For firms that bill for their people’s time
Firms that fall behind on AI are losing clients.
We find the work worth handing to a machine, build it, and prove what it gave back. Your team gets the hours, and you take on more clients without hiring.
What clients now say
- 78%
- of clients say better work through AI is important or essential when they pick a firm.
- 6%
- say most of the firms they already pay actually deliver it.
Source: Thomson Reuters, Future of Professionals 2026. 1,816 professionals across 62 countries, surveyed March to April 2026.
Work delivered for
- Singapore Government project
Why now
Your clients are already deciding.
These are not our numbers. They come from a survey of 1,816 professionals across 62 countries, published this year. Every figure below links to it.
78%
of clients say better work through AI is important or essential when they pick a firm.
6%
say most of the firms they already pay actually deliver it.
32%
will reconsider which firms they use inside the next twelve months.
$143bn
of US legal and accounting revenue is under active reconsideration, on Thomson Reuters’ own estimate.
Source: Thomson Reuters, Future of Professionals 2026. 1,816 professionals across 62 countries, surveyed March to April 2026.
Meanwhile 74% of professionals in firms like yours already use these tools every week, and 91% say their own firm gets less out of it than it should. Being busy with AI and being better because of it are two different things. Your clients can only see the second one.
Where the week goes
You are not short of clients. You are short of hours.
Every firm that bills for time loses it to the same six jobs. A client pays for none of them.
Onboarding a new client
The same details typed into four different systems.
Chasing people for things
Documents, approvals, signatures. Someone has to keep asking.
The monthly report
Numbers pulled out of three tools, then last month’s commentary rewritten.
Handing work between people
Sales to delivery. The context gets retyped and something is missed.
Proposals and scoping
Find a similar old job, rebuild the pricing, format the document.
Marketing that never happens
Case studies and posts, always last, because client work comes first.
Thomson Reuters put a figure on what getting those hours back is worth.
200 hrs
a year, per person, is what these tools are expected to give back.
$100k
a year in extra billable time, for one US lawyer, on the same forecast.
Source: Thomson Reuters, Future of Professionals 2024. more than 2,200 legal, tax, accounting and risk professionals, surveyed April to May 2024.
What changes
The same job, with fewer people moving it along.
Onboarding, as an example. The work still happens. Your team stops being the thing that carries it from step to step.
Bringing on a client today
Six steps, and a person has to start every one of them.
- Contract signed
- Chase documents by email
- Copy details into the CRM
- Set up folders and access
- Brief the delivery team
- Send the welcome pack
After a sprint
The judgement stays with your team. The chasing, copying and drafting does not.
- Contract signed
- Documents requested and chased
- CRM, folders and access created
- Draft brief written from the file
- Someone checks it and sends
What the research says
Picking the right job is the whole game.
Harvard and BCG ran the experiment properly. 758 consultants, 18 real tasks, half with AI and half without. On the jobs AI was good at, the people using it finished more work and finished it faster.
12.2%
more work finished.
25.1%
faster on the same tasks.
Then they measured the jobs AI turned out to be bad at. The same consultants, the same tool, and this result:
19 points
worse than using no AI at all, on the tasks it turned out to be bad at.
Source: Dell’Acqua and others, Organization Science. 758 Boston Consulting Group consultants across 18 real consulting tasks.
The tool was identical. The choice of job was the entire difference. That is the part most firms get wrong, and it is why we start by timing your work rather than by buying you software.
How we work
Identify. Build. Prove.
The same three steps every time, in the same order.
Identify
We time the jobs your team repeats. You get the short list worth fixing, and the longer list that is not.
Build
We build it into the tools your team already uses. No new system to learn.
Prove
We time the same job again. You get the before figure and the after figure in writing.
Where to start
7-Day AI Workflow Sprint, $2,500 fixed
One workflow. Seven days. A working system at the end, and the before and after figures beside it.
- One workflow
- Mapping and timing what happens today
- The build
- Integrations with your tools
- Testing on your real cases
- Deployment
- Documentation
- Handover to your team
- Before and after measurement
It is worth doing when one job takes your team ten hours a week or more between them. Below that the arithmetic does not work, and the page for it has a calculator that will tell you either way.
Flagship engagement
The exception queue nobody could read
A logistics operation at DHL scale generates more exceptions than any team can read. The opportunity was not a chatbot and it was not a dashboard. It sat in the gap between what the operational systems already knew and what the people handling exceptions were able to see in time to act. We found it by sitting with the people doing the work, built into the systems that were already running, and measured the result against the position before the programme started.
The programme contributed roughly $100M in annual operational value at DHL scale. A figure that size is never one person’s work, and the specific systems cannot be described publicly. The part worth talking about is the method that found it, which is the same one described above.
Read the DHL case studyFAQ
Questions people ask before they book
Next step
Bring the job your team complains about most. Thirty minutes is usually enough to say whether it is worth fixing.