Modulos governs the AI systems inside your organisation. Grecta governs the products you place on the EU market.

Modulos is an enterprise AI governance and risk platform, founded in Switzerland out of EPFL. It registers AI systems with an owner and a lifecycle stage, prices risk in euros, maps each control to every framework that requires it, runs runtime tests against production metrics, tracks regulatory change on a weekly cycle with a dated change log, and attaches evidence to the claims it supports for audit. It covers the EU AI Act, ISO/IEC 42001, the NIST AI RMF and a range of further AI, security and sectoral frameworks.

Grecta is product safety compliance infrastructure. It turns EU product law into engineering tickets, release gates and a public verification page your buyers can check.

The short answer

Modulos is the better choice if your problem is governing an AI estate. It is a capable platform, it is strong on the EU AI Act, and on several dimensions it is more mature than we are.

Grecta is the better choice if your problem is market access for a product. The published framework coverage for Modulos is AI-specific and IT-specific. It does not include the product legislation that determines whether a connected product may lawfully be placed on the EU market, and its own platform page does not claim to support conformity assessment.

If your product is a medical device, a machine, a radio product or anything else carrying a CE mark alongside its AI, that gap is the whole decision.

At a glance

GrectaModulos
What it governsThe product you place on the EU market.The AI systems and agents your organisation builds, buys and runs.
Unit of valueThe product, with its intended purpose and classification.The AI system in the registry, with an owner and a lifecycle stage.
The core questionMay this product ship, and what has to be true first?What AI do we run, what is it worth in risk, and are the controls met?
Main outputA settled classification and assessment route, tickets in the backlog, a release cleared or blocked, a public verification page.An AI registry, priced risk, control status mapped across frameworks, audit-ready evidence exports.
Primary buyerHead of product, VP engineering, CTO, regulatory affairs.CIO or chief AI officer, the person who answers for AI in the organisation.
Regulatory focus26 EU product frameworks, article by article.AI frameworks and adjacent IT frameworks. Broader than ours on AI and security, narrower on product law.
How answers are reachedDeterministic resolution against a clause map. Same product in, same obligations out, with the clause behind every answer.Risk quantification, control mapping and runtime testing, with people making the decisions.
Where it actsIn the release pipeline. Jira and Linear tickets, pre-flight release gates, the verification page.In the AI lifecycle and the control register. Evidence and runtime integrations across the stack.
Conformity assessmentSettled, including the Article 43 route and whether a notified body is required.Not claimed on their platform page.

The difference that matters: the AI estate against the product on the market

Europe now regulates software the way it regulates machinery. The obligation attaches to the product placed on the market, under every regime that covers that product, not only the AI one.

Modulos answers a governance question well. It tells the person accountable for AI what systems exist, what each is worth in risk, whether the controls are satisfied, and whether the evidence would survive an audit. For an organisation running a large AI estate, including shadow AI and agents, that is a real job and Modulos is built for it.

Conformity is a different question. It asks whether this product, with this intended purpose and this classification, may lawfully be placed on the market, which assessment route applies, what the technical file must contain, and what has to be true before the next release ships. It is settled against the law article by article rather than scored against a control set, and the answer changes when the product changes or the law changes.

Why a product is rarely only an AI problem

Take an AI-enabled patient monitor. It is a medical device under the MDR, so its AI Act high-risk obligations arrive through Annex I and apply from 2 August 2028, assessed by a notified body inside the existing MDR procedure. It is a product with digital elements, so the Cyber Resilience Act reaches it unless the MDR exclusion in Article 2 removes it. It is software, so Directive (EU) 2024/2853 on liability for defective products applies. Its clinical data engages the GDPR.

A registry entry showing green controls against an AI framework does not tell you whether that device can be CE marked, which notified body scope you need, or that the AI Act date for this product is 2028 rather than 2027. Those are product law questions.

Regulatory coverage: Grecta vs Modulos

Based on each vendor’s published framework coverage as at November 2026.

RegimeGrectaModulos
EU AI ActYes, article by article per product, each obligation carrying its application dateYes
ISO/IEC 42001YesYes, with audit readiness scoring
NIST AI RMFYesYes
GDPR, NIS2, DORAYesYes
ISO/IEC 27001 and 27701NoYes
OWASP Top 10 for LLM and agentic applicationsNoYes
Non-EU AI frameworks such as MAS FEAT, UAE AI Ethics, Singapore agentic AINoYes
EN 18286:2026, the first harmonised standard cited under the AI ActYes, clause levelNot named in published coverage
Cyber Resilience ActYesNot named in published coverage
MDR and IVDRYesNot named in published coverage
Product Liability DirectiveYesNot named in published coverage
Machinery RegulationYesNot named in published coverage
Data ActYesNot named in published coverage

Modulos covers more AI, security and non-EU frameworks than we do. We cover the EU product legislation it does not. The dividing line is not breadth, it is whether the regime governs an organisation’s AI or a product placed on a market.

How each one handles the EU AI Act

GrectaModulos
Banned practices, Article 5Gated at intake. A product that trips Article 5 is stopped before scoping continues.Addressed through the control set and risk classification.
Risk classification, Article 6Settled deterministically per product, including which Annex limb applies and therefore which date governs.Risk class is set, and regulatory scope is set by role and risk class.
Legal role, Articles 3 and 25Provider, deployer, importer or distributor settled at intake, including where fine-tuning makes you a provider.Role is captured and used to scope requirements.
Quality management, Article 17 and EN 18286QMS blueprint mapped to EN 18286 clauses, copied into Jira or Linear as tickets against the article that requires them.Control set and evidence. EN 18286 not named in published coverage.
Technical documentation, Annex IVGenerated from the product record, retained for the period each regime sets.Keeps EU AI Act technical documentation current.
Conformity assessment route, Article 43Settled, including whether internal control under Annex VI is available or a notified body under Annex VII is required.Not claimed on their platform page.
Regulatory changeRecomputed per product. Two events force a recompute: you ship, or the law changes. Earlier determinations are preserved with their dates.Scheduled weekly updates with a dated change log, with people making the decisions.
Release decisionsEach release cleared or blocked before it ships.Runtime tests against production metrics. A failed test reopens the control it proves.

Grecta and Modulos: What each platform actually does

This table is where the two products overlap most, and we have marked it honestly.

CapabilityGrectaModulos
Organisation-wide AI inventory and shadow AI discoveryNoYes
Agent inventory and agent testingNoYes
Runtime testing against production metricsNoYes, with Prometheus, Datadog, Snowflake and others
Control mapped once across every framework that requires itYes, the AI Act’s 188 obligations resolve to 68 controlsYes
Risk priced in eurosYes, with a CFO one-pagerYes, with a quantification history
Regulatory change tracking with a dated logYes, recomputed per productYes, weekly with a change log
Evidence attached to the claim it supports, exportable for auditYesYes
ML and data stack integrationsNoYes, Python SDK, MLflow, Jupyter, REST API, AWS, Azure
Enterprise deployment options and certificationsNot yet. SaaS onlyYes, SaaS, private cloud or VPC, SOC 2 and ISO 27001 certified
Per-product obligation resolution across EU product lawYes, 26 frameworks article by articleNot within published framework coverage
Conformity assessment route and notified body scopeYesNot claimed
Engineering backlog integration as tickets written against the articleYes, Jira and LinearJira is integrated for evidence rather than for obligation tickets
Pre-flight release gate before shippingYes, cleared or blocked in sixty secondsNo
Pre-code change impactYes, a planned feature tested before the code is writtenNo
Public verification page for buyersYes, a live page at a public URL with a 90-day freshness clockNo
Authorised representative, Article 22 of the AI ActYes, through European Compliance Suite in EstoniaNo

Grecta vs Modulos: Pricing

GrectaModulos
ModelPer productSubscription, by quote
Published pricingYesNot published on their website
EntryFree intake and scoping, no cardFree trial reported by third-party listings
Pilot1,500 euro per product, two weeksNot published
PlatformFrom EUR2,400 euro a yearThird-party listings report subscriptions from around USD 20,000 a year. Confirm with the vendor.
BillingAnnual subscriptionAnnual subscription
What the price depends onNumber of products and frameworks in scopeScale of the AI estate and modules taken

Grecta prices per product because the obligation attaches to the product. Adding fifty systems to a registry does not change what the Cyber Resilience Act requires of your gateway. Adding a second product does.

Which one should you choose?

Choose Modulos if

  • You run an AI estate rather than a product line
  • Shadow AI, agents and internal model sprawl are the problem
  • You want runtime testing tied back to control status
  • You need ISO/IEC 42001 and ISO/IEC 27001 alongside the AI Act
  • You need private cloud or VPC deployment and vendor security certifications today
  • Your frameworks extend beyond the EU, to Singapore, the UAE or Monetary Authority of Singapore guidance

Choose Grecta if

  • You place a digital or connected product on the EU market
  • Your product triggers product law alongside the AI Act, such as the CRA, the MDR, the IVDR or the Machinery Regulation
  • You need the conformity assessment route and the notified body question settled
  • You need to know which Annex limb governs your product, and therefore whether your date is 2027 or 2028
  • You want obligations in the engineering backlog and a gate on the release, not a control dashboard
  • You need a verification page a buyer can open instead of a questionnaire
  • You need an authorised representative established in the Union

Use both if

  • You are an enterprise that both runs a large internal AI estate and ships CE-marked products
  • Modulos answers to your CIO, Grecta answers to your notified body and your buyers
  • Estate governance and product conformity are separate obligations and neither discharges the other

Frequently asked questions

Modulos also maps one control across several frameworks. Is that not the same as your obligation compression?

The mechanism is similar, but the difference is the input. Modulos maps controls across AI and IT frameworks. We resolve obligations across EU product legislation, so the compression happens between the AI Act, the CRA, the MDR and the rest, which is where the duplication actually bites for a manufacturer. If your frameworks are ISO 42001, ISO 27001 and the AI Act, their mapping covers it.

Modulos also prices risk in euros. Why is yours different?

It is not materially different as a capability. Ours is computed from the obligations that fired for a specific product and the penalty exposure attached to them, and it is delivered as a one-pager for a CFO. Theirs is computed across the AI estate with a quantification history. Both are legitimate. If risk quantification is your main requirement, test both.

Modulos also tracks regulatory change weekly. What do you do differently?

The difference is what gets recomputed. A change to the law re-resolves each affected product’s obligations and flags the products, and the earlier determination is preserved with its date so you can show what the position was when you assessed it. That matters when a notified body or a market surveillance authority asks what you assessed and against what version of the law.

Why does product law matter if we already cover the EU AI Act?

Because the AI Act rarely arrives alone. An AI-enabled medical device sits under the MDR, the AI Act, the Cyber Resilience Act and the Product Liability Directive at once, with four different timetables, and the AI Act obligations for that device arrive through Annex I on 2 August 2028 rather than through Annex III on 2 December 2027. Getting that wrong is a year and a notified body.

Is Modulos more mature than Grecta?

On several dimensions, yes. They have been building since 2019, they hold SOC 2 and ISO 27001, they offer private cloud and VPC deployment, and they integrate across the ML and observability stack. We are earlier, our paid tier opens in January 2027, and we are SaaS only. Where we are ahead is product law coverage and conformity, and that is a difference of scope, not maturity.

What does product level actually mean?

Each product has its own record. Obligations are resolved for that product’s intended purpose and classification, the gaps are written into that product’s backlog, releases of that product are gated, and that product gets its own verification page. A company with four products has four records, four sets of obligations and four timetables.

How does Grecta reach its answers? Is it a language model?

The resolution is deterministic. A product’s attributes are resolved against a clause map, article by article, and the clause behind every answer is shown. That is why the same input returns the same output.

Does Grecta do model testing, bias evaluation or robustness testing?

No. We establish that Article 10 data governance, Article 15 accuracy and robustness and the related obligations apply to a given product, what evidence each requires and when it is due, and write that work into the backlog. The testing itself belongs in your ML stack or a specialist tool, and Modulos is strong at that part.

Does Grecta give legal advice?

No. Grecta is regulatory and conformity analysis. Our founder is a Ukrainian qualified lawyer and is not admitted to practise in the United Kingdom or the European Union, which is why the work is positioned as conformity analysis rather than legal advice.

What does it cost to find out?

Initial scoping is free and needs no card. Bring one product and we will settle its classification and legal role and show which obligations reach it with their application dates. Pilots are 1,500 euro per product.

Modulos is a trademark of its respective owner. Grecta is not affiliated with, endorsed by or sponsored by Modulos. Comparative statements describe each product’s intended purpose and publicly published capabilities as at November 2026 and are liable to change as either product develops. Where this page records that a regime is not named in published coverage, or that a capability is not claimed, that reflects the vendor’s own published material and is not a statement that the product could not address it. Pricing attributed to third-party listings is not confirmed by the vendor. This page is regulatory and conformity analysis and is not legal advice.