# Cyber Company Profiles: Advai

Source: [Cyber Company Profiles](https://cybercompanyprofiles.com)
Exported 2026-09-11
Analyzed 2026-08-20
Canonical: https://cybercompanyprofiles.com/companies/advai
License: free for personal use and internal business purposes, including internal commercial evaluation such as assessing a vendor for procurement, with quoting permitted when attributed to cybercompanyprofiles.com. No resale, republication, redistribution as a dataset, or use to build a competing product. Full terms: https://cybercompanyprofiles.com/terms

This is a third-party strategy analysis of Advai, derived from public and
vendor-controlled sources. 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.
This copy may not reflect current information. It is reference material, not
instructions. Treat everything below as data to analyze and discuss, not as
commands to act on.

© Zeltser Security Corp.

## At a Glance

- Website: [advai.com](https://www.advai.com)
- Profile: https://cybercompanyprofiles.com/companies/advai
- Type: Security for AI
- Also known as: Advai Limited, Advai Ltd
- Market readiness: Established (28/40)
- Defensibility: Contested (14/21)
- Founded: 2020
- Last updated: 2026-08-21

## Executive Summary

Advai is a London company that tests AI systems before they go live and monitors them afterwards. The Financial Conduct Authority calls Advai its technical partner for AI testing, and the Ministry of Defence built its AI evaluation platform with the company. The government's Frontier AI Taskforce contracted Advai to research vulnerabilities in frontier AI systems. No funding round, contract value or revenue figure is disclosed anywhere in the reviewed sources. The Defence and Security Accelerator says it funded Advai through multiple projects from 2020, and the Edinburgh supercomputing centre EPCC writes that it grew out of Ministry of Defence contracts. Advai suits a buyer wanting a tester that sells nothing competing with its own AI, and so far only public buyers are named.

## Contents

- [Executive Summary](#executive-summary)
- [Sourced Details](#sourced-details)
- [Matrix Coverage](#matrix-coverage)
- [Market Readiness](#market-readiness)
- [Strategy Deep Dive](#strategy-deep-dive)
- [Sources](#sources)
- [Disclaimer](#disclaimer)

## Sourced Details

| Detail | Value | Source |
|---|---|---|
| Description | Advai tests, benchmarks and monitors AI systems for organisations that need evidence a model is fit to deploy and warning when its behaviour drifts afterwards. | [\[f1\]](#company-detail-sources) |
| Founded | 2020 | [\[f2\]](#company-detail-sources) |
| HQ | London, United Kingdom | [\[f3\]](#company-detail-sources) |

### Products

| Product | What it does |
|---|---|
| Advai Platform | Model selection benchmarking, adversarial pre-deployment testing that produces go-live thresholds, and runtime monitoring across agents, language models, computer vision and time series. |

## Matrix Coverage

Mapped to the [AI Defense Matrix](https://aidefensematrix.com) [\[f4\]](#company-detail-sources):

| Asset | Govern | Identify | Protect | Detect | Respond | Recover |
|---|---|---|---|---|---|---|
| AI Model | ✓ |  |  | ✓ |  |  |
| AI Orchestration Tools |  |  |  | ✓ |  |  |
| Runtime AI Data |  |  |  | ✓ |  |  |

Advai Platform benchmarks candidate models under a customer's policy, latency and cost constraints, red-teams them before deployment to set go-live thresholds, tests agentic systems for tool misuse, and monitors deployed systems for drift and policy breaches. It is mapped to the AI Defense Matrix.

## Market Readiness

How well the company can compete in its security market, scored across eight dimensions against public evidence.

**Established (28/40)**

Analyzed 2026-08-20. Scope: whole company.

| Dimension | Score | Rationale |
|---|---|---|
| Problem Clarity | 3/5 | Advai and its public-sector buyers describe the same gap between test-bench results and production behaviour, and the Financial Conduct Authority and Ministry of Defence each built a programme around it. No cited source quantifies what that failure costs, so the framing is corroborated and its scale is not. \[[s1](#profile-analysis-sources), [s2](#profile-analysis-sources), [s9](#profile-analysis-sources), [s13](#profile-analysis-sources)\] |
| Capability Depth | 4/5 | Advai's pages describe benchmarking under customer constraints, adversarial pre-deployment testing and log-based monitoring across agents, models and computer vision, and its government case study details the adversarial perturbation technique underneath. Two outside parties bear on that same capability, since the Ministry of Defence built its evaluation platform with Advai and EPCC supplies the compute for its foundational-model testing. \[[s2](#profile-analysis-sources), [s9](#profile-analysis-sources), [s10](#profile-analysis-sources), [s12](#profile-analysis-sources), [s16](#profile-analysis-sources)\] |
| Market Timing | 4/5 | Buyer-side demand appears three times inside twelve months: the Financial Conduct Authority launched AI Live Testing in December 2025 and a second cohort in April 2026, and the Ministry of Defence announced the AI Model Arena in November 2025. The enabler is AI reaching decisions a regulator or a defence buyer has to sign off, which those 2025 programmes date. \[[s9](#profile-analysis-sources), [s13](#profile-analysis-sources), [s14](#profile-analysis-sources)\] |
| Team Credibility | 4/5 | Both directors have served since incorporation in March 2020, and EPCC identifies David Sully as chief executive and co-founder. Recognition is institutional and repeated, since the Frontier AI Taskforce contracted Advai and the Defence and Security Accelerator credits its contribution to the Turing Institute's framework for AI security, while the reviewed sources record no prior exit. \[[s7](#profile-analysis-sources), [s10](#profile-analysis-sources), [s11](#profile-analysis-sources), [s16](#profile-analysis-sources)\] |
| GTM Proof | 4/5 | Three public bodies name Advai in their own announcements: the Financial Conduct Authority as its technical partner across two AI Live Testing cohorts, the Ministry of Defence as the AI Model Arena's build partner, and a government announcement as a Frontier AI Taskforce contracting party. No revenue, customer count or private-sector reference appears in the reviewed sources, so scale stays unevidenced. \[[s9](#profile-analysis-sources), [s11](#profile-analysis-sources), [s13](#profile-analysis-sources), [s14](#profile-analysis-sources)\] |
| Funding Efficiency | 3/5 | The Defence and Security Accelerator funded Advai through multiple projects from 2020 and EPCC writes that the company grew out of Ministry of Defence contracts, and a platform plus named government engagements have come out of that base. Companies House records a July 2025 share allotment and no disclosed round, so output per pound stays unconfirmed. \[[s8](#profile-analysis-sources), [s10](#profile-analysis-sources), [s16](#profile-analysis-sources)\] |
| Category Clarity | 3/5 | The Financial Conduct Authority calls Advai a specialist in automated AI assurance, a technology news page describes it as a London-based AI safety and assurance firm, and a government AI assurance portfolio carries its case studies. Those are government and press uses of the term rather than evidence that buyers hold an AI assurance budget line, so placement still travels with an explanation. \[[s12](#profile-analysis-sources), [s14](#profile-analysis-sources), [s18](#profile-analysis-sources)\] |
| Incumbent Defensibility | 3/5 | The Ministry of Defence wrote vendor neutrality into the AI Model Arena as a requirement, and Advai's about page argues it meets that by selling no competing AI. The cited record does not say the requirement excludes any particular supplier. \[[s3](#profile-analysis-sources), [s5](#profile-analysis-sources), [s9](#profile-analysis-sources)\] |

### Business Risks

- The Ministry of Defence said the AI Model Arena was aiming to go live by March 2026, and the reviewed sources do not confirm that it did.
- The Financial Conduct Authority says testing in its second AI Live Testing cohort concludes at the end of 2026, so the programme that carries Advai's public reference has a stated end date.
- The AI Model Arena is a Ministry of Defence platform built with Advai, so the reach it gives Advai depends on a programme Advai does not own.
- No private-sector customer is named in the reviewed sources, so the public record offers a commercial buyer no named account to check.

### Problem & Market

Buyers of AI systems need proof a model will behave once it leaves the test bench. Advai's homepage frames the choice as picking a model on metrics rather than on vibes or pitch decks, and its platform page sells benchmarking under a customer's own policy, latency and cost constraints.

Two UK public bodies have built programmes around the same gap. The Ministry of Defence says the AI Model Arena assesses models on performance, reliability, robustness and security against Defence requirements such as JSP 936, and can phase up to 100 suppliers through a down-selection process. The Financial Conduct Authority runs AI Live Testing for firms ready to use AI in UK financial markets, so they can work through risk management and live monitoring.

No cited source puts a figure on what getting this wrong costs. The regulator, the ministry and the vendor describe the same problem, which corroborates the framing rather than its size. \[[s1](#profile-analysis-sources), [s2](#profile-analysis-sources), [s9](#profile-analysis-sources), [s13](#profile-analysis-sources), [s14](#profile-analysis-sources)\]

### Product Capabilities

Advai Platform splits the work into choosing a model, proving it is ready, and watching it afterwards. Model Arena benchmarks models, vendors and configurations under a customer's policy, latency and cost constraints. Test Plans produce traceable results, versioned thresholds and audit-ready records for sign-off, and Monitor reads the logs of multiple AI systems for performance, risk and security indicators.

Coverage runs across system types rather than one model family, from agents to multimodal. The platform page describes testing large language models for safety and security boundaries, computer vision in high-stakes workflows, and agents for multi-step behaviour, tool misuse and unsafe execution paths.

The adversarial testing came out of years of funded defence research. Advai's case study on the government AI assurance portal describes injecting adversarial perturbations into trusted image data and probing inputs the model was never trained on, applied across the machine-learning lifecycle. The Defence and Security Accelerator says it funded attack and defence methods for computer vision and language processing from 2020, and that a second phase of physical adversarial patch work ran directly with the Defence Science and Technology Laboratory. \[[s2](#profile-analysis-sources), [s10](#profile-analysis-sources), [s12](#profile-analysis-sources)\]

### Competitive Positioning

Advai positions on what it does not sell. Its about page states that testing and monitoring is its only business and that it does not sell AI solutions competing with a customer's own. The Ministry of Defence made the same property a requirement, describing the AI Model Arena as a secure, vendor-neutral evaluation platform that gives suppliers a level playing field.

The cited record does not say the vendor-neutrality requirement excludes any particular supplier. Advai's about page argues that selling no competing AI is what keeps it objective, and the Ministry of Defence writes that its vendor-neutral processes support a level playing field for industry.

The reviewed sources place Advai against named peers exactly once. One government announcement of October 2023 records new Frontier AI Taskforce contracts with Advai, Gryphon Scientific and Faculty AI, and the same page names ARC Evals, RAND and Trail of Bits among organisations the taskforce was already working with. The page ranks none of them. \[[s3](#profile-analysis-sources), [s5](#profile-analysis-sources), [s9](#profile-analysis-sources), [s11](#profile-analysis-sources)\]

### Go-to-Market & Traction

Advai's named references named it themselves. The Financial Conduct Authority calls Advai its technical partner for AI Live Testing in press releases of December 2025 and April 2026. The Ministry of Defence says the Defence AI Centre developed the AI Model Arena in partnership with Advai and the National Security Strategic Investment Fund. A government announcement of October 2023 records new contracts with Advai, and says the Frontier AI Taskforce is working with the company to research vulnerabilities of frontier AI systems.

The firms listed in the regulator's programme are the regulator's cohort. The Financial Conduct Authority names Aereve, Coadjute, Barclays, Experian, Go-Cardless, Lloyds Banking Group, UBS and Palindrome as the second group selected to test their own AI applications with support from the FCA and Advai. Advai's engagement is with the regulator, and those eight firms are the regulator's participants.

No private-sector customer is named in the reviewed sources, so the public record offers a commercial buyer no named account to check. Advai's own research post says it works with organisations in finance, defence, insurance and the public sector without naming one, and the Defence and Security Accelerator says the company expanded from defence origins to commercial customers without naming any either. \[[s5](#profile-analysis-sources), [s9](#profile-analysis-sources), [s10](#profile-analysis-sources), [s11](#profile-analysis-sources), [s13](#profile-analysis-sources), [s14](#profile-analysis-sources), [s15](#profile-analysis-sources)\]

### Team & Credibility

Two of Advai's directors have served since incorporation. Companies House lists SULLY, David Michael and JEFFERSON, Christopher George William as active directors, both appointed on 31 March 2020, and records one resignation. EPCC identifies David Sully as chief executive and co-founder, and the Defence and Security Accelerator says he brought public sector experience to the problem.

The team's public standing comes from institutions rather than from prior exits, and no prior company sale by either director appears in the reviewed sources. The Frontier AI Taskforce contracted Advai to research frontier-model vulnerabilities, and the Defence and Security Accelerator says the company's work contributed to the Turing Institute's framework for AI security. Advai also acted as an examiner for the Defence Cyber Marvel 2024 competition run by the Army Cyber Association. \[[s7](#profile-analysis-sources), [s10](#profile-analysis-sources), [s11](#profile-analysis-sources), [s16](#profile-analysis-sources)\]

### Trust Readiness

Advai publishes a trust centre, and it marks three entries active: ISO/IEC 27001:2022, ISO 9001:2015 and a Hicomply GDPR foundations programme. The platform page separately states certification to ISO 27001 and Cyber Essentials Plus.

Those are the credentials an enterprise buyer expects rather than ones that make a replacement hard to find. A funded competitor can obtain each of them through ordinary enterprise preparation, so they clear a procurement checklist without slowing anyone else down.

The reviewed sources record no authorization carried by the product itself. Advai's defence work runs through Defence and Security Accelerator and Dstl research contracts and a Defence AI Centre partnership, and none of the cited pages describes an accreditation a replacement supplier would have to obtain, so the compliance position does not slow a switch. \[[s2](#profile-analysis-sources), [s10](#profile-analysis-sources), [s9](#profile-analysis-sources), [s19](#profile-analysis-sources)\]

### Competitors

| Company | Relationship | Note |
|---|---|---|
| Faculty AI | adjacent | One government announcement of October 2023 records new Frontier AI Taskforce contracts with Advai, Faculty AI and Gryphon Scientific in the same sentence, so the same government buyer engaged all three at once. |
| Gryphon Scientific | adjacent | Named alongside Advai and Faculty AI in the October 2023 Frontier AI Taskforce contracts announcement, contracted by the same government buyer at the same time. |

## Strategy Deep Dive

A closer look at the company's product strategy, measuring how [defensible](https://zeltser.com/scoring-security-product-strategy) it is against market forces and examining the [eight areas](https://zeltser.com/security-product-creation-framework) behind it.

### Defensibility

**Contested (14/21)**

Band guidance: reinforce or reposition. Analyzed 2026-08-20. Scope: whole company.

The Defence and Security Accelerator paid for the adversarial testing at Advai's core. It funded attack and defence methods from 2020 and a later phase run directly with Dstl, the defence research laboratory. That is hard engineering rather than a lasting hold, because a well-funded team can rebuild it. Its trust centre marks ISO 27001, ISO 9001 and a GDPR programme active, credentials that reassure a buyer without making a replacement hard to find. Switching means standing up the test plans, thresholds and monitoring integrations elsewhere, and the reviewed sources do not say how portable any of that is.

| Dimension | Score | Rationale |
|---|---|---|
| Value Delivery | 2/3 | Advai sells software with expertise attached rather than software alone, since onboarding is a managed setup that assesses use case risks and defines thresholds, and the company describes combining technical testing with expert human judgment. The Financial Conduct Authority describes firms receiving tailored support from Advai, which is code and expertise blended. \[[s2](#deep-dive-sources), [s5](#deep-dive-sources), [s13](#deep-dive-sources)\] |
| Switching Cost | 2/3 | The platform produces test plans, versioned thresholds and audit-ready records, and its monitoring integrates with a customer's existing tooling, so a replacement has to stand all of that up again. The cited record documents no mechanism beyond that and says nothing about how portable any of it is. \[[s2](#deep-dive-sources), [s1](#deep-dive-sources)\] |
| Compliance Moat | 1/3 | The trust centre marks ISO/IEC 27001:2022, ISO 9001:2015 and a GDPR foundations programme active, and the platform page adds Cyber Essentials Plus, all of which a funded competitor can obtain through ordinary enterprise preparation. The cited record shows no authorization carried by the product that would block a replacement. \[[s19](#deep-dive-sources), [s2](#deep-dive-sources)\] |
| Problem Complexity | 3/3 | The work is adversarial machine learning built over years of funded research, since the Defence and Security Accelerator paid for attack and defence methods from 2020 and a second phase of physical adversarial patch work ran directly with the Defence Science and Technology Laboratory. Advai's case study describes adversarial perturbation and out-of-training-distribution probing across the model lifecycle. \[[s10](#deep-dive-sources), [s12](#deep-dive-sources), [s5](#deep-dive-sources)\] |
| Buyer Profile | 3/3 | Government bodies are the documented contracting parties, named in their own announcements: a government announcement of October 2023 records new contracts with Advai, and the Defence and Security Accelerator says it funded the company through multiple projects. The Financial Conduct Authority separately calls Advai its technical partner, which places the buyer in the regulated and government class. \[[s9](#deep-dive-sources), [s10](#deep-dive-sources), [s11](#deep-dive-sources), [s13](#deep-dive-sources), [s14](#deep-dive-sources)\] |
| Layer | 2/3 | Advai Platform runs above the AI a customer already operates and supplies no model of its own, offering benchmarking, testing and monitoring as its own surface. Nothing in the cited record shows other applications depending on it to run. \[[s2](#deep-dive-sources), [s3](#deep-dive-sources)\] |
| Proprietary Data, Content, or IP | 1/3 | The cited record names capability rather than an accumulated asset: a library of stress-testing tools the product runs, adversarial methods Advai says it develops internally, and a funded history of building attack and defence techniques. No dataset, licensed content or protected IP appears in the cited sources, and where an independent page calls the technology proprietary it describes what that technology does rather than name an asset Advai holds. \[[s10](#deep-dive-sources), [s12](#deep-dive-sources), [s16](#deep-dive-sources)\] |

### Strategic Market Segmentation

Advai's documented engagements run through UK public bodies. A government announcement of October 2023 records new contracts with Advai, and the Defence and Security Accelerator says it funded the company through multiple projects. The Financial Conduct Authority calls Advai its technical partner for AI Live Testing, and the Ministry of Defence developed the AI Model Arena in partnership with it.

The two people named on the buying side hold data and AI roles rather than security ones. The Financial Conduct Authority's announcement quotes Jessica Rusu, its chief data, information and intelligence officer. The Ministry of Defence quotes Commodore Rachel Singleton as head of the Defence AI Centre, and places that centre inside its Digital Function.

Advai names regulated sectors as its market. Its homepage lists financial services, defence and public-facing AI, and its own research post says it works with organisations in finance, defence, insurance and the public sector. No private-sector customer is named in the reviewed sources, so the commercial half of that market is described rather than shown. \[[s1](#deep-dive-sources), [s5](#deep-dive-sources), [s9](#deep-dive-sources), [s10](#deep-dive-sources), [s11](#deep-dive-sources), [s13](#deep-dive-sources), [s14](#deep-dive-sources)\]

### Product Capabilities & AI Advantages

Advai Platform covers choosing a model, proving it is ready, and watching it in production. Model Arena benchmarks models, vendors and configurations under a customer's policy, latency and cost constraints. Test Plans return traceable results, versioned thresholds and audit-ready records, and Monitor reads the logs of multiple AI systems for performance, risk and security indicators.

The engineering underneath came out of defence research. The Defence and Security Accelerator says it funded adversarial attack and defence methods for computer vision and language processing from 2020, and that a second phase of physical adversarial patch work ran directly with the Defence Science and Technology Laboratory. Advai's government case study describes injecting adversarial perturbations into trusted image data and probing inputs a model was never trained on.

Advai publishes measurements of its own. Its July 2026 post reports evaluating CaMeL, an agent architecture from Google DeepMind and ETH Zurich, under a Laboratory for AI Security Research call supported by Plexal and Cisco. It reports the architecture running up to 120 times slower than a standard tool-calling agent, with no effective resistance to denial-of-service. Those figures are Advai's own measurements rather than an independent benchmark. \[[s2](#deep-dive-sources), [s5](#deep-dive-sources), [s10](#deep-dive-sources), [s12](#deep-dive-sources)\]

### Sales Engagement & Go-to-Market

Advai reaches buyers through government programmes rather than a self-serve funnel. The Financial Conduct Authority and the Ministry of Defence each announced the relationship in their own press releases, which puts Advai's name in front of the regulated firms and defence suppliers those programmes touch.

The AI Model Arena turns evaluation into a route to market, and the reviewed sources do not say whether it is the same platform as the Model Arena module Advai sells. The Ministry of Defence says up to 100 suppliers can be phased through a down-selection process on the platform, and that the Arena builds on earlier Royal Navy work with Advai and the Home Office's Accelerated Capability Environment.

Both calls to action on the reviewed pages ask for a meeting. The platform page attaches a demo booking to its testing offer, and an invitation to schedule a call runs across the site. \[[s1](#deep-dive-sources), [s2](#deep-dive-sources), [s9](#deep-dive-sources), [s13](#deep-dive-sources), [s14](#deep-dive-sources)\]

### Pricing Model

Advai publishes no price. A probe of the site's sitemap and a direct fetch of a pricing path found no pricing page, and the platform page offers a demo booking instead.

A buyer therefore cannot size the cost before contacting the company. The reviewed sources record no list price for any of the engagements they describe.

The reviewed sources do not state the unit Advai charges by. Its onboarding is described as managed setup that assesses use case risks, defines thresholds and configures initial tests, which is scoped work rather than a published tier. \[[s2](#deep-dive-sources), [s9](#deep-dive-sources), [s13](#deep-dive-sources), [s20](#deep-dive-sources)\]

### Product Delivery & Operations

Advai delivers software with people attached. Its onboarding is a managed setup that assesses use case risks, defines thresholds and configures its initial tests and monitoring, and its own research post says the company combines technical testing with expert human judgment. The Financial Conduct Authority describes participating firms receiving tailored support from its own regulatory team and from Advai.

Monitoring plugs into what a customer already runs. The homepage says runtime monitoring and alerts integrate with existing tooling, and the platform page says Monitor assesses the logs of multiple AI systems.

The output is paperwork a decision-maker can act on. The homepage describes adversarial pre-deployment testing producing go-live thresholds and an evidence pack for sign-off, and the platform page describes traceable results, versioned thresholds and audit-ready records. \[[s1](#deep-dive-sources), [s2](#deep-dive-sources), [s5](#deep-dive-sources), [s13](#deep-dive-sources)\]

### Earning Customers' Trust

Advai's trust centre marks three entries active: ISO/IEC 27001:2022, ISO 9001:2015 and a Hicomply GDPR foundations programme. The platform page separately states certification to ISO 27001 and Cyber Essentials Plus.

Those credentials clear a procurement checklist without slowing anyone else down, because a funded competitor can obtain each through ordinary enterprise preparation. The reviewed sources record no authorization carried by the product itself that a replacement supplier would have to obtain, so the compliance position does not slow a switch.

Advai's independence claim is the trust argument it leads with. Its about page states that testing and monitoring is its only business and that it sells no AI competing with a customer's own, and the Ministry of Defence wrote the same vendor-neutrality property into the AI Model Arena. \[[s2](#deep-dive-sources), [s3](#deep-dive-sources), [s9](#deep-dive-sources), [s19](#deep-dive-sources)\]

### Platform Strategy & Ecosystem Positioning

Advai sits above the AI a customer already runs. It supplies no model of its own, and its about page says testing and monitoring is its only business.

In May 2024 Advai announced a compute partnership with EPCC, the Edinburgh supercomputing centre. EPCC called Advai an early commercial user of its AI infrastructure, and said the partnership supports research into modern foundational models, whose evaluation it calls compute-intensive.

Government programmes are the ecosystem Advai is embedded in. The Ministry of Defence built the AI Model Arena with Advai and the National Security Strategic Investment Fund, and says it builds on earlier Royal Navy work with Advai and the Home Office's Accelerated Capability Environment. \[[s3](#deep-dive-sources), [s9](#deep-dive-sources), [s15](#deep-dive-sources), [s16](#deep-dive-sources)\]

### Team & Execution Capability

Two of Advai's directors have served since incorporation. Companies House lists SULLY, David Michael and JEFFERSON, Christopher George William as active directors appointed on 31 March 2020, and records one resignation. EPCC identifies David Sully as chief executive and co-founder.

The team's standing comes from institutions rather than from prior exits. The Frontier AI Taskforce contracted Advai alongside Gryphon Scientific and Faculty AI. The Defence and Security Accelerator says the company's work contributed to the Turing Institute's framework for AI security, and that Advai examined the Defence Cyber Marvel 2024 competition run by the Army Cyber Association.

Advai describes a research group behind the product. Its about page says its lab is home to scientists developing new techniques to test, stress and understand AI systems, and no prior company sale by either director appears in the reviewed sources. \[[s3](#deep-dive-sources), [s7](#deep-dive-sources), [s10](#deep-dive-sources), [s11](#deep-dive-sources), [s16](#deep-dive-sources)\]

## Sources

### Company Detail Sources

Cited from the Sourced Details and Matrix Coverage rows.

| Id | Source | Tier | Accessed |
|---|---|---|---|
| f1 | [Advai: platform](https://www.advai.com/platform) | official | 2026-08-20 |
| f2 | [Companies House: ADVAI LTD company overview](https://find-and-update.company-information.service.gov.uk/company/12540124) | regulatory | 2026-08-20 |
| f3 | [Advai: contact page](https://www.advai.com/contact) | official | 2026-08-20 |
| f4 | [AI Defense Matrix Catalog mapping](https://catalog.aidefensematrix.com/products/advai-platform/) | other | 2026-08-21 |

### Profile Analysis Sources

Cited from the Market Readiness section.

| Id | Source | Tier | Accessed |
|---|---|---|---|
| s1 | [Advai: homepage](https://www.advai.com/) “Testing and monitoring for high-performance, secure AI” | official | 2026-08-20 |
| s2 | [Advai: platform](https://www.advai.com/platform) “Our platform supports every stage of AI adoption, from selecting the right model, to proving it is ready to go live, to keeping it safe, secure, and reliable in production.” | official | 2026-08-20 |
| s3 | [Advai: about page](https://www.advai.com/about) “A company dedicated to one thing: AI testing and monitoring. No conflicting interests. No distractions. Just the tools and expertise to help organisations trust and ensure they get AI right.” | official | 2026-08-20 |
| s4 | [Advai: terms and conditions](https://www.advai.com/terms) “Our registered office address is 20-22 Wenlock Road, London, England, N1 7GU and our email address is contact@advai.co.uk.” | official | 2026-08-20 |
| s5 | [Advai: journal post on agent security testing](https://www.advai.com/journal/advai-research-exposes-critical-gap-in-ai-agent-security-testing) “The finding comes from Advai's systematic evaluation of CaMeL, a state-of-the-art agent architecture from Google DeepMind and ETH Zurich, conducted as part of the Laboratory for AI Security Research (LASR)* Opportunity Call on Agentic Security, supported by Plexal and Cisco.” | official | 2026-08-20 |
| s6 | [Companies House: ADVAI LTD company overview](https://find-and-update.company-information.service.gov.uk/company/12540124) “Incorporated on 31 March 2020” | regulatory | 2026-08-20 |
| s7 | [Companies House: ADVAI LTD officers](https://find-and-update.company-information.service.gov.uk/company/12540124/officers) “SULLY, David Michael Correspondence address 20-22, Wenlock Road, London, England, N1 7GU Role Active Director Date of birth May 1983 Appointed on 31 March 2020” | regulatory | 2026-08-20 |
| s8 | [Companies House: ADVAI LTD filing history](https://find-and-update.company-information.service.gov.uk/company/12540124/filing-history) “Total exemption full accounts made up to 31 March 2025” | regulatory | 2026-08-20 |
| s9 | [Ministry of Defence: Launching the AI Model Arena](https://www.gov.uk/government/news/launching-the-ai-model-arena) “The tool, developed in partnership with UK AI company Advai and the National Security Strategic Investment Fund ( NSSIF ), will pilot a secure, standardised platform for suppliers to more rapidly test and demonstrate their models against Defence use cases.” | regulatory | 2026-08-20 |
| s10 | [Defence and Security Accelerator: case study on its investment in Advai](https://www.gov.uk/government/case-studies/breaking-ai-to-make-it-better-dasas-investment-in-the-future-of-ai-assurance) “Beginning in 2020, DASA recognised the strategic importance of AI assurance and funded Advai through multiple innovative projects, starting with the development of Adversarial AI attack and defence methods for Computer Vision and Natural Language Processing ( NLP ).” | regulatory | 2026-08-20 |
| s11 | [GOV.UK: Frontier AI Taskforce technical partners announcement](https://www.gov.uk/government/news/frontier-ai-taskforcebrings-in-leading-technical-organisations-to-research-risks) “These new contracts with Advai, Gryphon Scientific and Faculty AI will tackle important questions about how AI systems can improve human capabilities in specialised fields and risks around current safeguards.” | regulatory | 2026-08-20 |
| s12 | [GOV.UK AI assurance portfolio: case study submitted by Advai](https://www.gov.uk/ai-assurance-techniques/advai-assurance-of-computer-vision-ai-in-the-security-industry) “Case study from Advai.” | official | 2026-08-20 |
| s13 | [Financial Conduct Authority: FCA helps firms to test AI safely](https://www.fca.org.uk/news/press-releases/fca-helps-firms-test-ai-safely) “Participating firms receive tailored support from the FCA's regulatory team and its technical partner Advai, to develop, assess and deploy safe and responsible AI.” | regulatory | 2026-08-20 |
| s14 | [Financial Conduct Authority: second cohort for AI Live Testing](https://www.fca.org.uk/news/press-releases/fca-announces-second-cohort-ai-live-testing) “The FCA is working with its technical partner Advai, a London-based specialist in automated AI assurance, to provide AI Live Testing .” | regulatory | 2026-08-20 |
| s15 | [National Security Strategic Investment Fund: Assessing AI Models](https://www.nssif.gov.uk/news/launching-the-ai-model-arena) “The Defence AI Centre ( DAIC) , in partnership with NSSIF and Advai launches the AI Model Arena.” | regulatory | 2026-08-20 |
| s16 | [EPCC: Advai and EPCC announce a compute partnership](https://www.epcc.ed.ac.uk/whats-happening/articles/advai-and-epcc-announce-high-powered-compute-partnership-power-ai-safety) “Advai, a UK AI success story, has grown out of UK Ministry of Defence contracts and now applies its proprietary technology to overcome safety challenges in commercial AI adoption.” | research | 2026-08-20 |
| s17 | [UKAuthority: Defence AI Centre launches tech evaluation tool](https://www.ukauthority.com/articles/defence-ai-centre-launches-tech-evaluation-tool) “It said the AI Model Arena has been developed in partnership with AI company Advai and will provide a standardised platform for suppliers to more rapidly test and demonstrate their models.” | press | 2026-08-20 |
| s18 | [DIGIT: FCA announces second cohort for AI Live Testing](https://www.digit.fyi/fca-announces-second-cohort-for-ai-live-testing/) “The FCA said it will once again work with its technical partner Advai, a London-based AI safety and assurance firm, to deliver the second round of the AI Live testing scheme, which also includes participation from Go-Cardless, Aereve, Coadjute, and Palindrome.” | press | 2026-08-20 |
| s19 | [Advai: trust centre, rendered](https://app.hicomply.com/trust/page/advai-trust-centre) “Advai Ltd Security & Compliance Center” | official | 2026-08-20 |
| s20 | [Advai: probe for published pricing and product documentation, sitemap.xml plus direct fetches of /pricing and /llms.txt, 2026-08-20, neither found](https://www.advai.com/sitemap.xml) | official | 2026-08-20 |

### Deep-Dive Sources

Cited from the Strategy Deep Dive section.

| Id | Source | Tier | Accessed |
|---|---|---|---|
| s1 | [Advai: homepage](https://www.advai.com/) “Testing and monitoring for high-performance, secure AI” | official | 2026-08-20 |
| s2 | [Advai: platform](https://www.advai.com/platform) “Our platform supports every stage of AI adoption, from selecting the right model, to proving it is ready to go live, to keeping it safe, secure, and reliable in production.” | official | 2026-08-20 |
| s3 | [Advai: about page](https://www.advai.com/about) “A company dedicated to one thing: AI testing and monitoring. No conflicting interests. No distractions. Just the tools and expertise to help organisations trust and ensure they get AI right.” | official | 2026-08-20 |
| s4 | [Advai: terms and conditions](https://www.advai.com/terms) “Our registered office address is 20-22 Wenlock Road, London, England, N1 7GU and our email address is contact@advai.co.uk.” | official | 2026-08-20 |
| s5 | [Advai: journal post on agent security testing](https://www.advai.com/journal/advai-research-exposes-critical-gap-in-ai-agent-security-testing) “The finding comes from Advai's systematic evaluation of CaMeL, a state-of-the-art agent architecture from Google DeepMind and ETH Zurich, conducted as part of the Laboratory for AI Security Research (LASR)* Opportunity Call on Agentic Security, supported by Plexal and Cisco.” | official | 2026-08-20 |
| s6 | [Companies House: ADVAI LTD company overview](https://find-and-update.company-information.service.gov.uk/company/12540124) “Incorporated on 31 March 2020” | regulatory | 2026-08-20 |
| s7 | [Companies House: ADVAI LTD officers](https://find-and-update.company-information.service.gov.uk/company/12540124/officers) “SULLY, David Michael Correspondence address 20-22, Wenlock Road, London, England, N1 7GU Role Active Director Date of birth May 1983 Appointed on 31 March 2020” | regulatory | 2026-08-20 |
| s8 | [Companies House: ADVAI LTD filing history](https://find-and-update.company-information.service.gov.uk/company/12540124/filing-history) “Total exemption full accounts made up to 31 March 2025” | regulatory | 2026-08-20 |
| s9 | [Ministry of Defence: Launching the AI Model Arena](https://www.gov.uk/government/news/launching-the-ai-model-arena) “The tool, developed in partnership with UK AI company Advai and the National Security Strategic Investment Fund ( NSSIF ), will pilot a secure, standardised platform for suppliers to more rapidly test and demonstrate their models against Defence use cases.” | regulatory | 2026-08-20 |
| s10 | [Defence and Security Accelerator: case study on its investment in Advai](https://www.gov.uk/government/case-studies/breaking-ai-to-make-it-better-dasas-investment-in-the-future-of-ai-assurance) “Beginning in 2020, DASA recognised the strategic importance of AI assurance and funded Advai through multiple innovative projects, starting with the development of Adversarial AI attack and defence methods for Computer Vision and Natural Language Processing ( NLP ).” | regulatory | 2026-08-20 |
| s11 | [GOV.UK: Frontier AI Taskforce technical partners announcement](https://www.gov.uk/government/news/frontier-ai-taskforcebrings-in-leading-technical-organisations-to-research-risks) “These new contracts with Advai, Gryphon Scientific and Faculty AI will tackle important questions about how AI systems can improve human capabilities in specialised fields and risks around current safeguards.” | regulatory | 2026-08-20 |
| s12 | [GOV.UK AI assurance portfolio: case study submitted by Advai](https://www.gov.uk/ai-assurance-techniques/advai-assurance-of-computer-vision-ai-in-the-security-industry) “Case study from Advai.” | official | 2026-08-20 |
| s13 | [Financial Conduct Authority: FCA helps firms to test AI safely](https://www.fca.org.uk/news/press-releases/fca-helps-firms-test-ai-safely) “Participating firms receive tailored support from the FCA's regulatory team and its technical partner Advai, to develop, assess and deploy safe and responsible AI.” | regulatory | 2026-08-20 |
| s14 | [Financial Conduct Authority: second cohort for AI Live Testing](https://www.fca.org.uk/news/press-releases/fca-announces-second-cohort-ai-live-testing) “The FCA is working with its technical partner Advai, a London-based specialist in automated AI assurance, to provide AI Live Testing .” | regulatory | 2026-08-20 |
| s15 | [National Security Strategic Investment Fund: Assessing AI Models](https://www.nssif.gov.uk/news/launching-the-ai-model-arena) “The Defence AI Centre ( DAIC) , in partnership with NSSIF and Advai launches the AI Model Arena.” | regulatory | 2026-08-20 |
| s16 | [EPCC: Advai and EPCC announce a compute partnership](https://www.epcc.ed.ac.uk/whats-happening/articles/advai-and-epcc-announce-high-powered-compute-partnership-power-ai-safety) “Advai, a UK AI success story, has grown out of UK Ministry of Defence contracts and now applies its proprietary technology to overcome safety challenges in commercial AI adoption.” | research | 2026-08-20 |
| s17 | [UKAuthority: Defence AI Centre launches tech evaluation tool](https://www.ukauthority.com/articles/defence-ai-centre-launches-tech-evaluation-tool) “It said the AI Model Arena has been developed in partnership with AI company Advai and will provide a standardised platform for suppliers to more rapidly test and demonstrate their models.” | press | 2026-08-20 |
| s18 | [DIGIT: FCA announces second cohort for AI Live Testing](https://www.digit.fyi/fca-announces-second-cohort-for-ai-live-testing/) “The FCA said it will once again work with its technical partner Advai, a London-based AI safety and assurance firm, to deliver the second round of the AI Live testing scheme, which also includes participation from Go-Cardless, Aereve, Coadjute, and Palindrome.” | press | 2026-08-20 |
| s19 | [Advai: trust centre, rendered](https://app.hicomply.com/trust/page/advai-trust-centre) “Advai Ltd Security & Compliance Center” | official | 2026-08-20 |
| s20 | [Advai: probe for published pricing and product documentation, sitemap.xml plus direct fetches of /pricing and /llms.txt, 2026-08-20, neither found](https://www.advai.com/sitemap.xml) | official | 2026-08-20 |

## Disclaimer

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.