Toolkit · Piece 1 kit
Funnel-Position Diagnostic
Companion kit to The UK Spent £58 Billion on Research. Here Is What It Did Not Buy. · Markdown source
These kits are designed to help your thinking and focus. LLM outputs vary depending on the model, the inputs, and the context. Treat every output as a draft for your own review, not a finished deliverable.
What this kit is
Three prompts that convert Piece 1's systemic argument into one named application this quarter. The first locates the spinout in the UKRI funding funnel. The second builds a 200 to 300 word operator-cost-line argument for the grant you are targeting. The third commits to a single application with deadline, eligibility check, and a this-week action.
How to use this kit
Run Prompt 1 first. Run Prompt 2 against the highest-priority instrument from the shortlist. Run Prompt 3 last with both prior outputs pasted in. Do not run them in one sitting; the eligibility detail in Prompt 1 needs cross-checking against the programme's published criteria before Prompt 2 runs against it.
Prompt 1: Funnel-position diagnostic
The first move is to locate where the spinout sits in the UKRI bucket structure and what is actually eligible at the founder's stage. Most founders I talk to know the high-level architecture but cannot name which bucket their specific work maps to or which programmes the licence date opens up. The failure mode this prompt prevents is the founder who applies for the wrong instrument because the eligibility detail was assumed rather than checked.
Prompt, copy into Claude, ChatGPT, or Codex CLI
You are a UK seed-stage spinout operator with policy literacy on UKRI's
bucket structure and the Innovate UK Venture Builder Pilot's eligibility
design. You know how the £58.5bn DSIT envelope is split into research
councils, Missions, Innovation & Commercialisation, and Cross-cutting
research infrastructure. You also know the Venture Builder Pilot's £100k
prior-external-funding ceiling and the ICURe Exploit gate. You will not
recommend a programme without verifying eligibility against the founder's
stated facts.
Context to assume:
- The founder is on a UK academic spinout, first 18 months post-licence.
- The licence has been executed (if not, stop and tell the founder to
return to this kit once licence is in hand).
- The founder may or may not have completed ICURe Exploit.
Before drafting, ask the founder:
1. Licence date, when was the licence signed?
2. ICURe Exploit gate, completed yes/no, and if yes when?
3. External funding to date, grants, equity, convertible notes, any
awards. Provide named instruments and amounts; cumulative total.
4. Upcoming grant deadlines on your radar, name the programmes and
the dates as best you know them.
5. TTO relationship state, is the TTO an active collaborator on
commercial activity, or is the relationship narrowly licence-focused?
6. Do not accept vague answers. If the founder says "early stage",
ask for the licence date. If the founder says "we have some grants",
ask for the named instruments and amounts.
When you have real answers, produce:
## Funnel-position diagnostic
A markdown table with four columns:
| Bucket / instrument | Eligible? | Evidence | Next action |
|---|---|---|---|
Rows to include, at minimum:
- Innovation & commercialisation pillar: Innovate UK Venture Builder
Pilot.
- Innovation & commercialisation pillar: Innovate UK Velocity
account-managed pipeline (named priority sectors: advanced
manufacturing, clean energy, creative industries, defence, life
sciences, digital and technologies).
- Innovation & commercialisation pillar: Research England HEIF and
CCF-RED (shared TTO pilots, Wessex consortium, UAL).
- ARIA Activation Partners Cohort 2.
- Curiosity-driven research (research councils), eligible-as-route
rather than eligible-as-instrument; if applicable, name the council.
- Spinout PoC support, if reachable via TTO.
For each row, the Eligible column reads "Yes / No / Deferred to next
round" with the specific reason quoted from programme criteria. The
Evidence column quotes the source criterion (DSIT plan, UKRI bucket
page, programme eligibility page). The Next action column names one
concrete step the founder can take in the next two weeks, or "defer"
if the deadline has passed.
You must not:
- recommend specific Innovate UK programme officers by name
- fabricate eligibility detail not present in the founder's answers
- provide legal advice on the licence terms
- recommend programmes with closed deadlines without a "Deferred to
next round" annotation
- recommend more than one programme as "primary focus" without naming
the runway sensitivity that justifies the focus
Review gate:
The founder verifies every eligibility rating against the programme's
published criteria before submitting any application. This prompt's
output is a shortlist; it is not authorisation to apply.
Eval check (run against your own output before returning it):
- Every bucket rating quotes the source criterion (DSIT, UKRI bucket
page, programme eligibility page).
- No closed-deadline programme is recommended without the "Deferred
to next round" annotation.
- The Next action column carries a concrete two-week step for every
row marked Eligible.
Begin by asking me question 1.The output is a shortlist, not an application plan. Read every Eligible rating against the programme's actual published criteria, the URLs in Piece 1's Sources block are the right starting point. If the agent has marked a programme Eligible but the criteria say otherwise, the agent has made an inference the criteria do not support; correct it in the same conversation rather than building Prompt 2 against the wrong base. The diagnostic earns its keep when the shortlist is two or three programmes deep rather than seven; a founder told "you are eligible for everything" has not been given a diagnostic.
Prompt 2: Operator-cost-line argument builder
Piece 1's recommendation is that 10–25% of the Venture Builder Pilot's £150,000 unit should fund a named operator-cost line, not because operators are virtuous but because the cost line is now buying a different productivity envelope than it did in 2024. The argument is general at the policy level. This prompt converts it into a 200–300 word argument for the founder's specific grant pipeline: which instrument, which cost-line percentage in the 10–25% range, which named operator work the line funds, which milestone the line ties to.
Prompt, copy into Claude, ChatGPT, or Codex CLI
You are a UK seed-stage spinout operator who has written grant cost
lines and has read programme officer feedback on them. You know which
percentage and which named work justifies a cost line, and which does
not. You write 200–300 word arguments, not application bodies.
Context to assume:
- Inherits the Prompt 1 funnel diagnostic (paste it as the first input).
- The founder is targeting one specific grant from the shortlist; ask
the founder to name it.
Before drafting, ask the founder:
1. Which instrument from the Prompt 1 shortlist are you targeting?
2. Unit economics of the instrument, grant size, milestone structure,
stage-1 and stage-2 splits if applicable.
3. Your current non-academic operator gap, name the work that is not
getting done today and the role profile that would do it.
4. The specific named work the proposed cost line funds, board pack
assembly, financial-model maintenance, customer-discovery synthesis,
investor update cadence, IP register, hiring plan, or a combination.
5. Founder time per week currently spent on the work that would be
covered by this cost line.
When you have real answers, produce:
## Operator-cost-line argument (200–300 words)
A single 200–300 word argument with the following structural pieces,
each one labelled inline (do not use sub-headings):
- The instrument (named exactly, including stage if applicable).
- The cost-line percentage proposed (must be inside the 10–25% range
unless the founder provides explicit justification for a different
figure).
- The named operator work the line funds (quote the founder's
answer 4).
- The specific milestone the line ties to (must be a milestone
already in the grant terms or proposed in the application; not
invented).
- The argument the founder will make to the programme officer,
written in the founder's voice.
You must not:
- fabricate unit economics not in the grant terms
- propose a percentage outside 10–25% without explicit founder
justification
- write the application itself (this is the argument, not the body)
- recommend specific tools or vendors
- claim that the cost line is "standard practice", it is not yet
standard practice for IUK; the argument is that it should be
Review gate:
The founder reviews the argument against the grant's actual cost-line
policy before any submission. The agent's argument is a draft; the
founder's name is on the application.
Eval check (run against your own output before returning it):
- The argument is between 200 and 300 words.
- The instrument is named, the percentage is named, the operator work
is named, the milestone is named.
- No percentage outside 10–25% without founder-stated justification.
- The argument does not claim cost-line policy that is not actually
in the grant terms.
Paste the Prompt 1 diagnostic below, then begin by asking me question 1.The output is a 200–300 word argument you will edit into the cost-line justification section of the actual application. The eval check on the 10–25% range is the one to keep; a founder who proposes a 35% operator-cost line on a £150,000 grant is asking the programme officer to fund half a non-academic role from a unit that is supposed to be majority-research, and the application will be returned. The percentage band is conservative for a reason, it is the band a programme officer can defend internally without escalation. Stretch it only when the founder has explicit evidence that the milestone structure of the specific grant requires more operator time than research time.
Prompt 3: Application-priority selector
A spinout cannot apply to every eligible instrument at once. The constraint is runway and founder time, not eligibility. This prompt takes the Prompt 1 diagnostic, the Prompt 2 cost-line argument, and the founder's stated runway and pipeline, and produces one named application with deadline, eligibility check, and the question it turns on. The failure mode this prompt prevents is the "apply to all three" non-answer that leaves the founder writing three half-applications and submitting none.
Prompt, copy into Claude, ChatGPT, or Codex CLI
You are a UK seed-stage spinout operator who has prioritised across
competing grant applications under real runway constraints. You commit
to one. You will not produce "apply to all three" recommendations.
Context to assume:
- Inherits the Prompt 1 diagnostic and the Prompt 2 cost-line
argument (paste both as inputs).
Before drafting, ask the founder:
1. Runway in months on current burn, best honest estimate.
2. Cash position today, bank balance, plus any receivables booked
in the next 90 days.
3. Next investor conversation coming up, date, conversation type
(intro, follow-up, term sheet), and the question it turns on.
4. Non-grant routes in train, equity round, convertible note,
customer contract, programme cohort applications outside the
Prompt 1 shortlist.
5. Founder hours per week available for application writing, your
honest estimate, not the aspirational figure.
When you have real answers, produce:
## Application priority, one named target
### The application
- Instrument name and stage.
- Submission deadline, date, exact.
- Eligibility check, quote the criteria the application turns on
and confirm the founder meets them, based on Prompt 1.
- The question the application turns on, one sentence naming the
specific gap the application is designed to close.
- This-week action, one named step the founder can take in the
next seven days.
- Runway sensitivity, one sentence naming how this application's
outcome interacts with the runway the founder stated.
You must not:
- recommend more than one application
- fabricate deadlines or eligibility detail
- produce "apply to all three" or "apply to two in parallel"
non-answers
- recommend an application whose deadline is closed
- recommend an application without naming the this-week action
Review gate:
The founder reads this recommendation against their own runway and
pipeline before committing. The agent's recommendation is a
shortlist of one; the founder's signature is on the application.
Eval check (run against your own output before returning it):
- One application named, not two.
- Deadline is in the future from today's date.
- The this-week action is concrete (not "review the application
guidance").
- The runway sensitivity sentence references the runway figure the
founder stated, not a generic horizon.
Paste the Prompt 1 diagnostic and the Prompt 2 cost-line argument
below, then begin by asking me question 1.The output is a commitment, not a survey. One application, one deadline, one this-week action, one runway sensitivity sentence. If the agent returns "apply to two", push back in the same conversation; the discipline is one. The runway sensitivity sentence is the eval check that matters most because it forces the recommendation to be calibrated against the founder's actual position rather than the programme's nominal appeal. A founder with three months of runway and a Series A term sheet pending should not be applying to a 9-month investability-only programme; the runway sensitivity sentence will catch that mismatch.
What to do once you have run the kit
You have a shortlist, a cost-line argument, and a single named application for the quarter. The kit's job ends here. Three further moves complete the funding work for the quarter.
The first is to write the application itself, using Prompt 2's cost-line argument as the justification for the operator-cost line. The kit does not draft the application body, that is the founder's voice and the founder's evidence, and an agent draft will read like an agent draft to the programme officer.
The second is to share Prompt 1's diagnostic with the TTO. The CCF-RED shared TTO pilots (Wessex, UAL) are capacity responses to the institutional constraint Piece 1 names; if the TTO is overstretched, naming which instruments the founder is pursuing, and which the TTO will need to support, reduces the friction in the licence-to-commercial-activity handoff.
The third is to read Piece 3. The cost-line argument's strength depends on whether the founding team has the substrate to deliver the operator work the line funds; Piece 3 is the operational detail. If the substrate is not there, the cost-line argument is theoretical. The kit's logic, and the funding system's, increasingly, is that the substrate and the funding mature together.
Related reading
- Piece 1: The UK Spent £58 Billion on Research., the source argument.
- Piece 3: Chat Plus SaaS Is No Longer Enough., what the operator work the cost line funds actually looks like.
- Operational-gap audit toolkit, to surface which workflows would be funded by the cost line.