Holistic AI Engineering Team - May 2026
About Holistic AI
Holistic AI is the leading enterprise AI governance platform, founded at University College London - the same institution behind foundational research on retrieval-augmented generation and a long lineage of modern AI work - by Dr. Adriano Koshiyama and Dr. Emre Kazim.
Based in London, the company helps global enterprises discover, test, and monitor their AI systems across the full lifecycle. The platform covers everything from shadow AI detection and bias testing to EU AI Act compliance and automated policy enforcement.
Recognised as an AI governance market leader by IDC in 2025, Gartner Cool Vendor for AI Security and a Major Contender by Everest Group, Holistic AI serves organisations that need to govern AI at scale while keeping pace with deployment speed.
The Challenge
Enterprise AI governance unlocks organisation wide AI adoption, but it must also stand up to regulators, auditors, and boards across thousands of AI systems and every jurisdiction a customer touches.
Frameworks like the EU AI Act, NIST AI RMF, and ISO 42001 each demand their own controls, with new guidance, amendments, and codes of practice landing almost every month.
Every recommendation the platform produces (a bias mitigation, a control mapping, a third-party model assessment) has to cite current peer reviewed research, regulatory text, and technical standards. Hundreds of times a day, per customer.
Manual curation can't keep up: hand-built corpora drift within a quarter, and LLM summaries hallucinate citations that don't exist. The wrong hallucination in a regulatory inquiry is a material legal problem.
Holistic AI needed an evidence layer that was live, audit-grade, and pluggable.
The Solution: Valyu as the evidence layer
Holistic AI integrated Valyu in days. The Search API covers academic literature, regulatory texts, technical standards, and the broader web in a single call, with source scoping configurable per call. It handles the high frequency, low latency lookups that sit behind everyday governance work.
For continuous, multi-step research, Holistic AI uses Valyu's DeepResearch API. It's the backbone of Holistic AI Atlas, a global horizon scanning tool that tracks new AI regulation, legislation, penalties, and incidents, linking every entry back to its primary source.
Together, Search and DeepResearch keep the platform's outputs anchored in current evidence. Governance outputs need to reflect what the field looks like today, not what a content team last uploaded six months ago. Every claim carries a citation a compliance team can defend.
"Building an AI governance platform turns out to be, in large part, the work of citing the world correctly. Frameworks cite laws. Assessments cite methods. Mitigations cite research. Reports cite all three."
- Kleyton da Costa, AI Engineer, Holistic AI
What Valyu Got Right
Several details made Valyu a good fit for a regulated industry product.
Source scoping is composable. Switching from a general web search to a peer reviewed only academic search is a one line change. Many APIs lock source scope at the contract or tier level.
Breadth of indexed content. Regulatory documents, peer reviewed papers, technical standards, and PDFs are all indexed and directly retrievable. Governance evidence lives in formats the open web doesn't surface cleanly - the team pulls from the source instead of relying on third party summaries.
Replayable transactions. Every search response carries a transaction ID that gets persisted alongside generated reports. When a regulator asks where a citation came from months later, the team can trace it back to the original retrieval.
Cost is predictable, not token metered. Search is priced per call; DeepResearch is flat per task, so a run costs the same whether it touches five sources or fifty. For a platform billing customers per assessment, that turns evidence retrieval into a known unit cost.
Results & Impact
Since integrating Valyu, Holistic AI has moved from manually curated evidence to live, research backed outputs across the platform.
Every recommendation is defensible. Governance outputs now come with traceable citations from peer reviewed and regulatory sources, giving customers an evidence trail they can present to auditors and regulators with confidence.
Evidence stays current. The team no longer maintains static corpora of PDFs. Valyu searches the latest research and regulatory content on demand, so recommendations reflect the current state of the field rather than whatever was last manually uploaded.
One integration, multiple features. What started as a single feature experiment now powers multiple core capabilities - from per-call citations on every governance recommendation through to Atlas, Holistic AI's global horizon scanning tool for new AI regulation and enforcement activity across jurisdictions.
"We expected to use it for one feature. We are now planning around it."
- Kleyton da Costa, AI Engineer, Holistic AI
Why Valyu
For Holistic AI, governing AI at scale is, fundamentally, an evidencing and citation problem. Every output the platform produces has to be backed by current, verifiable sources, and the volume of evidence grows with every customer and every new framework.
Valyu gave Holistic AI two complementary APIs against the same index: Search for high frequency lookups grounding individual recommendations, and DeepResearch for the continuous, multi-step research behind Atlas. Underneath both sit the infrastructure features a regulated industry product needs: transaction IDs, composable source scoping, and predictable per call and per task costs.
Both companies share roots at UCL: Valyu's search is built for research grade retrieval, Holistic AI's platform for research grade governance.
"If you are building anything where the answer has to reflect what the field looks like this week, not last quarter, Valyu is the layer we would recommend taking a hard look at."
- Emre Kazim, Co-founder/Co-CEO, Holistic AI
Holistic AI's platform helps enterprises govern, audit, and monitor their AI systems. Learn more at holisticai.com.
Want to try Valyu yourself?
Get started in our playground or dive into the documentation.
Questions? Reach us at contact@valyu.ai.