ZynoviqWe find where your AI fails — before your customer does.
About 100 synthetic cases on one workflow you choose, written and certified by qualified professionals. Your AI is evaluated against them over about four weeks, for one fixed price, quoted. You receive every failure with the rule behind it, an executive report and a regression set you keep.
Request a diagnostic using the form below, or see the four weeks step by step.
Between a free look and a full assessment
The free 25-case sample test tells you whether there is a problem. A capability assessment of 300–500 cases maps a whole domain. The diagnostic sits between them: deep enough to show the pattern of failures on the workflow that matters most to you, small enough to decide on quickly.
- About 100 cases, in one domain, on one workflow.
- About four weeks from the workflow decision to your report.
- One fixed price, quoted after a scoping call and agreed before work starts.
Domains you can choose
| Domain | Status | Example workflow |
|---|---|---|
| GST and Indian indirect tax | Live | Input tax credit eligibility, blocked credits, reverse charge, place of supply |
| Payroll under the labour codes | Active — on request | Wage definitions, statutory deductions, payroll questions from HR |
| Company law | Active — on request | Compliance questions, filings and board processes |
For payroll and company law we confirm on the scoping call that we have qualified professionals for your workflow before we quote.
Who it is for
- Teams whose AI already answers or acts on Indian tax, payroll or company-law work — in a product, an internal copilot or an agent inside an enterprise system.
- Teams that have tried the free sample, or seen enough of their AI's mistakes, and need evidence before a release, a model change or a budget decision.
- Product and enterprise AI teams that want a first regression set without committing to a full benchmark.
It is not for organisations only adopting off-the-shelf AI tools. Who we serve.
What you receive
- A limited set of expert cases — about 100, built for your chosen workflow and certified through our quality chain.
- An evaluation of your AI against those cases: structured fields marked by rule, free-text reasoning marked by professionals.
- A failure analysis — every failure with the correct answer, the rule behind it and how serious it is.
- An executive report and a walkthrough call. See a sample report.
- A reusable regression set — the cases your AI failed, ready to re-run after every change.
No answer is written or certified by AI. Marking today is rule-based and human; AI-assisted marking of free text is planned, opt-in and always checked by people.
The four-week plan
- Week 1 — you choose the workflow
On a scoping call we agree the workflow, the domain, how questions reach your AI in a test environment and the fixed price.
- Weeks 1–2 — we build the expert test set
Qualified professionals write about 100 synthetic cases, with expected outcomes and rationale.
- Week 2 — independent review and adjudication
Each case is solved independently and reviewed; disagreements are adjudicated into one certified answer.
- Weeks 2–3 — your AI is evaluated
We run the cases through your AI and mark every response.
- Week 3 — failures analysed
Each failure is traced to the rule it breaks and given a severity.
- Weeks 3–4 — you receive the results
Benchmark, failure report, scores, recommendations and a reusable regression suite, with a walkthrough call.
Timings are typical and are fixed in the written scope you approve. The four-week diagnostic, week by week, on the packages page.
How it compares
| Free sample test | AI Quality Diagnostic | Capability assessment | |
|---|---|---|---|
| Cases | 25 | About 100 | 300–500 |
| Domains | GST | GST (live); payroll and company law on request | One domain per assessment |
| Scope | Standard GST topics | One workflow you choose | The whole topic map of a domain |
| Timeline | About two weeks | About four weeks | Four to six weeks |
| You receive | Score, three most serious mistakes, call | Every failure analysed, executive report, regression set | Full baseline across the topic map, every failure explained |
| Price | Free | Fixed — quoted | Fixed — quoted |
| Page | Sample test | This page | Capability assessment |
Would you like us to run this every time you change your AI?
Your AI will change: a new model, a new prompt, new knowledge, a new tool. The regression set from your diagnostic is the start of a permanent check that shows whether each change made it better or quietly worse. How regression testing works → · Regulatory Regression Service
AI Quality Diagnostic — FAQ
How is the diagnostic different from the free sample test?+
The sample is 25 GST cases with a short report and the three most serious mistakes. The diagnostic is about 100 cases on the workflow you choose, with every failure analysed, a full executive report and a regression set you keep.
How much does it cost?+
It is a fixed price, quoted after a short scoping call and agreed before work starts. The price does not change once the scope is approved. How we price.
Which domains can we choose?+
GST and Indian indirect tax are live. Payroll under the labour codes and company law are active on request — we confirm on the scoping call that we have qualified professionals for your workflow. Domains and status.
Do we need to share data?+
No. Every case is synthetic. Your AI is run in a test or sandbox environment, never production. How tests run, security and data handling.
Is the report a certification of our AI?+
No. It shows where your AI is right and wrong against professional ground truth. It is not a certification or assurance of your system.
What happens after the diagnostic?+
Nothing, unless you want more. Teams usually either widen the test into a capability assessment or re-run the regression set every time they change their AI. The diagnostic week by week, and the packages after it.