MiViA
MiViA is a browser-based AI tool that analyses metallographic micrographs. It measures grain size, retained austenite, ferrite, cracks, decarburisation and layer thickness in seconds, giving metallography labs objective, reproducible results that no longer depend on the operator.
What is MiViA?
MiViA is an artificial intelligence tool for metallographic microstructure analysis, published by MiViA GmbH in Freiberg, Saxony. A laboratory uploads a micrograph and receives a quantified, reproducible measurement in seconds, in place of a manual evaluation that takes minutes to hours and varies with whoever performed it. The company frames the problem it addresses in three parts: manual analyses are subjective, they consume valuable working time, and qualified metallographers are increasingly hard to recruit.
Seven analysis modules are available today: crack detection, grain size, grain size with twins, retained austenite, ferrite content and distribution, edge decarburisation, and layer thickness. Two more are announced but not yet usable — a bainite module described as still in training, and an inclusion rating module at feasibility stage. Requests for further modules are invited by email.
The product runs entirely in the browser and is presented as a progressive web app, so nothing has to be installed. Results come out as vector-based PDF reports that the company says are legible even to non-specialists, and as CSV exports for building up data over time. An OpenAPI connection allows a microscope to feed captured images directly into the software, and API access is listed as included in all three pricing tiers, although no public documentation for it exists.
MiViA GmbH describes itself as a 2021 spin-off from the research Dr. Grzegorz Korpala began at TU Freiberg in 2012 on industrial process monitoring. Its own timeline records roughly €1 million of EXIST funding, a beta phase in 2022, the MiViA 1.0 release in 2023 and a team of ten by 2024. The customer base is industrial and named publicly: VDM Metals, KRW Leipzig, CTB Trattamenti Termici, Vulkan INOX, Waelzholz and others, drawn from steel production, heat treatment, bearing manufacturing and automotive supply. Published case studies carry figures rather than testimonials — roughly one minute per sample at KRW, analyses fifteen times faster at CTB, and one study claiming 3,000 microstructures processed in a single day.
What it does
- Upload a micrograph and receive a quantified measurement in under a minute
- Measure grain size, including in samples containing twins
- Quantify retained austenite in steel samples
- Detect cracks and measure their length and depth
- Determine decarburisation depth with an automatically placed measurement line
- Measure coating and layer thickness along the whole section
- Export results as vector-based PDF reports or as CSV for longitudinal analysis
When to use MiViA / When not to
A quick filter to help you decide if MiViA is the right fit.
When to use MiViA
- In-house metallography laboratories running a steady daily sample load, where manual evaluation has become the bottleneck
- Quality managers who must document analyses against tightening customer requirements and want results that hold up to scrutiny
- Heat treatment and steel producers measuring grain size, retained austenite, ferrite or decarburisation depth as routine controls
- Labs short of experienced metallographers, where analysis quality currently depends on who happens to be at the microscope
- Materials research groups and university labs that need reproducible, statistically comparable microstructure measurements
When not to use MiViA
- Anyone working outside metals: no module covers polymers, ceramics, composites or biological samples
- Labs needing bainite classification or inclusion rating today, since both modules are still announced rather than available
- Buyers who require published terms and conditions or a licence agreement before committing, as the site publishes none
- Teams wanting to build an integration immediately, because the API is sold in every plan but has no public documentation
- Occasional users hoping for instant self-service access, since the free trial is unlocked only after a scheduled setup call
How to use MiViA
A typical end-to-end flow, from setup to results.
- Book an introductory call of about 30 minutes, which the company presents as a mutual fit assessment rather than a sales pitch
- Attend the account setup appointment, which is mandatory before trial access is granted and covers the technical requirements for uploaded images
- Receive access to the free trial: one month, 20 analyses, full functionality, no credit card required
- Prepare and microscope your samples exactly as you do today, with no change to sample preparation
- Upload the micrograph through the browser interface, with nothing to install locally
- Select the analysis module matching the measurement you need, such as grain size or retained austenite
- Review the quantified result and the measurement overlay the AI has placed on the image
- Export the outcome as a PDF report for documentation or as CSV for longitudinal data collection
- Optionally connect your microscope through OpenAPI so captured images arrive directly in the software
- Move to a paid plan and complete the individual module configuration and onboarding provided with it
Pros & Cons
Pros
- Reproducible results that do not depend on the operator, confirmed in writing by several named industrial customers
- Documented speed gains from real deployments: around one minute per sample at KRW, fifteen times faster at CTB
- Seven modules covering the routine metallographic measurements on steels, rather than a single narrow function
- Prices published openly for all three tiers, which is uncommon in this corner of the industrial software market
- Nothing to install: the browser is enough, and the vendor makes a point of avoiding lengthy internal IT approval
- Substantial human support, with three months of onboarding on the Basic and Enterprise tiers
- German publisher with European hosting, a named data protection officer and a full GDPR privacy policy
Cons
- No terms and conditions, licence agreement or dedicated contact page is published anywhere on the site
- The API is billed as included in all three tiers but has no public documentation, so an integration cannot be scoped in advance
- The free trial is not self-service: a scheduled setup call must take place before access is granted
- Monthly billing costs roughly twice the annual rate, which makes the annual commitment effectively mandatory
- The Essential tier is capped at 200 analyses and excludes user management and panorama images
- The privacy policy covers the website only and says nothing about how uploaded micrographs are handled
- No published position on whether customer data trains the models, and no documented way to opt out
Pricing & Plans
There is no permanent free plan. A free trial is offered, limited to one month and 20 analyses, with full functionality and no credit card required, but access is granted only after a scheduled setup call. Three paid tiers are published. The cheapest recurring entry point is the Basic tier at €499 per month with annual billing, rising to €1,000 per month if billed monthly. A lower absolute entry point exists in the Essential tier, a one-time payment of €999 covering 200 analyses on a single workstation. The Enterprise tier costs €3,499 per month with annual billing, or €7,000 per month billed monthly. Prices are identical on the German and English pages.
- Essential — €999
- one-time payment
- 200 analyses
- 1 workstation
- all modules
- API included
- one module configured
- 3 hours of guidance plus support
- Basic — €499 per month billed annually (€1
- 000 per month billed monthly)
- unlimited analyses
- 1 workstation
- three modules configured
- panorama images
- 3 months of onboarding
- Enterprise — €3
- 499 per month billed annually (€7
- 000 per month billed monthly)
- unlimited analyses
- 10 workstations
- all modules configured
- user management
- panorama images
Data, GDPR & hosting
A consolidated view of how MiViA handles your data.
GDPR overview
GDPR implementation is concrete and above average for a company of this size. MiViA GmbH is established in Germany, so the regulation applies directly and no Article 27 representative is needed or named. The privacy policy names a data protection officer, Dr.-Ing. Grzegorz Korpala, and enumerates the full set of data subject rights: withdrawal of consent, objection under Article 21, complaint to a supervisory authority, portability, access, rectification, erasure and restriction. Transfers outside the EU, which arise through Calendly, LinkedIn, Stripe and Google, are covered by the European Commission's standard contractual clauses. SSL/TLS encryption is stated. The important caveat is scope: this policy governs the website. It does not describe how micrographs uploaded to the analysis platform are processed.
Who owns the data?
The published documentation does not settle this. MiViA GmbH names itself as the data controller for the website and has appointed a data protection officer, Dr.-Ing. Grzegorz Korpala, reachable at info@mivia.ai. However, the privacy policy covers the marketing website, not the analysis platform, and the company publishes no terms and conditions or licence agreement anywhere on the site. There is therefore no public clause stating who owns the micrographs a laboratory uploads, nor what MiViA may do with them. Prospective customers should ask for this in writing before uploading production images, since nothing on the public site answers it.
Reuse rights
For the website, MiViA states that it processes personal data to perform the contract (Art. 6(1)(b) GDPR), on legitimate interest grounds (Art. 6(1)(f)) or with consent, and that customer and contract data are collected to establish and manage the commercial relationship. Data are passed to third parties only where the contract requires it, for example to the institution handling payment, and the policy states explicitly that data will not be shared for advertising without express consent. Processors are engaged under data processing contracts. What the documentation does not address is the customer's side: there is no published statement on whether a laboratory may reuse, export or redistribute the analysis output beyond the PDF and CSV reports the product generates, and no licence terms exist to define it.
Data retention & training
Hosting summary
The jurisdiction is German. MiViA GmbH, based in Freiberg, Saxony, is the stated controller and names OVH GmbH as the host of its website, giving the provider's corporate address in Cologne, Germany. That address identifies the supplier, not the location of the servers, and no hosting country or region is stated anywhere for customer data. Independently, the domain's A record resolves to an address geolocated in Dunkirk, France, on OVH's network, so the website at least is served from French infrastructure. The two facts are not contradictory, but neither establishes where uploaded micrographs are stored. The policy adds that the host processes data only as far as necessary to meet its obligations and on MiViA's instructions, that SSL/TLS encryption is in place, and that transfers to the United States arising through Calendly, LinkedIn, Stripe and Google rest on the European Commission's standard contractual clauses. Anyone with data residency requirements should obtain a written statement of the storage location.
Things to keep in mind
Risks and trade-offs to weigh before adopting MiViA.
- The free trial is gated behind a scheduled Calendly appointment, so budget for a delay rather than expecting instant access
- The API is sold in every tier but undocumented publicly: confirm endpoints, authentication and limits in writing before planning an integration
- No terms and conditions are published, so the contractual framework cannot be reviewed before committing
- Monthly billing is roughly double the annual rate, so a short-term or trial-length commitment is disproportionately expensive
- The privacy policy covers the website, not the analysis platform: ask explicitly what happens to uploaded micrographs and who may access them
- Treating an automated measurement as unquestionable is the classic trap with tools like this; keep enough in-house metallographic judgement to recognise when a result is wrong
- No support email address is published, and the visible address is presented for prospective rather than existing customers, so agree a support channel as part of onboarding
Setup & Integrations
Technical difficulty
Low on paper, moderate in practice. The software runs in a browser with nothing to install, and the vendor stresses that no lengthy internal IT approval is required. Against that, the models impose specific requirements on uploaded images — enough that the company makes a setup appointment mandatory before trial access, to walk through them. Module configuration is carried out by the vendor, from one module to all of them depending on the tier, and onboarding runs to three months on Basic and Enterprise. Connecting a microscope through OpenAPI is optional and would need vendor assistance, since no documentation is published.
Deployment
Behind MiViA
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What can MiViA actually measure?
Do I need to install any software?
Is there a free trial, and how do I get it?
Is there a permanent free plan?
What does MiViA cost?
Does MiViA offer an API?
Can I connect my own microscope?
Where is my data hosted?
Are my images used to train the models?
Are there published terms and conditions?
Should you pick MiViA?
MiViA is a narrow, serious tool that does one difficult thing well. For a metallography laboratory measuring grain size, retained austenite, ferrite or layer thickness day in and day out, the case is strong and unusually well evidenced: seven working modules, named industrial customers willing to be quoted, and published figures rather than vague promises. The reproducibility argument is the real one. Removing operator variance from a measurement that has always depended on experience is worth more to a quality department than the speed gain, even though the speed gain is what the marketing leads with.
What holds the tool back is documentation rather than capability. There are no published terms and conditions, no licence agreement and no contact page. The API is billed as included in every tier yet has no public documentation, which makes it impossible to scope an integration before signing. The privacy policy is thorough but governs the website, leaving the question of what happens to uploaded micrographs unanswered — including whether they train the models, on which the company takes no published position. For a laboratory handling proprietary alloy work, that is not a small gap.
The commercial terms deserve attention too. Monthly billing costs roughly twice the annual rate, so the annual commitment is effectively compulsory, and the €999 Essential tier is a capped one-time purchase rather than a cheap subscription. The trial is genuine but gated behind a scheduled call.
The practical conclusion: worth a serious look if you run a metals lab with real volume, provided you use the mandatory setup call to get written answers on data ownership, model training and API access. Those answers are not on the website, and they should be before anything proprietary is uploaded.
- Choosing a selection results in a full page refresh.
- Opens in a new window.