AI time value check

Add up the work your team does every week, choose what AI could help with, and see a cautious range for the value in year one.

A planning estimate, not a forecast. The saving rates are our own assumptions, set low on purpose, and real results depend on how well tools and training are put into practice. Nothing you enter leaves this page.

1. Your team

People doing this work

Average salary (£)

Regular AI users today: 20%

People who use an AI tool at least weekly for work.

2. Where the hours go

Tick the tasks your people do, and set hours per person per week.

3. Costs and how it lands

AI tool, £ per person per month

Training, £ per person, one-off

Our Practice Lab starts at about £660 per person (from £10,500 for a group of up to 16).

How much saved time turns into real value: 50%

Freed hours only count when they are used on something useful.

Months until people work at full speed: 4 People get structured, role-based training

Assumes more people become regular users (up to 80%) and the ramp is one month shorter.

How this is worked out

Each task has three saving rates (cautious, expected, hopeful). With every task ticked, the expected rates add up to about 2.2 hours saved per regular user per week, in line with what users reported in the St. Louis Fed survey. Hours freed = people × share who are regular users × hours per week × saving rate × 46 working weeks, with a month-by-month ramp. Only part of those hours is counted as value, because research has found that saved time does not always show up in pay or hours worked. Hourly cost = salary × 1.2 (employer costs) ÷ 1,650 working hours. The structured-training option adds 25 percentage points to regular use (up to 80%) and shortens the ramp by a month: that is our assumption, not a published figure.

Sources we checked:

UK median full-time earnings were £39,039 in April 2025 (provisional): ONS, Annual Survey of Hours and Earnings
US generative AI users reported saving 5.4% of work hours (about 2.2 hours of a 40-hour week); across all workers it was 1.4%: Federal Reserve Bank of St. Louis (2025). These are self-reported figures from the US.
Danish records show no measurable effect of chatbots on earnings or recorded hours, ruling out effects above 2% two years on: Humlum and Vestergaard, NBER w33777
Regular use is sharply higher for employees with at least five hours of training and coaching, and only a third say they are properly trained: BCG, AI at Work (2025)
Employer costs: HMRC rates and thresholds and GOV.UK workplace pensions (basis for the 1.2 uplift, which is a rounded estimate)

These studies give direction rather than inputs to the arithmetic, and none is a forecast for your organisation.