Work out what to do with AI, then do it
Training, strategy, governance and engineering, in one conversation.
Most AI advice stops at “you should use AI.” The questions that decide whether it works come after: which process it goes into, and who owns it once the consultants leave.
The first thirty minutes are free, and can end with us telling you not to hire us.
Most first calls start with one of these
Five practices, and almost nobody buys all five. Find the sentence you would use about this week, and read what the work behind it actually involves.
“Everyone says we should be doing something with AI. Nobody can say what.”
We look at how the work moves through your business and find the places AI changes the numbers. You get a short list, in order, with a cost and a reason against each one.
Strategy and transformation
- Leadership session, one day
- Transformation support, 3 to 12 months
“We bought the licences. People went back to working the old way.”
Tools rarely fail on their own. We build the training around the work your team already does, then redraw the roles and handovers the tools changed.
AI training and organisation
“The pilots went well. None of them shipped.”
A pilot usually skips the dull part: who owns it, who gets the data, and what happens when the model is wrong. We pick the two worth finishing, and finish them.
Bespoke AI build
“We are paying for a lot of AI and cannot tell what it returns.”
We total the licence and token spend, check who uses what, and put a number against each use. Then you can cancel the rest.
AI audit
“A customer sent us an AI questionnaire and nobody could answer it.”
We classify each system you run, close the gaps, and leave you the documentation. You answer the next questionnaire yourself.
Governance and compliance
Six systems we have built in defence, tunnels, clinics, autonomous fleets and recruiting, with what each one takes in and hands back.
Four steps, and you can stop after any of them
Most AI work stalls because nobody can say what good looks like for one specific product. We start from evidence about your own systems rather than a questionnaire about your intentions.
Find out what you have
You end up with: A written picture of where you stand
We go through the tools, the spend and the work itself, including the licences bought on somebody's card. Two to three weeks, and you can stop here.
Decide what is worth doing
You end up with: A short plan, in order, with numbers
One working session with your leadership team. We rank the options by what each costs against what it returns, and we name the ones to drop.
Build it
You end up with: A working system, and the record of how it was made
Where the answer is software, our engineers write it. Where it is policy, we write it with you. Either way the evidence trail goes in from the first day.
Hand it over
You end up with: A team that does not need us again
We train your people on their own work and redraw the roles the new tools changed. Then we leave, and the thing keeps running.
Questions people ask first
If yours is not here, send it over. One of us answers it, rather than routing you into a sales call.
We are not sure we even need help. Where do we start?
Take the free thirty-minute call. Tell us how the work moves through your business and where it jams, and we will tell you which parts AI would change and which it would not. A fair number of those calls end with us saying you are fine for now, and here are the two things to watch.
Is AIAuditSense a substitute for a real audit?
No. It points you at what to look at first. Classification under the AI Act depends on facts no automated tool can see from a repository or a website, so a defensible assessment involves people, documents and interviews. Where the legal exposure is real, engage counsel.
What does the training look like?
We start from your material rather than a syllabus. That means the documents your team works on, the tools they already have licences for, and the tasks in front of them this week. Sessions are cut to the role, because finance, legal, operations and engineering carry different risk and the law calibrates the duty to a person's function. Attendance and assessment are recorded as we go, which is what turns the training into evidence you can show a customer or an auditor. It runs as a half or full day workshop, or as a programme across cohorts over four to eight weeks.
Do you only work on compliance?
No. Compliance is one of five practices, and the one fewest engagements start with. Most open with the question of what to build at all, or with an audit of what you already run, and go where the findings point. That is usually spend, adoption or how the team is organised, well before it is law.
Do you do the implementation, or only the advice?
Both. Our engineers and coding agents build the systems, and the controls go in during development rather than after. If you already have a development team, we work alongside it and hand over the documentation.
We are outside the EU. Does the EU AI Act reach us?
Often, yes. The Act covers providers and deployers established outside the Union where the output of the system is used inside it. A product with European customers can be in scope with no EU entity and no EU staff. Whether it applies, and in which role, turns on who puts the system on the market, whose name is on it, and where the output lands.
What does it cost?
The first call costs nothing. Audits and diagnostics are fixed price, quoted after that call, so the first step never carries open-ended risk. What follows depends on how many systems you run, how many people you employ and which sector you are in.
How small is too small?
We have worked with teams of fifteen. Below that, a workshop and a short audit tend to cover it and a full engagement would be poor value for you. We will say so rather than sell you one.
You do not have to talk to us to find out
AIAuditSense is the audit engine we built for our own consulting work. Point it at a website, a repository or a product description. It classifies the system, scores each governance dimension, and hands back a list of what to fix with the clause references attached. Beside this is a page of one it produced.
It takes about two minutes, asks for no sign-up, and costs nothing. If what comes back is worth an argument, that is what the call is for.
Or call +91 98732 63331. It reaches a director, not a sales desk.
AIAuditSense helps you get oriented. It is not legal advice. Classification under the AI Act turns on facts nobody can see from code or copy alone, so bring in qualified counsel before you rely on any assessment.
Candidate screening and job-fit platform
High risk · Art. 6 / Annex III(4), employment and worker management
16out of 100
14out of 100
14out of 100
Weakest dimensions, of twelve assessed
- Technical documentation and record-keeping5 out of 100Non-compliant
- Quality management and conformity5 out of 100Non-compliant
- Registration and post-market monitoring5 out of 100Non-compliant
- Risk management system8 out of 100Non-compliant
- Accuracy, robustness and cybersecurity10 out of 100Non-compliant
What to fix first
Document the Annex III classification
Write down why the product falls under employment and worker management, why the narrow-task derogation does not apply, and hand the memo to every enterprise customer during onboarding.
Art. 6 · Annex III(4) · effort: medium
Get a legal opinion on emotion inference
Establish whether the interview analysis infers a candidate's emotional state. If it does, remove that output from employment contexts, because workplace emotion recognition is a prohibited practice.
Art. 5(1)(f) · effort: medium
A page from a real report, with the assessed company removed. The full version scores twelve dimensions against each framework and carries the evidence it read.