forta.studio

AI Proof Pilot

Automate your most painful process, proved on your own numbers

Pick the single task that eats your team's week. We build an intelligent AI system that does it on your real data, for a fixed fee, with the success metric agreed up front.

Live on one process in about two weeks. Proved on your metric in 30 days. One extra month on us if needed. £9,950. Hit the metric, or you don't pay.

The problem

Smart, expensive people doing work that software should handle

Your best people lose hours every week to the same repetitive admin: copying data between systems, formatting the same reports, chasing information, triaging the same inbox. It's expensive, it's draining, and it's exactly the kind of work that doesn't need a human. Most teams know this. They just don't have a safe, proven way to hand it over.

The offer

One painful process, automated and backed by a metric

Pick the repetitive task draining the most hours from your team. We scope it, lock the success metric in writing at kickoff, and build a working AI system that does it on your real data for one fixed fee. Once access and data are in place, it's typically live in about one to two weeks. Then you get 30 days to see it hit the number on real work. If we can't get there, you don't pay.

The Pilot // Process
03 Steps
Process

Live in about two weeks, proved over 30 days

One process, from baseline to live in about one to two weeks once access is in place, then measured against the metric we locked at kickoff.

01

Scope & lock the metric

We pick one painful process, lock the single metric that defines success in writing at kickoff, and measure where you are today. Without a baseline there's no way to prove the saving, so this comes first.

02

Build & go live

We build on your real data and your real edge cases, not a sanitised demo. Once access and data are in place, the workflow is typically live in about one to two weeks. You see it working through the build, not just at the end.

03

Prove it on real work

You get a 30-day prove window to hit the metric on real work. Hit it and the balance falls due, including if we get there early. Fall short at day 30 and we carry on for up to another 30 days at no extra pilot cost. If we still can't get there, you don't pay.

What you get

Something working, and the proof it works

01

A working AI system

Not a slide deck or a throwaway prototype. A system doing your chosen process on real data, in production.

02

Evidence against the locked metric

The before-and-after numbers: how long the process took, how long it takes now, and the hours reclaimed, measured against the metric we agreed at kickoff. Evidence, not assertion.

03

Hand-over docs and a runbook

Everything your team needs to run, monitor and trust the system: how it works, what to watch, and what to do if something looks off.

04

Recommendations to scale

Where the next highest-leverage automation sits and what it would take, so you decide what comes after with eyes open.

What a pilot looks like

Two example workflows

The flow is simple by design. Making it reliable on your real data is the work. That's what a pilot builds and proves.

Illustrative examples of the shape a pilot can take, not real client systems.

Flow 01

Supplier & contract onboarding

Onboarding

You stay in control. Anything the system isn't sure about gets routed to a person.

Documents arrive

Whatever the format: PDFs, scans, emailed forms, each supplier's own layout. The system takes them as they come.

The system processes it

Reads each document, pulls the key terms, and cross-checks them against your existing records and rules, the part that quietly eats hours of a senior person's week today.

Human review

Only the cases the system flags as uncertain reach a person. Everything clear-cut is handled automatically. You set where the line sits.

Recorded across systems

Written into your CRM, finance and document systems consistently, with nothing missed or keyed in twice.

Flow 02

Quote & proposal preparation

Quoting

You stay in control. Every quote is checked by a person before it goes out.

Request comes in

From wherever they land (inbox, web form or CRM), captured automatically so nothing slips.

The system prepares it

Pulls the right pricing, terms and history, and drafts a complete quote, the work that normally ties up your best person for an afternoon.

Human review

Your team reviews and adjusts before anything goes out. The system does the heavy lifting; the judgement stays with you.

Sent to the client

Delivered, logged and tracked, so follow-up never falls through the cracks.

Every pilot runs on the Forta Toolkit, the same AI infrastructure behind all our client systems

  • Observability

    Every model call traced

  • Safety

    PII redacted, guardrails on

  • Cost control

    Spend capped per workflow

  • Durability

    Retries, recovery, no lost work

The guarantee

We carry the risk, not you

Before we build, we lock the one metric that defines success, in writing, and measure your starting point. Once it's live, you get 30 days to see it hit that number on real work, and one further month on us if we're short. We hit the metric, or you don't pay, and you keep everything we've built. The guarantee holds as long as you hold up your side: the access, data and timely feedback we agree at kickoff.

The metric, locked at kickoff

We define the one number that means success, in writing, before any build starts. No moving the goalposts later.

Hit it and the balance is due

The second half is only ever due when the metric is hit and evidenced. Get there before day 30 and it still counts. You never pay out the clock.

Short at day 30? One more month on us

Up to another 30 days on the same scope and metric, at no extra pilot cost. If we still can't get there, you don't pay and you keep everything we built.

Proof

What our clients say

Our vision was completely understood and our goals became Forta Studio's goals. The end result is nothing short of incredible.

Sean Brattley

Founder, MerchantLink

Proper quality work from people who clearly care about what they put their name to.

Harley Sanders

Founder, Incus Audio

The maths

How it pays for itself

One person doing eight hours a week of repetitive admin, at a loaded cost of around £50 an hour, costs you roughly £20,000 a year, every year, indefinitely.

A pilot is a one-off £9,950. If it works, that saving keeps landing, year after year. Even after the running costs of keeping it live, the pilot pays for itself comfortably inside a year, and the system keeps running long after.

And if it doesn't hit the number we agreed, even after the free extension? You don't pay. That's the whole reason the guarantee exists: the maths only works in your favour.

A worked example

Here's an illustrative shape of a typical pilot:

A growing firm brings on new suppliers and clients regularly, and each one means onboarding: collecting documents, checking the details, cross-referencing them against existing records, and setting everything up across their systems. It's careful, repetitive work, and it's usually done by someone senior enough to catch the mistakes, which makes it expensive.

A pilot would scope that one process, agree the metric (say, cut the hours spent per onboarding by 80%) and build a system that reads each document, extracts and validates the key details, sets up the records automatically, and flags only the cases that genuinely need a human.

At around 12 hours a week reclaimed from senior staff, that's roughly £37,000 a year back before running costs. Against a £9,950 pilot, the saving is the easy part: the only question we'd answer first is whether your version of this is as clean as it sounds. That's what the scoping call is for.

At a glance

~12 hrs

Reclaimed each week

~£37,000

A year, before costs

£9,950

One-off pilot fee

Run your numbers

Put your own figures in

1
8h
The fully-loaded hourly cost of the person doing the work: salary plus on-costs like National Insurance, pension, benefits, tools and overheads. Usually well above their headline pay rate.£50
How it runsRun & Maintain is £500/month for us to host, monitor and run it. Hosting and AI usage stay on your own accounts, billed at cost, either way. Self-host and you keep that fee but carry the monitoring, updates and your team's time when it breaks.
The per-use cost of the AI models, billed to your own AI provider account on separate billing. You pay this whether we run it or you self-host, and it's never marked up.£150/mo
Cloud hosting and infrastructure to keep the system live, on your own accounts and billed at cost. You pay this whether we run it or you self-host, and it's never marked up.£50/mo

Reclaimed hours are worth

£20,800

a year, gross

Net saving after running costs

£12,400

with Run & Maintain · £700/mo

Pilot pays back in

10 months

9.6 months to be precise

Reclaiming this is worth ~£20,800 a year. Hosting and AI usage run on your own accounts at cost (about £200/month); Run & Maintain adds £500/month for us to run it, so it's £700/month all in. Your net saving is £12,400 and the pilot pays back in 10 months. You'd carry hosting and AI usage self-hosting anyway; the £500/month is just for us to run it, monitor it and fix it.

Pricing

One fixed fee to prove it, then a simple way to keep it running

The pilot

£9,950+ VAT

Fixed fee. One process, live in about two weeks and proved over 30 days.

What's included

  • Scoping workshop and the success metric locked in writing
  • A measured baseline of the current process
  • The working AI system, built on your real data
  • Hand-over documentation and a runbook
  • A 30-day prove window on real work
  • A readout of the proven result against the baseline
  • A further 30 days at no extra pilot cost if we're short at day 30
  • Recommendations on what to automate next

The metric is locked in writing at kickoff. We hit it, or you don't pay. The number that defines success is fixed before we start.

How it's paid

  • 50% deposit on signature, due before kickoff.
  • 50% balance when the agreed metric is hit and evidenced, in the prove window or the extension, including early.
  • No extra pilot fees during the free extension.
  • If we still can't hit it, the deposit is refunded in full, the balance is never invoiced, and you keep the deliverables, provided the access, data and feedback we agreed at kickoff were in place.

Keeping it running

Optional

Run & Maintain

From£500/month + VAT

Once it's proven, we keep it doing its job, so the saving keeps landing month after month. We host, monitor and maintain it for a fixed monthly fee. Hosting and AI usage run on your own accounts, managed by us and billed to you at cost, never marked up. The natural next step, never a condition.

What it covers

  • Hosting and infrastructure managed by us, billed to your accounts at cost
  • Monitoring and alerting via the Toolkit's observability layer
  • Error handling, retries and incident response
  • Model and prompt updates as providers change
  • Small tweaks as the process evolves, up to 2 hours/month included
  • AI usage stays on your own AI accounts, billed to you, never marked up

Or run it yourself

You own the system and the code, so you can take the handover and self-host it on your own infrastructure. Running costs go straight to your own accounts, never marked up. The trade-off is simple: if you self-host, it isn't supported, monitored or maintained by us. Run & Maintain exists for the teams who'd rather that stayed our problem.

We only take two pilots a month

Each one gets our full attention from kickoff until the metric is hit. That's the only way the guarantee works, extension included. When the month is full, it's full.

FAQ

Questions, answered

How it works
  1. We lock the metric. Before anything starts, we define the success metric and the baseline together, in writing, at kickoff. You know exactly what we're aiming at and how it'll be measured.
  2. You secure your slot. A 50% deposit books your place in the build schedule. It's fully refunded if we can't hit the agreed metric, so it's tied to your result, not just a payment.
  3. We build and go live. We automate the one workflow we've scoped and get it running in your business, typically about one to two weeks after access and data are in place. You keep everything we build, whatever happens.
  4. We prove it over 30 days. Once it's live you get a 30-day prove window on real work. When the metric is hit and evidenced, the balance falls due, and that counts if we get there early. Fall short at day 30 and we carry on for up to another 30 days at no extra pilot cost. If we still can't get there, you don't pay: deposit refunded in full, balance never invoiced, and you keep everything we built.
What's included in the fixed fee, and what isn't?

The fee covers everything we do to deliver the pilot: scoping, building, testing, and handing over a working system on your data, plus documentation. What it doesn't cover is running the system once it's live. After the pilot you choose how it runs:

Managed (Run & Maintain). We host, monitor and maintain it from £500/month + VAT. Hosting and AI usage stay on your own cloud and AI accounts, managed by us and billed to you at cost, so they're never marked up and you keep full visibility. The fee is for the running and support, not the running costs themselves.

Self-hosted. Take the handover and run it on your own infrastructure and accounts. All running costs go directly to you, never marked up. The trade-off is that it isn't supported or maintained by us.

Either way there's no markup on usage. We'll give you a clear estimate of expected monthly running costs during scoping, and cost controls are built in as standard so there are no surprises.

What happens to my data?
Your data stays yours. We work on it only to build and test your pilot, we never use it to train any AI model, and we don't share it. The system runs on your own infrastructure and accounts in both cases, so your data lives with you, not us. If you self-host, we have no ongoing access once it's handed over. If you take Run & Maintain, we operate it on your infrastructure on your behalf, which makes us your data processor: that's covered by a data processing agreement we sign before any data changes hands. Our standard safety layer scans inputs and protects sensitive information as part of every build, and anything we held for development is deleted on request.
What if you don't hit the metric?
First, we keep going. If we haven't hit the agreed metric by the end of the initial 30-day prove window, we carry on at no extra pilot cost for up to another 30 days on the same scope and metric. If we hit it in that window, the balance is due. If we still can't get there, you don't pay, and you keep everything we built. The second half is only ever due when the metric is hit, which can be before day 30 if we get there early. The guarantee assumes you've held up your side: the access, data and timely feedback we agree at kickoff.
How does payment work?

The pilot fee is fixed at £9,950 + VAT, split in two and tied to the result rather than the calendar:

  • 50% deposit on signature, due before kickoff. It books your slot in the build schedule.
  • 50% balance when the agreed metric is hit and evidenced. That applies in the initial prove window or the extension, and it counts if we hit it early. You never wait out the clock to pay for something already proven.

There are no additional pilot fees during the free extension. If we still can't hit the metric by the end of it, and you've met the access, data and feedback dependencies we agreed at kickoff, you get the full deposit back, the balance is never invoiced, and you keep the deliverables. If those dependencies aren't met, the guarantee path doesn't apply and the deposit is non-refundable under the agreement.

Who owns the code?
You do. On hand-over you get the working system, the documentation and the runbook. No lock-in, no per-seat licence, no hostage situation.
Do I have to take the retainer?
No. You own the system and the code either way. You can take the handover and self-host it on your own infrastructure — just note that if you self-host, it isn't supported or maintained by us. Run & Maintain (from £500/month + VAT) is there for teams who'd rather we kept it hosted, monitored and up to date. Hosting and AI usage stay on your own accounts either way, billed at cost and never marked up.
Can it really go live in a couple of weeks?
Usually, yes, because the scope is deliberately narrow (one process) and the hard infrastructure already exists. Observability, safety, cost control and durability are built into the Forta Toolkit, so the build time goes on your problem, not on plumbing. The clock starts when access and data are in place — that's the part that most often moves the date, which is why we sort it during scoping rather than after kickoff.
Can we automate more processes later?

Yes, and it costs considerably less than the first one. Most of a first pilot's fee goes on foundations rather than the process itself: connecting to your systems, getting access to your data, the safety and monitoring layers, and understanding how your business actually works. That's all built and paid for after the first pilot.

A second process only pays for the process. Additional workflows start from £4,950 + VAT, around half a standalone pilot, with the exact figure depending on complexity and how many new systems it needs to touch. Same fixed fee, same agreed metric, same guarantee.

Adding workflows increases what we're running for you, so the Run & Maintain fee steps up accordingly, and we'll always price that before you commit.

I've finished a pilot and want to go further: do I need another audit?
It depends how far you're going. If you're extending the process we already proved, we skip straight to a proposal, no need to re-cover ground we've covered. If you want to take AI business-wide, across processes the pilot never touched, we run an Expansion Audit: a focused review priced at £1,500 + VAT, lighter than a full discovery because we already know your data and stack. It maps the broader opportunity and, like our AI Opportunity Audit, is credited in full against the full AI system build that follows.

Book your pilot call

Pick a time to discuss your pilot build with us

A short, no-pressure call to find the right process and confirm the pilot is a fit. If it isn't, we'll tell you.

Prefer to reach out first? Get in touch via the contact form.