SmartGT

Consulting • process audit • AI rollout

An AI rollout —measured in person-hours, not in promises

We find where AI takes the routine away, and calculate the effect before the work starts.

Get the numbersGet in touch
  • We calculate the effect before the work starts
  • Figures with sources
  • No upselling

What the research says

Three figures from published research — not from a vendor's slides.

Just hand over the tool
5,2%
of their time is saved by the average AI user
Details

Across all workers, including those who do not use AI at all, the saving is only 1.4%. A survey of employed people in the USA.

Federal Reserve Bank of St. Louis, 2025 ↗
Build it into the process
+14%
more enquiries resolved per hour after an AI assistant was introduced
Details

Not a laboratory but live support work: 5,179 agents.

Brynjolfsson, Li, Raymond · NBER ↗
Where the newcomers are
+34%
gain among the least experienced workers
Details

On experienced staff the effect was minimal — hence the choice of where to roll out.

Brynjolfsson, Li, Raymond · NBER ↗

How many hours that is in your case

How many hours a month does your team spend on enquiries and routine?

hours a month
The tool just handed over
≈ 5% of working time
hours a month
Built into one process
≈ 12% of working time
hours a month
A team with many newcomers
≈ 25% of working time
Calculate it for your teamA benchmark from research, not a guarantee
Where the figures come from and how we calculate

5,2% — that is how many more hours the average user of generative AI would have had to work to do the same job without it (across all workers, including those who do not use it, the figure is 1.4%): Bick, Blandin, Deming (2025), Federal Reserve Bank of St. Louis, WP 2024-027.
+14% і +34% — the increase in enquiries resolved per hour across 5,179 support agents: Brynjolfsson, Li, Raymond, “Generative AI at Work”, NBER WP 31161. To get the hours freed up, we convert the productivity gain into time: the same work takes 1/1.14 and 1/1.34 of the former time, that is ≈12% and ≈25% of the hours.
What this does not mean. It is not a guarantee for your business but a benchmark from somebody else's measurements. The difference between 5% and 12% is not a different AI model but a rebuilt process. We calculate the real effect on your own data during the audit.

How we work

First we work out whether there is anything to save. If no gain is visible we will say so rather than sell you a rollout.

The process auditWe take apart the team's working day — enquiries, correspondence, reports, moving data by hand — and count the hours for each item.
Selecting for AIWe take processes with frequent, uniform repetition and a clear criterion for a correct answer: that is where the effect is largest and most predictable.
A pilot on one processWe launch on a narrow slice and measure before and after on your data. Whether to scale up is decided by the pilot's result.
Handover to the teamWe build it into the working tools, train the people and leave the instructions. The goal is for it to work without us.

Start with a process audit

During the audit we calculate the effect on your own figures. The result is yours — even if you go on to do the work yourselves.

1 200 грн/год

Our development and rollout rate. The first conversation is free.

Discuss the task
  • An honest answer if there is no gain
  • Calculated on your data