Questions engineers
and managers ask us.

Kovelta builds manufacturing AI applications and delivers the implementation around them. The starting point is always the expertise your engineers already have: their specifications, measurement data, judgment calls and review habits. We turn that into software the team can run every day, with people keeping the final call.

Manufacturing teams that live with measurement data and engineering decisions: process engineering, quality, new product introduction (NPI), equipment and technical service. The current public demonstration is built for a process engineer analysing measurement results, but the company scope is manufacturing AI software and implementation more broadly.

Kovelta Limited is based in Taiwan and works with manufacturers in Taiwan and internationally. Conversations and pilots can be run in English or Traditional Chinese.

No. The design principle behind every Kovelta application is that people keep judgment, set the standard, handle exceptions and own the result. The software does the collecting, comparing, tracing and reporting so engineers spend their time on decisions rather than on moving data between systems.

The Process Engineering Workspace at demo.kovelta.com is a working measurement-analysis workbench. You can import CSV data, adjust specification limits, switch between box plots and histograms, trace an out-of-spec reading back to a region of interest on a reference image, record an engineering review and export an HTML report. It runs in the browser and needs no account.

No. Every reading, image and drawing in the demo is independently generated synthetic data, clearly labelled as such. No customer or former-employer data is used. Bring your own CSV if you want to see the workflow on your numbers; it stays in your browser.

Not yet. The current demo is deterministic statistics and workflow: linear-interpolated quartiles, overall standard deviation, Ppk against the limits you set, and a rule-based assistant. There is no large language model, AutoML, automatic root-cause analysis or MES connection in it. We say plainly which parts are implemented and which are illustrative, and the AI capabilities are what pilots are scoped around.

English by default, Traditional Chinese and Vietnamese. Switching language keeps your data, limits, selected readings and notes. The interface language is also used for the exported report.

In the demo, no. Imported files are parsed and analysed locally; refreshing the page clears the session and nothing is uploaded. In a pilot, data handling, storage and access are agreed with you before any data is shared.

With one line, one measurement item and data you already have. We work through real cases with your engineers to make the evidence, decisions and exceptions explicit, agree acceptance criteria up front, test the workflow in your environment, and only then decide where the next application belongs.

Existing measurement exports (CSV is enough to begin), the specification limits you actually use, the person who currently makes the call on that data, and a rough picture of the current process from measurement to sign-off. No MES or ERP integration is required to start.

Kovelta does not publish list prices. A pilot is scoped and priced around one defined workflow; ongoing software and implementation follow from what the pilot proves. Ask for a demo and we will walk through the scope with you.

Use the contact page or email owen@kovelta.ai. Requests for a demo get a reply from a person, not an automated sequence.