[ 01 ]
Workflow map
The workflows you named, drawn as they run today: steps, systems, hand-offs and who touches what.
[ AI WORKFLOW AUDIT ]
Find out where AI pays off before you spend on it.
Two weeks, a fixed price and a report you can act on with us, with someone else or on your own. If nothing is worth automating, the audit says so and saves you the licences.
2 weeksFixed scope, fixed priceYours even if you never build
[ 01 — WHAT YOU KEEP ]
What you keep, even if you never build with us.
Five results you hold in your hands after the readout. None of them depends on hiring us again.
- [ 02 ]
Baseline per workflow
Time, volume and error rate, measured or sampled. This is the number every later result is compared with, whoever builds it. - [ 03 ]
Ranked candidates
Each candidate gets hours saved, effort, risk and data sensitivity in one line, so the order is defensible in front of management. - [ 04 ]
Architecture and data-risk notes
For the top candidates: which model type, which provider or local option, where the data has to stay and what your security and compliance teams will ask. Written so they can review it. - [ 05 ]
A go or no-go
A scoped 4-week pilot with owner, guardrails and number to beat, or an honest note that nothing is worth automating yet.
[ 02 — NO STRINGS ]
Why this is the low-risk first step.
Most AI spend starts with a licence or a build. This one starts with a measurement.
- [ 01 ]
Fixed price, fixed scope
Agreed before we start. No open hours and no surprises on the invoice. - [ 02 ]
Nothing to install, nothing to buy
We only read. No licences, no new platform, no changes to your systems. - [ 03 ]
The report is yours
Take it to management, to another provider or into your own project. There is no exclusivity and no obligation to continue. - [ 04 ]
A no is a result
If nothing is worth automating yet, we say so. That is cheaper than a pilot that fails.
[ 03 — PROCESS ]
Two weeks, four steps.
You spend a few hours of the right people's time. We do the measuring, the analysis and the writing.
- 01
Scope
A 30-minute call to agree which teams and workflows are in, which systems we may read and who decides at the end.Day 1 - 02
Interview and observe
We talk to the people who run the workflow and watch a normal week's work, not the process on paper.Week 1 - 03
Baseline and rank
We time the workflow as it runs today, sample real cases and rank every candidate.Week 2 - 04
Readout
A working session with management: the ranked list, the baseline and a recommendation for one pilot.End of week 2
[ 04 — YOUR SIDE ]
Who should be in the room, and how much time it takes.
Plan for a few hours of your people's time in total, spread over two weeks. We do the rest.
- [ 01 ]
The people who run the workflow
One interview each, roughly 1 hour, plus short checks while we watch normal work and review samples. They know where the process really breaks. - [ 02 ]
One person with system access
Someone in IT who can grant read access to the systems involved, for about 1 hour at the start. Access can be withdrawn at any time. - [ 03 ]
A decision-maker
The person who owns the budget or the process, for the scoping call and the readout. They leave with a decision, not a homework list. - [ 04 ]
IT, security or compliance
Optional, but useful: one conversation early on lets us write the data-risk notes in the language they need.
[ 05 — SCOPE & PRICE ]
Fixed scope, agreed before we start.
AI workflow audit · 2 weeks
CHF 4,900fixed scope · fixed price · excl. VAT
Excl. VAT. About 4 consultant days. The scope is confirmed in the intro call before you commit.
For the price of about 4 consultant days you get a measured baseline and a ranked decision instead of an opinion. Compare that with the alternative: buying licences or commissioning a build on a guess.
- Interviews with the people who run the workflow
- Baseline measurement per workflow
- Ranked list with architecture and data-risk notes
- Go or no-go and pilot recommendation, presented to management
Not included
- Building agents: that is the pilot
- Licences and new platforms
- Changes to your systems: we only read
[ 06 — QUESTIONS ]
Before the audit.
Do we need Copilot licences or any other AI tool first?
No. The audit comes before any licence or build decision, and it is not tied to a vendor.
What data do you look at, and does it leave our environment?
Only what the workflow touches, with read access you grant and can withdraw. Where possible we work on samples, and working files stay in your environment. We do not put your data into public AI tools, and we agree any use of a model in writing before we start.
Can we sign an NDA first?
Yes. Ask for it when you get in touch and we sign before you share details.
What if nothing is worth automating?
Then the audit says so. That is a result too, and it saves you the licences and the build.
What comes after the audit?
Optionally one pilot: one workflow, four weeks, a baseline and a number to beat, agreed before we start. You can also take the report and do it yourself or with another provider.
Our IT team is cautious about AI. Is that a problem?
No, it helps. Bring them in. Access is read-only, and the data-risk notes are written for security and compliance to review.
What do we get that we could not do ourselves?
An outside baseline, a consistent way to rank candidates and the view of people who build these workflows, so the risk notes are realistic. Teams that do this alone often measure too late or rank by enthusiasm.
Start with the measurement.
30 minutes to agree the scope. Two weeks later you have a ranked list and a decision.