Verification as a Service

We are building the epistemic infrastructure for trustworthy AI
Today's AI systems are remarkably good at finding and generating information, but they still struggle with a fundamental problem: determining what should be trusted, how confident they should be, how to handle conflicting sources, and how to explain the reasoning behind an answer.

Our goal is to build the ontology, knowledge graph, and verification layer that sits between information retrieval and AI-generated outputs.
Rather than simply retrieving documents, this layer enables claims to be connected to evidence, provenance, confidence levels, and alternative interpretations.

We are using environmental sustainability as our first proving ground, because it combines some of the most complex regulatory frameworks, scientific evidence, standards, and reporting requirements globally.
If verification can work in a domain with evolving regulations, multiple standards, uncertainty, and competing interpretations, the same infrastructure can later be applied to AI governance, policy analysis, compliance, investment due diligence, scientific knowledge, and other high-impact domains

Our long-term vision is to make verification a standard component of AI systems, helping models reason from evidence rather than simply generating plausible answers. We believe trustworthy AI will require not only access to information, but access to mechanisms that evaluate the quality and reliability of that information.

Who we are

A team of eight volunteers bringing together expertise in software engineering, emerging technologies, ethics and business development.
Collectively, we bring:

• 20+ years of experience in emerging and cutting-edge technologies\
  • 20+ years experience in cutting -edge tech\
  • 40+ years of software development
  • 10+ years of technology startup experience
    As a founder, this is the third technology developed in the last 10 years from scratch. As a volunteer-led project, funding will be directed primarily towards infrastructure, data acquisition, standards access, human expert validation and platform/API development rather then salaries.

Where we are now

Over the past year, we have developed and validated the foundations of the approach, including:

• A machine-readable ontology and semantic framework

• A structured knowledge graph

• An initial verification engine

• More than 1 million structured knowledge 'units' ingested from authoritative sources

  This work has demonstrated that complex claims can be mapped to supporting evidence and evaluated through a transparent, scalable verification process

What's the output we work for

During the grant period, we will:

• Expand EU source coverage from approximately 35 global authoritative sources to 100-120 sources
• Grow the knowledge graph from 1M to 7-10 Million structured knowledge units

-Extend ontology coverage across at least five industries

- Integrate extra EU authoritative datasets, regulations, standards and scientific sources covering at least South and North American jurisdictions
-Generate source attribution, confidence assessments and contradiction detection for evaluated claims
-Develop an MVP that allows users to upload reports and receive evidence-based verification outputs
-Release a production-ready API for integration with AI systems, RAG architectures, AI agents, and automated reporting workflows

How our work reduces AI risks

We believe one of the major challenges for advanced AI is not access to information, but the ability to evaluate information.
Current AI systems can retrieve documents and generate convincing responses, yet they often lack robust mechanisms for assessing evidence quality, handling uncertainty, identifying contradictions, or explaining why a claim should be trusted.
As AI becomes increasingly embedded in decision-making, these limitations create risks ranging from hallucinations and misinformation to poor governance and flawed policy decisions.
Our hypothesis is that the future trustworthy AI systems will require a shared epistemic layer that makes authoritative knowledge machine readable, traceable and verifiable.
By connecting claims and statements to evidence, provenance, confidence and uncertainty, VaaS aims to help both humans and AI systems make better-informed judgments.

We see environmental sustainability as the first proof of this approach, but the infrastructure is being designed to scale across domains wherever trustworthy reasoning, evidence assessment, and transparent decision-making are required.