Services
Three fixed-price stages. Buy one, or all three.
Most clients start with the review. Each stage ends with something you can act on, and leaves your team more able to do the next one without us.
Stage 1
Pilot Stall Review
A clear decision on every pilot. Stopping one is a good outcome if it stops the spend.
- ForAny business with stalled pilots
- Duration2 weeks
- PriceFixed price · from £[PRICE_REVIEW]
What happens
- Days 1 to 2Agree with your sponsor which pilots are in scope (up to three), and get access to the documents, data samples and build.
- Week 1Interviews with the sponsor, process owner, IT and risk for each pilot. A hands-on look at the data and the build.
- Week 2Each pilot assessed against the six stall points, with the cost and time to production for those worth fixing.
- Day 10Playback to your sponsor, and the written recommendation.
What you get
- A stall diagnosis for each pilot
- A fix-or-stop recommendation, with the cost and time to production
- A one-page summary for the board or investment committee
If a pilot should be stopped, we'll say so in writing, and say why.
Stage 2
Production Push
One pilot, taken to a live production slice with real users.
- ForOne pilot worth fixing
- Duration6 to 10 weeks
- PriceFixed price · from £[PRICE_PUSH]
What happens
- Week 1Name the business owner and agree the success measure. Write the testable spec and build the test set.
- Weeks 2 to 3Set up the data path for the production slice. Agree the architecture and controls with IT and risk.
- Weeks 3 to 7Build the production slice with your team, tested against the test set at every change.
- Weeks 6 to 8Evidence pack to risk for sign-off. Release to the first real users.
- Final weeksRoll out with the users, measure weekly use against the success measure, and fix what gets in the way.
A small senior team embeds with yours. Some people call this forward deployed engineering (FDE): engineers working inside your team and environments, rather than building elsewhere and handing over.
What you get
- A testable spec and test set
- A working data path
- The production build
- A governance and evidence pack for risk sign-off
- An adoption plan
- Weekly usage measured against the agreed success measure
Stage 3
Capability Handover
So the next pilot doesn't stall.
- ForMaking the next pilot land
- DurationUp to 3 months
- PriceFixed price · from £[PRICE_HANDOVER]
What happens
- Month 1We pair with your team on the next pilot and set up the reusable patterns, evaluation harness and governance templates.
- Month 2Your team leads. We review, coach and step in where needed.
- Month 3Your team runs it to production with us on call, and we hand over a short internal playbook.
What you get
- Coaching and pairing with your team
- Reusable patterns, an evaluation harness and governance templates
- A short internal playbook
- Your team running the next pilot to production, with Opta6 on call, not in charge
Why bring someone in
Partnered builds land more often.
The same MIT research that found most pilots stall also found that bought or partnered AI solutions succeeded about two-thirds of the time, against about one-third for internal builds.1
Partnering doesn't mean handing over. We work inside your team, and every stage leaves them able to do more of the next one without us.
Questions
What people ask before they start
Do you replace our team?
No. We work alongside your people and leave them able to run what we build. The aim is that you need us less each time.
What if the pilot should be stopped?
We'll tell you, in writing, with the reasons. Stopping a pilot that won't pay back is a good outcome: it stops the spend and frees the budget for one that will.
Do you work with our existing vendor?
Yes. If a vendor built the pilot, we work with them. Our job is to get it into production, not to replace whoever started it.
Who does the work?
The people you meet. We're a small senior team and don't hand work to juniors after the sale.
What do you need from us?
A sponsor, a process owner, and access to the data and environments within two weeks.
- MIT NANDA, The GenAI Divide, 2025, as reported in Fortune, August 2025. ↑
For private equity
Portfolio companies reporting AI pilots but no EBITDA impact? We run the same Pilot Stall Review across several of them, with comparable scoring, so you can see where the problem sits and which pilots are worth fixing. Then we take the pilot with the clearest return through a Production Push.
Find out where your pilot is stuck.
Twelve questions, five minutes, an answer on screen. Or talk it through with the people who'd do the work.