in our team to work on the allergen model. The current round closes 14 Oct, too soon.
  • SMART: SCOTLAND: a replacement fund is expected late autumn, if we have a Scottish base.
Why not yet
  • SBRI Healthcare: no catering or allergy challenge is open.
  • FSA: its innovation programme ended in May 2026 and it is not funding new projects.
Next step
  • Line up a university partner for the next KTP round.
  • Ask the FSA for a letter of support rather than money.

Claim regardless: R&D tax relief and SEIS/EIS

HMRC · not competitive, no deadline pressure

Do now
Why we qualify
  • The engine, Mia and the allergen model are qualifying R&D.
  • Loss-making, R&D-heavy SMEs can get a cash credit.
The catch
  • Claims must follow the accounting year and current HMRC rules.
  • Have an accountant confirm the scheme and rate.
Next step
  • Brief the accountant this month.
  • Apply for SEIS/EIS advance assurance before the next raise.

Beyond restaurants

The same engine refuses an unsafe meal wherever food is served to someone with a declared allergy. Each row below is a market where SafePlate works as it is today. It is also a deployer type that opens funding a restaurant cannot.

WhereWho serves the foodSafePlate surfacesEligible BridgeAI deployer?
HospitalsNHS trusts, hospital catering teamsEngine on patient meal orders; Passport attached to the patient; kitchen display on the ward kitchenYes, health
Care homesCare groups (medium or large)Resident allergy profiles; Mia for kitchen checks; owner approvalsLikely; confirm health covers social care
DefenceMilitary dining facilities and their contractorsEngine and kitchen display at scale; full audit trailYes, defence
Workplaces and eventsContract caterers and facilities firmsTill plus Passport for staff; estate-wide owner viewPossible, as business services
Food productionReady-meal and dark-kitchen manufacturersIngredient database and recipe-change tracking; label checksPossible, as advanced manufacturing
InsuranceInsurers of hospitality venuesNear-miss and override data as a risk signal; cheaper cover for SafePlate venuesYes, financial services
TravelAirlines, rail, cruise caterersPassport that follows the traveller; engine at the galleyNo; grants via other routes
UniversitiesCampus cateringStudent Passports across every outlet on campusNo; good KTP partner
Everyone elsePOS vendors, menu apps, delivery platformsThe engine and ingredient data as an APINo; this is the platform business

Rules that decide what we can do at the same time

One BridgeAI lead

SafePlate can lead one BridgeAI application, across all three strands. A deployer can join up to three, each for a different use case.

One application per research call

Frontier AI and Controllable AI each allow one application per SME. They are separate competitions, but writing both at once is not realistic.

No subcontractors

All three Innovate UK calls ban subcontractors. Work has to be done by consortium members or our own team.

Start dates line up

All three would start by 1 April 2027. If more than one is won, the team has to deliver them in parallel.

Sources