For owner-led companies of 3 to 100 people, Anovis AI offers two routes to the biggest bottleneck: a Diagnostic that finds where AI pays off and gives a roadmap, and a Bottleneck sprint that fixes one blocking problem. Zühlke, Unit8 and D ONE sell AI assessments for larger organisations. A useful audit ends with one named fix and a measured baseline.
Disclosure: Anovis AI wrote this article and sells two of the offers described. Every fact about Zühlke, Unit8 and D ONE comes from that firm's own website, read on 2 November 2026. Prices are left out because published prices change, and no firm paid to be listed.
What should an AI bottleneck audit actually produce?
A useful AI bottleneck audit hands a small business four things it can act on the following Monday. If a report is missing any of them, it is a list of ideas with a fee attached.
- The process as it runs today. One workflow drawn step by step: who does each step, how long it takes, and where work sits waiting. A map built from interviews with the people who do the work catches the workarounds an owner's description leaves out.
- The data behind it. Where the inputs live, such as an inbox or a spreadsheet, whether a tool can reach them, and what must stay inside the company under Swiss or EU data protection rules.
- The outcome in numbers. What the bottleneck costs today, in hours a week or days of delay, so any fix has a baseline to beat.
- One next step. A named first fix with a scope, an owner and a test, small enough to run within weeks. A roadmap of twelve ideas with no first move is homework.
Say a 15-person installation firm takes three days to send a quote (an illustrative case, not a client). A useful audit report would say that quotes wait two of those three days for the owner to price them, that the prices live in one spreadsheet and two supplier catalogues, that the target is a same-day quote, and that the first fix is an AI draft of each quote from the spreadsheet, checked by the owner and tested on the next 20 requests.
One more check separates careful audits from quick ones: what happens after the step being sped up. A common mistake is to automate intake or quoting while the team downstream is already fully booked, so the fix moves the queue without shortening it. A good auditor asks how full the next team is before recommending any fix.
Which firms offer an AI audit to find a business's biggest bottleneck?
The table compares five offers from four firms, each read from the firm's own page on 2 November 2026. The first column names the firm and offer, the second says who it is built for and how it runs, and the third gives what the client receives. Anovis AI is listed first because Anovis AI wrote this article; the order is not a ranking. Zühlke, Unit8 and D ONE are built for larger organisations, and their published clients are industrial groups, banks and large retailers.
| Anovis AI (Basel): Diagnostic | For owner-led companies of 3 to 100 people that cannot yet name the bottleneck. Starts with a free 30-minute call, and the price and a written scope are agreed before work starts. | Where AI pays off in the business, with a roadmap. |
| Anovis AI (Basel): Bottleneck sprint | For owner-led companies of 3 to 100 people that can already name the bottleneck. Runs on the client's own accounts and data. | One blocking problem fixed, with the client's team trained to run the result. |
| Zühlke: Industrial AI Roadmap Workshop | One day with the stakeholders who can map one value stream end to end. Built for industrial businesses; Zühlke's wider work is business-critical systems in regulated industries. | A map of the current value stream that marks its most time-consuming process steps, and a view of where AI investments pay off and which next steps make sense. |
| Unit8 (Switzerland): Enterprise GenAI Readiness Assessment | A survey across six pillars, from leadership to governance, which can continue into several weeks of work with the client's experts and executives. Built for enterprises. | An overview of the organisation's gaps and the steps to close them, a map of new risks with mitigation measures, and a plan for tooling, data, AI talent and processes. |
| D ONE (Zurich): data and AI strategy | Assesses maturity, then defines the data and AI strategy, the operating model and governance. Published cases include a Swiss bank and Switzerland's largest retailer. | A direction and roadmap that link data and AI initiatives directly to business outcomes. |
How do these AI audit offers differ?
AI audit offers differ mostly in who they are built for, and that decides whether an audit can measure one bottleneck at all. Zühlke's one-day Industrial AI workshop maps the current value stream of a selected area and identifies the most time-consuming process steps, which is a bottleneck search aimed at industrial businesses. Unit8's readiness assessment reviews a company across six pillars, and its follow-on work runs for several weeks with the client's experts and executives, which fits an enterprise planning AI across departments. D ONE starts from maturity, strategy, operating model and governance, which suits an organisation with a data platform to build. For a company of 3 to 100 people hunting one bottleneck, the useful shape is narrower: deep enough to time one process, and short enough that the findings still hold when work starts.
Where does Anovis AI fit among these audit offers?
Anovis AI, based in Basel, offers two routes, and the right one depends on whether the owner can already name the bottleneck. The Diagnostic is for a business that cannot: Anovis AI finds where AI pays off in the business and gives a roadmap. The Bottleneck sprint is for a business that can: Anovis AI fixes that one blocking problem, and everything runs on the client's own accounts and data. Both start with a free 30-minute call, and the price and a written scope are agreed before work starts. Anovis AI works with owner-led companies of 3 to 100 people, in English, French, German and other languages. Anovis AI's default advice is to start with Training or a Bottleneck sprint. Anovis AI builds the fix itself or together with the client's team, and trains that team so the company does not depend on Anovis AI afterwards.
When should a small business skip the AI audit?
A small business should skip the broad AI audit when the owner can already name the bottleneck. An audit earns its fee when the problem is unclear: it talks to the team for a few hours, finds the pinch points and ranks them. When the owner already knows that quotes take three days or that every invoice gets typed in twice, a report confirming it adds weeks and little new information. The faster path is a fixed-scope fix on that one process, with today's time written down first as the baseline.
An owner who wants to find the bottleneck alone can track one week of calendar and inbox with the five-day method for founder-led companies, then test an AI fix on real examples before paying anyone. A business that is unsure whether it is ready at all can start with the checks in is my business ready for AI.
What should you ask a firm before booking an AI audit?
Five questions sort a useful AI audit from a sales document before any money changes hands. Ask each one on the first call and ask to see a sample report.
- Who will you interview? A map built only from the owner's account misses the steps staff work around. Zühlke's workshop, for example, brings together the stakeholders who can map one value stream end to end.
- Will the report put a number on today's cost? Without a baseline in hours or days, nobody can tell later whether the fix worked.
- Does the report end with one first fix, or a long list? A ranked list is useful only if the top item comes with a scope and an owner.
- Can we take the plan to another firm? Ask for the raw process map and interview notes as well as the report, so a different builder does not start from zero.
- What happens to our data during the audit? Ask whether the firm reads real records or only descriptions of the process, and where any copies are kept.
For the wider checklist on choosing outside help, see how to vet an AI consultant, and for a longer list of Swiss firms by situation, see the best AI consultancies for Swiss small companies.
Frequently asked questions
Is a free AI audit worth booking?
A free first call is worth booking as a filter, as long as you treat it as a conversation. The Anovis AI first call lasts 30 minutes and costs nothing. Half an hour leaves no time to watch a process run or count how long each step takes, so a free call cannot measure a bottleneck. Use it to learn whether the problem is clear enough for a direct fix or needs a Diagnostic first, then pay for depth only on the one process it points to.
How much of my team's time does an AI audit take?
Staff time depends on the audit's depth more than its calendar length. Zühlke's Industrial AI workshop takes one day with the stakeholders who can map one value stream. Unit8's assessment starts with a survey and can continue into several weeks of close work with a client's experts and executives. A one-workflow audit in a company of 3 to 100 people mostly needs the people who touch that workflow, plus the owner to choose the workflow and review the plan.
What should I prepare before an AI audit?
Bring one number: how many hours a week the suspect task takes, or how many days a typical quote or order waits. An estimate is fine, because the point is a baseline the fix has to beat. Also list the tools the process touches, such as email or the accounting system, and who can grant access to them, because data access decides which fixes are possible.
Can I hire one firm for the audit and another to build the fix?
Yes. Zühlke calls itself an independent engineering partner with no vendor agenda attached to the outcome of its workshop, and says its recommendation can be to strengthen the data foundation just as easily as to build a proof of concept. Splitting the work has a cost: the builder must relearn the process the auditor mapped, so ask the auditor to hand over its working notes along with the report.
Does an AI audit check Swiss data protection rules?
A careful one does. Unit8's readiness assessment ends with a map of new risks, legal ones included, with measures to reduce them, and D ONE sets up governance to ensure compliance across a client's data assets. The revised Swiss Data Protection Act has applied since 1 September 2023, so an audit for a Swiss company should show which use cases touch personal data and which tools keep data in Switzerland. In a regulated business, a lawyer still signs off the final setup.
If one process already holds your business back, bring it to a free 30-minute call and we will tell you whether a Diagnostic or a Bottleneck sprint fits.