All analysis was generated autonomously, without human review. Scores are analytical opinions drawn from the cited public sources, without hands-on testing. They are not audits, certifications, investment reports, purchasing advice, or evaluations of quality.
Holistic AI sells enterprises software that finds the AI systems they run and tests those systems for bias, safety and security. The software maps the results to rules such as the EU AI Act and NYC Local Law 144, a bias-audit rule. The company also audits other organizations' AI systems as an outside reviewer. Founded in 2020, it has taken investment from Mozilla Ventures. It reports that Gartner named it a Challenger in a 2026 ranking of AI governance platforms. Its named customers include Unilever, MAPFRE, Starling Bank, Allegis and the Wikimedia Foundation. The foundation publishes the audit reports Holistic AI completed for it in 2024 and 2025 under the EU Digital Services Act. That outside-audit work is the part of its position a rival would take longest to reach.
| Description | Holistic AI's governance platform lets enterprises discover, assess, and govern AI systems from shadow AI to agentic workflows, identifying risk and bias and enforcing regulatory compliance. | [f1] |
|---|---|---|
| Founded | 2020 | [f2] |
| HQ | San Francisco, United States | [f3] |
| Deployment | SaaS | [f4] |
| Product | What it does |
|---|---|
| Holistic AI | AI governance platform that discovers AI systems including shadow AI, audits models and LLMs for bias, robustness, and data leakage, and enforces regulatory compliance across the AI lifecycle. |
| Endlayer | Device-level control plane that surfaces the AI tools, agents, models and MCP servers people use, monitors what each reaches and costs, and governs them against a versioned policy. |
AI Defense Matrix
| Govern | Identify | Protect | Detect | Respond | Recover | |
|---|---|---|---|---|---|---|
| AI-Workload Platforms Inference servers, training platforms, vector DB platforms, and the model-loading supply chain. | ||||||
| AI Coding and Orchestration Tools AI coding tools and agentic orchestration tools on user devices, plus their plugins, skills, hooks, system prompts, scaffolding, harnesses, configuration settings, and MCP clients. | ||||||
| AI-Generated Code Code produced by AI tools, AI-assisted reviews, AI-generated infrastructure-as-code and tests, and vibe-coded apps that bypass CI/CD. | ||||||
| AI Traffic Paths to external and self-hosted AI services, MCP tool channels, agent-to-agent calls, model-registry downloads, and egress to unapproved AI services. AI gateways, LLM routers, and MCP gateways steer traffic along those paths. The steering decisions and their inputs belong here too: routing policies, MCP server registries, and agent naming and discovery services. | ||||||
| AI Model Model weights, fine-tuning checkpoints, model cards, registries, an AI bill of materials (AIBOM), and the third-party LLMs your enterprise consumes. | ||||||
| Training Data Datasets used for training, fine-tuning, and continued learning. | ||||||
| Runtime AI Data User prompts, inference inputs, RAG content, vector DB content, persistent agent memory, and interaction history. | ||||||
| AI Agent Identities AI agents as non-human principals, plus credentials, keys, permission scopes, service accounts, and delegation chains across agents and tools. |
Holistic AI's governance platform discovers AI systems and audits models and LLMs for bias, robustness and data leakage, and its Endlayer device agent inventories AI tools, coding agents and MCP servers and blocks unsafe prompts and secrets. It is mapped to the AI Defense Matrix. [f5]
How well the company can compete in its security market, scored across eight dimensions against public evidence.
| Dimension | Score | Rationale |
|---|---|---|
| Problem Clarity How precisely the company defines its problem, with evidence the problem exists at the scale claimed. | 5/5 | The problem is established outside the vendor. TechTarget's independent 2026 survey says AI governance has become a boardroom priority, that compliance is becoming enforceable, and that AI governance platforms are becoming a defined market. The Wikimedia Foundation, speaking as a buyer, states that its platform must be audited annually by an independent auditor under the Digital Services Act and that Holistic AI completed that audit. A named category, a regulatory mandate and a buyer describing the same problem in practice meet rung 5. [s10, s12, s1] |
| Capability Depth How specific the technical capabilities are, with evidence beyond marketing claims such as docs and third-party validation. | 4/5 | The platform page sets out an architecture, naming read-only discovery across more than fifteen integrations, automated bias, robustness, efficacy, privacy and transparency testing, agentic red teaming, an Agent Graph over agents, tools, tasks and data flows, and Sentinel and Operative agents that monitor and intervene inline in production. The Endlayer page carries the same detail for its device agent. Validation from outside is the Wikimedia Foundation's own record of two Digital Services Act audits Holistic AI delivered, the 2024 one rated Positive with comments. [s2, s14, s12, s8] |
| Market Timing Whether the market is ready for this product, with evidence that buyers are actively seeking solutions. | 4/5 | Two kinds of buyer-side signal are documented and dated. Analyst category, June 2026: Gartner built a Magic Quadrant for AI Governance Platforms and evaluated thirteen vendors, reported by Holistic AI, with TechTarget's independent 2026 survey describing the same category. Regulatory driver, 2025: the Wikimedia Foundation's own compliance page states that a very large online platform is audited annually by an independent auditor and names the auditor it engaged, which is a buyer acting on a rule rather than a vendor describing one. Both are independent of each other in kind. [s5, s10, s12] |
| Team Credibility Demonstrated domain expertise with public signals such as prior exits, publications, and industry recognition. | 4/5 | Both co-CEOs appear as authors on a 2024 Royal Society Open Science paper on algorithm auditing that Semantic Scholar indexes with 69 citations, which is publication standing on an independent record rather than a company claim. NIST lists Holistic AI among the members of its AI Consortium. The AiThority interview adds more than 70 papers for Adriano Koshiyama and an advisory seat on the OECD AI and Global Partnership on AI group. That is plurality with independent evidence. The cited record names no prior startup exit for either founder. [s11, s9, s6] |
| GTM Proof Evidence of actual traction (customers, revenue signals, partnerships) beyond stated intentions. | 4/5 | The customers page names Allegis, the Wikimedia Foundation, Unilever, MAPFRE, Starling Bank, MindBridge, Bryq and Hired, and the Allegis case study quotes a senior IT leader on the platform. One of those names is corroborated away from the company's pages: the Wikimedia Foundation records that Holistic AI completed its Digital Services Act audit reports for 2024 and 2025. Neither revenue nor a customer count is disclosed in the cited record, so the size of that traction stays unreported. [s3, s4, s12] |
| Funding Efficiency Whether funding matches go-to-market ambition, with signs of capital-efficient growth. | 2/5 | Mozilla Ventures announced an investment in March 2024 and the cited record carries no amount, no round label and no later raise. The output side is documented in the same record: a platform with named enterprise customers, a second product in Endlayer, and statutory audit work the Wikimedia Foundation confirms. Evidence that documents output and stays silent on capital is one-sided. Whether the money is proportional to the enterprise motion cannot be judged, because the cited record carries no raise size to judge it against. [s7, s14, s12] |
| Category Clarity Whether the company creates or fits a recognizable category that buyers can quickly place in their stack. | 4/5 | TechTarget places Holistic AI inside a category it calls dedicated AI governance platforms, next to IBM watsonx.governance, ServiceNow AI Control Tower, Credo AI, OneTrust AI Governance, ModelOp, Cranium AI and LatticeFlow AI, with no vendor coaching in the framing. Gartner built a Magic Quadrant for the same category in 2026 and placed Holistic AI in it as a Challenger, which the company reports. Gartner placed it as a Challenger rather than a Leader, so the record does not show it defining the category. [s10, s5] |
| Incumbent Defensibility How vulnerable the core value proposition is to absorption as a feature by a platform vendor. | 3/5 | TechTarget's survey shows the absorption risk directly, listing GRC and privacy vendors and cloud and AI platform vendors that embed governance alongside the dedicated platforms. What resists that is the statutory audit work: the Digital Services Act requires an independent auditor, and the Wikimedia Foundation appointed Holistic AI twice. The cited record does not show that role converting into lock-in on the software, so the friction is real without being something a bundled rival could not eventually match. [s10, s12, s8] |
Holistic AI sells governance software to enterprise governance, risk, compliance, engineering, business-owner and security teams. Its platform page addresses governance leads, risk and compliance staff, data and ML engineers, business owners, and information security and third-party risk teams. The product sets out to discover every AI system an enterprise runs, including shadow AI, test those systems for risk and bias, and produce the compliance evidence those teams have to show.
The pain the company describes is tied to named regimes. The platform assesses systems against the EU AI Act, the NIST AI Risk Management Framework, ISO 42001 and NYC Local Law 144. TechTarget, writing about the 2026 market, calls inventory, policy mapping and audit evidence relatively mature while runtime controls, shadow AI discovery and agent governance develop quickly and stay inconsistent across vendors.
Governments treat this work as a discipline with its own techniques. The UK Department for Science, Innovation and Technology publishes a Holistic AI case study in its Portfolio of AI Assurance Techniques, under headings that include compliance audit, bias audit and conformity assessment. The entry is a case study Holistic AI supplied, so what the portfolio establishes is the company's inclusion in a government reference rather than an outside review of its method. [s2, s10, s8]
The platform runs as three modules. Identify discovers and inventories models, agents, APIs and pipelines across cloud, code and vendors through more than fifteen integrations the company describes as read-only with no agents to install. Protect runs automated testing for bias, robustness, efficacy, privacy and transparency, adds agentic red teaming for jailbreak, toxicity, hallucination and counterfactual bias, and maps agents, tools, tasks and data flows end to end through what the company calls an Agent Graph. Enforce turns policy into workflows and produces full audit trails, version history and on-demand reports.
Two agent types work at production time. Sentinel Agents monitor what a customer's AI says in production and catch prompt injection, jailbreaks, data leakage, hallucination and toxicity without interrupting workflows. Operative Agents act inline once risk crosses a defined threshold, governing which tools an AI system calls, what it can access and how much it can spend.
Holistic AI also offers Endlayer for devices. Endlayer surfaces every AI tool, agent, model and MCP server people use, monitors what each one reaches and costs, and governs all of it against a versioned policy. It installs as one unprivileged agent through fleet tooling the customer already runs, reads what the device user can already read, and follows the same four setup steps whether the estate is one laptop or fifty thousand.
Validation from outside the company covers the assessment work. The Wikimedia Foundation publishes that Holistic AI completed its 2024 Digital Services Act audit report and its 2025 report, and records the 2024 report's overall rating as Positive with comments. Holistic AI separately reports that Gartner named it a Challenger in the inaugural 2026 Magic Quadrant for AI Governance Platforms and ranked it first for the AI Risk and Compliance use case at 3.90 out of 5.0. A reader sees that ranking through the company's own announcement. [s2, s14, s12, s5]
TechTarget's 2026 survey of AI governance tools puts Holistic AI in a category it calls dedicated AI governance platforms, alongside IBM watsonx.governance, ServiceNow AI Control Tower, Truyo, Credo AI, OneTrust AI Governance, Monitaur, Airia, ModelOp, Saidot, Cranium AI, Relyance AI, Trustible, LatticeFlow AI, Modulos and Lumenova AI. The same piece names four further vendor categories whose products overlap, among them GRC, privacy and data governance vendors and cloud and AI platform vendors that embed governance. That is an outside account of both the slot and the crowding.
Statutory audit is the part of the business a governance module inside another vendor's stack does not answer, because the rule names an independent auditor rather than a tool. The Wikimedia Foundation's compliance page states that a very large online platform is audited annually by an independent auditor, and it names Holistic AI as its auditor for two successive compliance periods. The UK government's assurance portfolio describes the same practice in four steps and ends at an assurance step in which the company's team declares that a system conforms to predetermined standards, practices or regulations.
Standing in the standards community is not what separates Holistic AI here. NIST's AI Consortium member list carries Holistic AI, and the same list carries Cranium AI and Credo AI. [s10, s12, s8, s9]
The customers page names Allegis, the Wikimedia Foundation, Bryq, MindBridge, Starling Bank, Hired, Unilever and MAPFRE. The Allegis case study quotes a senior IT leader at the company saying the governance platform gave Allegis Group a unified, scalable solution for managing AI risk and regulatory compliance, and describes better visibility into AI usage across the enterprise.
One reference is corroborated away from Holistic AI's own pages. The Wikimedia Foundation's compliance archive records that Holistic AI completed its 2024 Digital Services Act audit report and its 2025 report, and states that the 2024 report's overall rating of the foundation's compliance was Positive with comments. That is a customer speaking for itself about a repeat engagement.
Mozilla Ventures published its own investment announcement. An interview with co-CEO Adriano Koshiyama names Allegis, Mapfre, Michelin, Unilever and Wikimedia. Mozilla Ventures announced an investment in March 2024 without an amount. Holistic AI reports its own Gartner Challenger placement. No revenue figure and no customer count appear anywhere in the cited record. [s3, s4, s12, s6, s7]
The founders' research record is verifiable without going through the company. Semantic Scholar indexes Towards algorithm auditing, a 2024 Royal Society Open Science paper carrying 69 citations, and lists both co-CEOs, A. Koshiyama and Emre Kazim, among its authors. Algorithm auditing is the method the product applies, so the publication sits on the company's own subject rather than beside it.
The AiThority interview adds scale to that record. It says Adriano Koshiyama has published more than 70 papers in international conferences and journals and is responsible for the early research on AI auditing, assurance and risk management, and that he is an active advisory member on the AI Risks and Accountability group at the OECD AI and the Global Partnership on AI. The AiThority interview states those figures.
Participation in the standards process appears on an independent record. NIST lists Holistic AI among the members of its AI Consortium. What the cited record does not carry is a prior startup exit for either founder, so a buyer weighing the team has a published research record to go on and nothing in the cited record about a prior company outcome. [s11, s6, s9]
The product is itself an assurance instrument, so its own readiness matters twice. The platform produces full audit trails, version history and on-demand reports, and assesses customer systems against the EU AI Act, the NIST AI Risk Management Framework, ISO 42001 and NYC Local Law 144. On its platform page Holistic AI describes the product as a SOC 2 platform.
Because the software reads across an enterprise's cloud, code and SaaS estate, a security review will ask what it touches. Part of the answer sits in the design the company publishes. The Identify module's integrations are described as read-only with no agents to install, while Endlayer does place an agent on the device and is described as unprivileged, reading what the device user can already read and never elevating.
Separately from that access question, part of the company's assurance work is on public record because customers and governments publish it. The Wikimedia Foundation names Holistic AI as the auditor of its 2024 and 2025 Digital Services Act reports and states that the 2024 report rated the foundation's own compliance Positive with comments, which is Holistic AI's opinion on Wikimedia rather than Wikimedia's opinion on Holistic AI. The UK Department for Science, Innovation and Technology publishes the company's general audit method in its assurance portfolio. [s2, s14, s12, s8]
| Company | Relationship | Note | Compare |
|---|---|---|---|
| Credo AI | competes with | TechTarget's 2026 survey lists both inside its dedicated AI governance platform category, so the two meet on the same enterprise governance shortlist. | |
| Cranium AI | competes with | Named in the same TechTarget category list, and both appear on the NIST AI Consortium member roster, so they compete for the same governance buyer. | N/AWe captured the evidence for these companies under different evidence-model versions (v1 vs v2), so the totals were scored under different conditions and are not directly comparable. |
| TrojAI | competes with | Competes on the testing and risk-assessment side of the same buyer's governance programme. | N/AWe captured the evidence for these companies under different evidence-model versions (v1 vs v2), so the totals were scored under different conditions and are not directly comparable. |
| Arthur AI | competes with | Contests the same enterprise AI oversight buyer. | |
| LatticeFlow AI | competes with | TechTarget names it in the same dedicated AI governance platform category. | N/AWe captured the evidence for these companies under different evidence-model versions (v1 vs v2), so the totals were scored under different conditions and are not directly comparable. |
Add analyzed competitors to compare them side by side with Holistic AI.
A closer look at the company's product strategy, measuring how defensible it is against market forces and examining the eight areas behind it.
press the advantage
Founded in 2020, Holistic AI sells AI governance software to enterprises and also performs third-party AI audits. Mozilla Ventures invested in Holistic AI. The software discovers a customer's AI systems, including unapproved ones, and tests them for bias, safety and security. It maps the results to the EU AI Act, the NIST AI Risk Management Framework, ISO 42001 and NYC Local Law 144. A second product, Endlayer, governs AI use on employee devices. Its customer list names Allegis, the Wikimedia Foundation, Unilever, MAPFRE, Starling Bank, MindBridge, Bryq and Hired. The Wikimedia Foundation records on its own pages that Holistic AI completed its Digital Services Act audits for 2024 and 2025. That independent-auditor role is the part of the position a rival would take longest to reach.
| Dimension | Score | Rationale |
|---|---|---|
| Value Delivery Does the product sell software as the product, or judgment, trust, or accountability with software as the delivery mechanism. | 2/3 | Alongside the platform, Holistic AI performs audit and assessment engagements the customer sought externally: the Wikimedia Foundation's Digital Services Act audits for 2024 and 2025, the MindBridge audit, Bryq and Hired bias-audit work, and the Starling Bank assessment. The UK government's portfolio entry, a case study the company supplied, describes an assurance step in which its team declares that a system conforms to predetermined standards, practices or regulations. Its product pages sell the software as the purchase, so it sells software and expertise together rather than judgment alone. |
| Switching Cost How expensive leaving is for a customer: data portability, integrations, learned workflows, network effects, regulatory data residency. | 2/3 | The record documents a system-of-record mechanism: one live registry of models, agents, datasets and endpoints, plus full audit trails, version history and on-demand reports that accumulate while a governance programme runs. What the record does not do is size the exit. It states no migration duration, no parties and no non-portability, and it describes the discovery integrations as read-only with no agents to install, so the switching mechanism is documented and the cited record does not size the migration. |
| Compliance Moat Whether certifications, liability acceptance, or audit trails block an easy replacement. | 2/3 | The Wikimedia Foundation confirms that platforms under the Digital Services Act are audited annually by an independent auditor, with reports submitted to the European Commission, and that Holistic AI completed its audits. That is regulator-facing audit work beyond routine attestation. It stops short of rung 3 because the Act requires an independent auditor rather than this one, and the record shows no authorization or retained liability a qualified replacement could not also take on. |
| Problem Complexity Whether the product requires ML, optimization, real-time systems, or years of specialized expertise. | 3/3 | The record documents real-time work rather than forms and dashboards. Operative Agents act inline once risk crosses a threshold, governing which tools an AI system calls, what it can access and how much it can spend, while Sentinel Agents watch production output for prompt injection, jailbreaks and data leakage. Endlayer runs a 3 MB unprivileged device agent across a fleet the page sizes at up to fifty thousand machines. A 2024 Royal Society Open Science paper on algorithm auditing lists both co-CEOs as authors. |
| Buyer Profile Whether buyers are SMB operators, mid-market IT teams, or regulated enterprises and governments with procurement gates. | 3/3 | The named buyers include Starling Bank for a customer onboarding assessment, the Wikimedia Foundation for Digital Services Act audits, plus Allegis, Unilever and MAPFRE. Mozilla Ventures describes the customer base as Fortune 500 corporations, SMEs, governments and regulators. Procurement and legal review sit between that class of buyer and a replacement. |
| Layer Whether the product is an end-user application, a platform with application features, or infrastructure other applications depend on. | 2/3 | Holistic AI inventories and governs AI systems that belong to other teams and other vendors, connecting through read-only integrations with AWS, Azure, GitHub, Databricks and others. Endlayer adds a device agent that reads what the device user can already read. Both sit above the models, agents and applications they watch, and nothing in the record shows another application depending on Holistic AI to run. |
| Proprietary Data, Content, or IP Whether the product accumulates datasets, content licenses, or IP that a rival cannot recreate from scratch. | 1/3 | No cited source names a retained dataset, content licence or granted patent. The compliance mappings encode public regulation and the founders' algorithm-auditing research is published in an open-access journal, so what the record shows the company holding is method and published work rather than an accumulated asset. The cited pages do not state what the platform retains from a customer's audit evidence. |
Holistic AI sells to the enterprise team answerable to regulators and boards for the AI a company builds, buys and runs. Its platform page names governance leads, risk and compliance staff, data and ML engineers, business owners, and information security and third-party risk teams as the people who use the product. Mozilla Ventures describes Holistic AI as serving Fortune 500 corporations, SMEs, governments and regulators.
The named accounts fit that description. The customers page carries Allegis, the Wikimedia Foundation, Bryq, MindBridge, Starling Bank, Hired, Unilever and MAPFRE, spanning staffing, a non-profit under EU platform regulation, HR technology, financial audit software, UK banking, consumer goods and insurance.
Segment boundaries show in the two products. The governance platform addresses the AI an enterprise builds and deploys. Endlayer addresses the AI its people adopt on their own machines. Its page describes those tools as arriving one person and one laptop at a time.
The platform covers discovery, testing and enforcement rather than one point feature. Identify discovers and inventories models, agents, APIs and pipelines across cloud, code and vendors through more than fifteen read-only integrations. Protect runs more than forty specialised tests covering bias, safety, security and performance, adds agentic red teaming for jailbreak, toxicity, hallucination and counterfactual bias, and maps agents, tools, tasks and data flows through an Agent Graph. Enforce produces full audit trails, version history and on-demand reports.
Two agent types work at production time. Sentinel Agents monitor what a customer's AI says in production and catch prompt injection, jailbreaks, data leakage, hallucination and toxicity without interrupting workflows. Operative Agents act inline once risk crosses a defined threshold, governing which tools an AI system calls, what it can access and how much it can spend. The homepage adds deployment gates, approval workflows and kill switches under the same Guardian Agents heading.
Endlayer takes the same idea to the device. Its page describes one agent in three weights deployed to the devices that need them. The 3 MB Sensor inventories AI tools, local models, agents, extensions and MCP servers. The 30 MB Decoder terminates TLS on the device so a prompt can be blocked in real time. The 300 MB Guardian watches agents call each other and blocks a request on the machine. The page says nothing decrypted ever leaves the device.
The founders' research is on an independent record, though no cited source ties a published method to a shipped test. Semantic Scholar indexes Towards algorithm auditing, a 2024 Royal Society Open Science paper carrying 69 citations and listing both co-CEOs among its authors. The company's own case study, published by the UK government, says the audits are derived from extensive academic research and industry expertise.
The named customers are the core of the traction record. The customers page carries Allegis, the Wikimedia Foundation, Bryq, MindBridge, Starling Bank, Hired, Unilever and MAPFRE. The Allegis case study quotes a senior IT leader saying the governance platform gave Allegis Group a unified, scalable solution for managing AI risk and regulatory compliance.
One of those relationships is documented by the customer rather than by the vendor. The Wikimedia Foundation's compliance archive records that Holistic AI completed its 2024 Digital Services Act audit report and its 2025 report, and states that the 2024 report rated the foundation's compliance Positive with comments.
Analyst placement is the other new signal, and it reaches the reader through the company. Holistic AI reports that Gartner named it a Challenger in the inaugural 2026 Magic Quadrant for AI Governance Platforms, one of thirteen vendors evaluated. It also reports that Gartner ranked it first for the AI Risk and Compliance use case at 3.90 out of 5.0. TechTarget independently lists the company among dedicated AI governance platforms. Mozilla Ventures announced an investment in March 2024 and named no amount. No revenue figure and no customer count appear anywhere in the cited record.
No price appears on any page cited here, and the published route in is a demo request. That pattern fits the negotiated enterprise deals the buyer list implies, and it leaves a buyer without a public anchor for budgeting.
The unit of value reads as the governed AI estate rather than a seat. Identify counts models, agents, APIs and pipelines, Protect runs tests per system, and Endlayer charges its attention to devices, quoting a fleet range from one laptop to fifty thousand. Which of those the company actually meters is not stated in the cited pages, so treat the estate reading as inference rather than a published unit.
Holistic AI delivers a software platform alongside named audit and assessment engagements. The platform connects to a customer's environment through read-only integrations with AWS, Azure, GitHub, Databricks and others, discovers AI systems, tests them, and produces compliance evidence. The cited pages do not state its hosting model, its tenancy or who operates it day to day.
The engagements are the other half and they carry their own delivery shape. The UK government's portfolio entry sets out four steps, triage, assessment, mitigation and assurance, ending with the company's assurance team declaring that a system conforms to predetermined standards, practices or regulations. The Wikimedia Foundation's record separately shows statutory audit work running to a published outcome, naming Holistic AI as its auditor for two compliance periods.
Endlayer changes the installation story. Instead of connectors, a local unprivileged agent installs through fleet tooling the customer already runs, reads what the device user can already read and never elevates, with the same four setup steps at any fleet size.
The product is an assurance instrument, so its own trustworthiness is part of what a buyer is assessing. The platform produces full audit trails, version history and on-demand reports, and assesses customer systems against the EU AI Act, the NIST AI Risk Management Framework, ISO 42001 and NYC Local Law 144. On its platform page Holistic AI describes the product as a SOC 2 platform.
Access is the question a security review will reach first, because the software reads across cloud, code and SaaS. Part of the answer is in the published design: the Identify integrations are described as read-only with no agents to install, and the Endlayer agent is described as unprivileged, reading what the device user can already read and never elevating.
Two of the records a buyer can read come from parties other than the company. The Wikimedia Foundation names Holistic AI as the auditor of its 2024 and 2025 Digital Services Act reports and states that the 2024 report rated the foundation's compliance Positive with comments. The UK Department for Science, Innovation and Technology publishes a case study the company supplied describing its audit method. NIST lists Holistic AI among the members of its AI Consortium.
Holistic AI positions itself as a neutral governance layer over an enterprise's AI rather than a feature inside one vendor's stack. The homepage says the platform detects models, agents, APIs, pipelines and workflows across AWS, Azure, GitHub, Databricks and more than twenty integrations. The platform page says more than fifteen, so the company's pages differ on the exact integration count.
Neutrality is the claim and also the exposure. TechTarget's 2026 survey names cloud and AI platform vendors with embedded governance, and GRC, privacy and data governance vendors with embedded governance, as two of five categories whose products overlap this one. A customer already inside one of those stacks can reach for governance there.
Endlayer widens the surface in a different direction. Its page says desktop apps, coding agents and local models run on the machine and never touch the gateway, and it names MCP servers alongside them, which is the surface the device agent is built to reach.
The founding pair is research-rooted and the record is checkable outside the company. Semantic Scholar indexes Towards algorithm auditing, a 2024 Royal Society Open Science paper with 69 citations, listing A. Koshiyama and Emre Kazim, the two co-CEOs, among its authors. Algorithm auditing is the method the product applies, so that publication sits on the company's own subject.
The AiThority interview describes Adriano Koshiyama's publication and advisory record. It says Adriano Koshiyama has published more than 70 papers in international conferences and journals. It credits him with early research on AI auditing, assurance and risk management. It identifies him as an active advisory member on the AI Risks and Accountability group at the OECD AI and the Global Partnership on AI. It says he founded the company in 2020. That biography comes from the company, relayed by an interview publication.
Institutional participation is on a government record. NIST lists Holistic AI among the members of its AI Consortium, a list that also carries Cranium AI and Credo AI, so membership places the company in the standards conversation without separating it from its rivals. The cited record names no prior startup exit for either founder.
| Id | Source | Tier | Accessed |
|---|---|---|---|
| f1 | Holistic AI: An End-to-End AI Governance Platform | official | 2026-09-21 |
| f2 | AiThority interview with Adriano Koshiyama, co-founder and co-CEO of Holistic AI | research | 2026-09-21 |
| f3 | About Holistic AI | official | 2026-06-14 |
| f4 | AI Defense Matrix Catalog entry | other | 2026-06-09 |
| f5 | AI Defense Matrix Catalog mapping | other | 2026-06-23 |
This site is an experimental research aid created by Zeltser Security Corp. All its data gathering and analysis was performed autonomously without human review, and it can contain errors of fact, interpretation, and judgment that a human reviewer might catch.
The analyses are statements of opinion, not statements of fact. Machine analysis produced the scores, summaries, and matrix placements by weighing the public sources each page cites, and reasonable people can weigh the same sources differently. Where a page states a fact, it cites the public source and the date it was checked, and the statement is only as accurate as that source. Unless a profile expressly says otherwise, the analysis involves no hands-on testing and no independent validation of any company's products or services.
Nothing here is professional, security, legal, financial, investment, or purchasing advice, and nothing here is a recommendation to invest in, do business with, or avoid any company. Inclusion of a company is not an endorsement, and absence of a company is not a judgment about it. Reading this site creates no advisory or client relationship. Verify any detail you plan to act on against the vendor's current materials.
The content is provided "as is" and "as available," with all warranties disclaimed, express or implied, including merchantability, fitness for a particular purpose, accuracy, and non-infringement. No entry is warranted to be complete, current, or correct. Companies change, vendors update their claims, sources can be wrong, and automated analysis can misread them.
To the fullest extent permitted by law, the operator, Zeltser Security Corp, is not liable for any damages that arise from using this site or relying on its content, including direct, indirect, incidental, special, and consequential damages and lost profits, even if advised that such damages were possible. If you are dissatisfied with the site or disagree with these terms, your remedy is to stop using it.
Entries link to vendor pages, press coverage, and other external sites that Zeltser Security Corp does not control and is not responsible for. A link is not an affiliation with the destination or an endorsement of it. Product and company names and trademarks are the property of their owners, used here nominatively to identify the companies described. Short quotations from cited sources appear for identification and commentary.
Use, quotation, automated retrieval, and redistribution of the content are governed by the Terms of Use at cybercompanyprofiles.com/terms, which permit personal and internal business use with attribution and prohibit republication and resale.