An AI plan tied to this year's targets

We start from this year's targets, pick the processes where AI moves them, and give each team tested setups, rules and training.

Our builds, June to October

1 to 9 days

from the start of a build to people using it, across 7 systems we built. The plan comes from the engineers who'd build it, so it's sized for what can go live.

Browse the case studies

Where to start with AI

If your AI licences haven't shortened the work, start from a target such as a faster month-end close or a shorter support backlog. Then place each candidate process by the hours it takes and the judgement it needs. The processes with the most hours and the least judgement go first.

Company of 200 peopleHours a month against the judgement each process needs
  • Supplier invoice entry180 hours a month, low judgement
  • Request triage150 hours a month, low judgement
  • Month-end reconciliations115 hours a month, medium judgement
  • Employee onboarding60 hours a month, low judgement
  • Weekly reporting30 hours a month, low judgement
  • Contract review75 hours a month, high judgement
  • Screening job applicants45 hours a month, high judgement
  • Credit decisions20 hours a month, high judgement

Screening job applicants and credit decisions are high-risk uses under the EU AI Act, so they come later, with a person deciding.

What changes for your teams

The same finance team with the same licences, before and after a goal, shared setups and rules.

AreaBeforeAfter
GoalUse AI moreClose the month in 5 working days, down from 8
SetupsEveryone writes their own promptsShared setups tested on month-end tasks
DataPeople guess what's allowedOne page of rules on data and review
ResultsNobody records what workedUse and hours saved reported each month

What our plans draw on

Plans come from people who build and run AI systems, using data and setups we use ourselves.

  1. AI exposure data

    Every task in 923 jobs scored for how much AI can take on today, with a second model checking a sample. A plan starts from your teams' real tasks.

    Works with O*NET task data, our own scoring
  2. Team pilots on the gateway

    A team starts on its own keys and budget, so use and cost show from the first day.

    Works with The private AI gateway
    First pilot: 3 data engineers on $100 a month each
  3. Rules for coding assistants

    Repository templates and rules that let coding assistants work safely in a codebase.

    Works with GitHub, CI
    Used across our own builds
  4. Products we run

    Flotta builds and runs fleets of websites, and Squadra HR runs a company's HR.

    Works with Cloudflare, company sign-in

What a plan is worth on licences you already pay for

A 200-person company that bought an AI licence for everyone and sees about 30% of them used each week.

  • 2 monthsto pay back the programme
What a plan is worth on licences you already pay for: what is counted, how, and the result
What we countHowResult
AI licences200 seats at €25 a month€60,000 a year
Seats used each weekAbout 30% before enablement60 people
Seats used each week afterAbout 70%, with shared setups and training140 people
Hours saved80 more weekly users, 1 hour a week, 46 weeks3,680 hours
Staff time back each year3,680 hours at €40 an hour€147,200

What each team gets

  1. 01

    Goals tied to this year's targets

    The targets, and the processes where AI moves them.

  2. 02

    Setups tested on real work

    Shared assistants and prompts for each team, tested on its own work.

  3. 03

    Rules on data and review

    Which data may go where, what needs review, and how to report a problem.

  4. 04

    Training on the team's tasks

    Sessions on each team's real tasks, with a monthly review of use.

Questions about AI consulting

Is this a strategy deck?

No. Every recommendation names a process, a measure and an owner, and we can build it with you.

Who is the training for?

Each team, on its own work. Leadership gets a shorter session on what to expect and how to measure it.

Do you cover governance?

Yes. Our AI governance service sets owners, permissions and records for the AI in use.

Find the processes where AI moves your targets

Bring this year's targets to a 30-minute call. We'll point to the processes where AI would move them.

Ask your AI assistant about us

Open our summary in the assistant you use, or copy the prompt.

Read the prompt