The hours AI saves
are the hours you bill.
Margin pressure leaves an accountancy firm no choice but to automate. The billable hour punishes it for succeeding. H3 works with firms on all three at once: adoption inside each service line, the core work rebuilt, and a new earning logic tested before it scales.
Everyone has a licence.The year-end still takes the same weeks.
The technology works in every demo. The firm around it has not moved.
Adoption without depth. People tried it once, for an email or a summary. Inside the audit file, the compilation or the tax return, the work runs as it did. Habit stopped at the safe first try.
Pilots in the hub, not in the service line. A central team builds promising tools. The partners who own the work never asked for them, so nothing carries volume.
A regulator in the room. Every shortcut looks like a risk to the file. Without governance that says where experimenting is free and where production is strict, speed itself feels indefensible.
And underneath all of it sits the paradox nobody at the table wants to name: every hour AI saves is an hour the firm can no longer invoice, and salaries do not fall with the hours.
The question is not whether AI works in an accountancy firm. It is who owns the hour it frees, and what you charge for once you stop selling it.
Four parts, one flywheel.
Not four workstreams running side by side. Each part feeds the next, and the service line, not a central office, is where the work changes.
Adoption, deepened per service line
We meet everybody where they are on the adoption ladder: a safe first try, then habitual use, then depth inside their own discipline, in clinics built on the workflows people actually run. Each service line names an ambassador and two sherpas, and the firm makes those recognised positions with protected time.
The core work, rebuilt per service line
Every service line writes its own transformation roadmap, starting from the end-to-end process rather than a list of use cases. A small innovation team of five part-time roles builds it, following one route from idea to production through a single governance gate that decides on the spot. The hub owns the standard and hands the process over across three cycles.
A new earning logic
One service line, ring-fenced, end to end, with a limited set of clients. Three candidates on the table: charging for the outcome, for access, or for the product. No model gets chosen before the pilot has run, and an explicit go/no-go comes before anything scales.
Leadership goes first
On every rung, before anyone below is asked to climb it. A change story per service line that names the job-loss fear instead of soothing it, and a dashboard that measures value and enjoyment of the work, not logins.
A 950-person firm, firm-wide, voluntary.
Nine in ten people across the firm now work with AI. Nobody was ever ordered to. The full case, with what each horizon delivered, stands on Our work.
- 869 active users against a headcount of 950, counted in the system
- 162 minutes a week freed per person on recurring work, self-reported in the first pilot, against a 45-minute target
- 79% said the work became more enjoyable. 96% of the 105 volunteers stayed active
- A fifth of all platform traffic now runs through assistants the firm's own people built, led by tools for audit files and year-end compilation
- The billing question is open. The experiments run inside a ring-fenced service line, with the go/no-go still ahead
Adoption is measured. Process redesign is running. The new business model is under way, and that is the decision that determines whether the transformation pays for itself.
The workflows a firm repeats.
These are examples, not a menu. In a firm, the first redesigned workflows come from the service lines themselves, and they are the ones that carry volume every week.
The evidence gathered, cross-referenced and flagged before the reviewer opens the file.
From ledger to draft accounts with every adjustment traceable to its source.
Exceptions surfaced across the whole population, not a sample.
Know-your-client checks and engagement letters drafted from what the firm already holds.
What the annual accounts say about the client's business, written for the conversation after the figures.
Recurring questions answered from the firm's own positions, with a person signing off.
What changed, who it affects and which files need a second look.
Value and enjoyment of the work per team, not logins.
Governance built to speed the work up.Not to slow it down.
In a firm a regulator audits, the pace has to be defensible. That is a design question, not a reason to wait.
- Free experimentation in a sandbox. Anyone in a service line can try, on the work they know, without asking
- Strict rules in production. Data, review and sign-off are explicit before anything touches a client file
- One gate that decides on the spot. A single route from idea to production, so nothing waits three committees for an answer
- The hub owns the standard, the service line owns the work. Leading the first cycle, coaching the second, stepped back by the third
A firm sits across all three horizons.
Depth inside each service line is adoption. The audit file and the year-end rebuilt is redesign. And a firm that charges for something other than the hour is a new business model. It is the same climb, run inside one firm, with the earning logic decided last and tested first.
Four ways to start.
A firm-wide programme is not the only way in. Most firms start with one service line and one workflow, and the smallest step still ends in a decision.
The workflow baseline
First step · three weeksIn three weeks we put a number on one workflow: what it could save, where quality and work satisfaction can improve, and what the change would take. Enough to decide whether to proceed, without committing to an open-ended programme.
What you get
- A substantiated range for the time and cost that could be freed up, with every assumption made visible
- Where quality and work satisfaction can improve: fewer errors, fewer handovers, less daily friction
- The levers that have to move: the work, the technology supporting it, and the business rules around it
- A first delivery approach with the decisions and owners it requires, and an indicative cost range for implementation
- A go or no-go that we act on ourselves
Firms that can see the potential of AI in one service line, but not yet what it would deliver.
Executive workshop
1 to 2 daysOne room, the partners who have to carry the change, and a decision written down at the end.
What you get
- Where value is moving in accountancy
- Your three horizons mapped on your own service lines
- The decision written down, with owners against it
- What you do in the next 30 days
Boards and partner groups that must agree on direction before committing budget.
Use case sprint
2 to 4 weeks · small cycleOne workflow rebuilt with your people, running in production when we leave.
What you get
- One workflow rebuilt, not advised on
- Your people in the build from day one
- The baseline re-measured after the rebuild
- Handover to a named owner
Service lines with one process that visibly costs more than it should.
Roadmap
6 to 12 weeks · large cycleThe path across the three horizons for the whole firm, with ownership assigned per service line and the first cycle already run inside it.
What you get
- The three horizons mapped for your firm
- One cycle run inside the roadmap, not after it
- Governance that survives our leaving
- A board session your people run, with us in the room
Firms moving from a stalled initiative to an operating capability.
Bring us the service line where the hours are.
Tell us which work your people repeat every week, and what the firm would do with the capacity if it came back. Some of what lands on this page turns out to be a pricing question, not an adoption one, and that is worth finding out in an hour.