Feasibility Study Agent
States what would have to be true, and what happens when it is not
- 8hard rules
- Things it must never do, enforced in every run
- 5handover rules
- Conditions where it stops and gives the work to a person
- 19output fields
- A structured result, not a paragraph you have to re-read
- ≤6steps per run
- Bounded work: at most 5 tool calls, enforced by the runtime
Read-only by design. This agent has no path to sending, spending, publishing or committing anything. Where that limit is the product, removing it would remove the reason to trust the output.
Overview
Almost every feasibility study concludes that the project is feasible. That is the genre's defect: it is commissioned by somebody who has decided, and read by a bank that needs a document. Eighty pages arrive, "viable" is the verdict, and nobody can say afterwards which number the whole thing was resting on.
A study is worth something only if it could have concluded the opposite, and if the reader can find the assumptions that decide the outcome. Every project has two or three. A restaurant is a bet on covers per night and rent as a share of revenue. A factory is a bet on utilisation and input price. An app is a bet on acquisition cost against retention. The rest is scenery.
So the output is inverted. Not "the project returns 24%" — a figure carried to two decimals from inputs that were guesses — but "this works while food cost stays under 31% and Tuesdays fill 40 covers; here is what happens at 35% and 30; and here is which of the two to go and measure before signing anything."
It covers the whole structure a lender or a partner expects — market, operations, technical, legal, organisation, financials, risk — but every section is written to be falsifiable, and it separates the figures you supplied from the ones it assumed and from the ones nobody has yet.
It gives no recommendation to invest, no valuation, and no financial or legal advice.
Benefits
A document a lender can interrogate
Because every figure names where it came from.
The real risk, early
Found in an afternoon rather than in the second year of trading.
A decision you can defend
Including the decision not to proceed.
What it does
Finds the two or three numbers that decide it
Every project rests on a handful of assumptions and eighty pages of scenery. It names which is which before anything else.
Break-even on each, not a single verdict
How far each assumption can move before the answer flips. A return carried to two decimals from guesses is theatre.
Three cases, honestly separated
Conservative, expected and optimistic — with what changes between them stated, not just the totals.
Marks what was supplied, assumed, or unknown
Most studies blend all three into one confident column. Knowing which is which is the difference between a study and a brochure.
Says what to go and measure first
The one or two facts worth the cost of finding out before committing, ranked by how much they move the outcome.
Can conclude that it does not work
And states the conditions under which it would. A study that cannot fail is not a study.
What this changes
What this changes about your week
- It takes over
- Producing the study a lender or a partner expects, with the assumptions exposed rather than buried.
- Done today by
- A consultant at several thousand, or the founder at midnight in a spreadsheet nobody else can follow.
- How often
- Before every lease, licence, loan or new line.
- Typical manual effort
- Weeks, and eighty pages that conclude "feasible" because that is what the commission was for.
Why Professional Services in particular. Almost every study concludes the project is feasible. One that names the two numbers deciding it, and what happens when they move, is a different document entirely.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Feasibility Study Agent has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-opus-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Demo
Demo dataIllustrative sample output, abridged.
Input
{
"project": "A 40-cover restaurant in a mid-size city centre",
"currency": "USD",
"capitalAvailable": 180000
}Output
{
"cases": [
{
"name": "conservative",
"worksOrNot": "Does not work with the capital available. It runs out in month eleven, which is before it fails on trading — the shortfall is cash timing, not the business.",
"whatChanges": "29 midweek covers, food cost at 34%, three months to reach steady trade instead of two.",
"peakCashNeed": 214000,
"monthsToBreakEven": 19
},
{
"name": "expected",
"worksOrNot": "Works, with 13% of the capital available as headroom. That is thin for a first year.",
"whatChanges": "34 midweek covers, food cost at 31%.",
"peakCashNeed": 156000,
"monthsToBreakEven": 11
}
],
"risks": [
{
"risk": "Midweek trade does not develop",
"effect": "The single largest exposure. Below 29 covers the project does not reach break-even at any point in the horizon.",
"earlyWarning": "Midweek covers in weeks 6 to 10, once opening curiosity has passed."
}
],
"market": "…",
"escalate": true,
"technical": "…",
"conclusion": "Feasible in the expected case and not in the conservative one, and the gap between them is a single number nobody has measured: midweek covers. This is not a decision to make on the model — it is a decision to make after four evenings of counting.",
"disclaimer": "Assumptions and break-evens from the figures supplied. This is not investment, legal or financial advice, contains no valuation, and is not a recommendation to proceed. Figures marked assumed were not researched or supplied.",
"operations": "…",
"measureFirst": [
{
"question": "What do comparable midweek covers actually run at in this location?",
"howToFindOut": "Count arrivals at two comparable venues within 400m on four midweek evenings. Two evenings of somebody's time, against a decision worth $180,000.",
"whyItMatters": "It is the assumption with the least headroom and the one the whole model rests on."
}
],
"organisation": "…",
"stillUnknown": [
"Local midweek demand",
"Extraction consent status for the site"
],
"assumedFigures": [
"Covers per night — assumed from venue size and typical turn, not from any measurement of this location.",
"Food cost percentage — assumed at a common range for this format."
],
"capitalRequired": 156000,
"suppliedFigures": [
"Rent",
"Capital available",
"Site size"
],
"whatThisRestsOn": [
{
"source": "assumed",
"headroom": "15% — the thinnest margin in the model.",
"assumption": "Covers on a Tuesday and Wednesday",
"assumedValue": "34 per night",
"whyItDecides": "Weekends fill in almost any plausible case and are not what decides this. Midweek carries the fixed costs, and the difference between 29 and 34 covers is the difference between the project working and not.",
"breakEvenValue": "29 per night"
},
{
"source": "supplied",
"headroom": "Comfortable at the rent supplied — but the lease term matters more than the rate.",
"assumption": "Rent as a share of revenue",
"assumedValue": "9%",
"whyItDecides": "It is the only cost here that cannot be reduced once signed.",
"breakEvenValue": "13%"
}
],
"escalationReason": "The conservative case fails on cash before it fails on trading, and the deciding assumption has never been measured.",
"legalAndRegulatory": [
"Food business registration, a premises licence if alcohol is served, and an extraction consent that is frequently the item that delays opening. Whether each applies to your site is a question for your local authority, not for this study."
],
"rampAndSeasonality": "The model reaches steady trade in month three. Peak cash need is month eight, not month one — a project that works at steady state and runs out in month eight is not feasible, and an annual average hides exactly that.",
"conditionsUnderWhichItWorks": [
"Midweek covers hold above 29.",
"Food cost stays under 33%.",
"At least $30,000 of the capital is held back for the month-eight cash trough rather than spent on fit-out."
]
}Requirements
Integrations
No integrations required.
Before you start
- What the project is, where, and at what scale
- Any costs, prices or volumes you already know
- The capital available and where it comes from
- Any constraint that is fixed — a site, a licence, a deadline
- Requires the Starter plan or higher
Use cases
New venture or branch
Before a lease, a licence or a loan is signed.
Bank or investor submission
The structure a lender expects, with the assumptions exposed.
A new product line or market
Whether the existing business can carry it.
Comparing two options
The same falsifiable treatment applied to both.
Grant and programme applications
Where the assumptions have to be stated to be assessed.
Pricing
Starter
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Will it tell me whether to invest?
No. It states what would have to be true and how far each assumption can move before the answer changes. Recommending an investment is regulated financial advice in most places and requires a licensed adviser who knows your circumstances.
Where do its numbers come from?
From what you supply. Anything it assumes is labelled as an assumption with the reasoning, and anything neither supplied nor safely assumable is listed as unknown rather than filled in. It does not invent market sizes or benchmark costs and present them as findings.
Does it value the business?
No. It produces no valuation, no multiple and no share price. Those are opinions with legal weight and they belong to a qualified professional.
Can it work with very little information?
Yes, and it will say so plainly. With a thin brief it produces the structure, the assumptions that would decide it, and the list of what to establish — which is genuinely useful and is not the same as a study.
Does it cover legal and regulatory feasibility?
It flags the permits, licences and regulatory questions a project of that type usually faces, so none is a surprise. It does not give legal advice or confirm that any specific requirement applies to you.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent