<!-- https://cyberessentialssoftware.com/method/ai/ | read 18 Nov 2026 -->

Criterion, weight 10%

# AI that does the work under control: how 14 Cyber Essentials platforms score

Read 18 Nov 2026

AI that does the work under control carries 10 percent of the main ranking. On 18 Nov 2026 the highest score is 10 out of 10, held by Comp AI. Scytale scores 7, 3 below the top. The criterion measures AI or an agent that drafts answers and policies and does compliance tasks, with review and approval before anything is used. Sprinto leads the main ranking at 5.75; Scytale ranks 7th of 10 at 4.60. Scytale publishes no price, and sells its expert services separately from its plans, with no certification body named for Cyber Essentials. Sprinto is ahead mainly on guided route to the certificate: 5 against Scytale's 1 at a weight of 25 percent.

## How do the 14 platforms score on AI that does the work under control?

Scores are listed highest first. Spread of scores, with the number of platforms at each: 10 (1), 7 (10), 0 (3).

| Platform | Score | Reason | Sources |
| --- | --- | --- | --- |
| [Comp AI](/platforms/comp-ai/) not ranked | [10](/platforms/comp-ai/#c-ai) | Comp AI documents agents that pull evidence and verify controls in a browser, policies generated from onboarding context, and questionnaire answers reviewed before export. Policies and AI edits need approval before going live. TechCrunch quotes the CEO that "a person still reviews and approves" an agent-drafted policy, so the cap does not apply. | [Comp AI: home page](https://www.trycomp.ai/) [Comp AI: security questionnaire](https://www.trycomp.ai/docs/security-questionnaire) [Comp AI: AI policy editor](https://www.trycomp.ai/docs/ai-policy-editor) [TechCrunch: Comp AI and an agentic future for compliance](https://techcrunch.com/2026/09/17/comp-ai-sets-eyes-on-a-continiously-agentic-future-for-security-and-complaince/) |
| [Sprinto](/platforms/sprinto/) | [7](/platforms/sprinto/#c-ai) | Sprinto names agents including an Evidence Agent and a Fix it agent, lists AI-assisted policies and AI questionnaires (20 a year on Foundation), and says "Your team reviews, approves, and steers". Vendor-only evidence caps the total of 10 at 7. | [Sprinto: Sprinto AI](https://sprinto.com/sprinto-ai/) [Sprinto: pricing](https://sprinto.com/pricing/) [Sprinto: product updates, July and August 2026](https://sprinto.com/blog/sprinto-product-updates-july-august-2026/) [Sprinto: security](https://sprinto.com/security/) |
| [Secureframe](/platforms/secureframe/) | [7](/platforms/secureframe/#c-ai) | Secureframe documents AI control mapping and evidence validation, an AI writing assistant in the policy editor, and AI-drafted questionnaire answers with cited sources for the user to verify and edit. Its evidence validation states no manual review is needed. The third-party interview confirms only the MCP server launch, so the total of 10 is capped at 7. | [Secureframe: AI features](https://secureframe.com/features/ai) [Secureframe: MCP server announcement](https://secureframe.com/newsroom/secureframe-mcp-server) [AIFN: Shrav Mehta of Secureframe on building AI in production](https://aifn.co/shrav-mehta-of-secureframe-on-building-ai-in-production/) [Secureframe: overview of AI capabilities](https://support.secureframe.com/en/articles/15111072-overview-of-ai-capabilities-in-secureframe) |
| [Thoropass](/platforms/thoropass/) | [7](/platforms/thoropass/#c-ai) | Thoropass has an MCP server so AI agents do work in its audit workflow, and the team verifies that work before it moves forward; it lists Due Diligence Questionnaires powered by GenAI. AI policy drafting: none found (read 18 Nov 2026). Vendor-only evidence caps the total of 8 at 7. | [Thoropass: MCP and AI](https://www.thoropass.com/solutions/mcp-ai) [Thoropass: compliance AI](https://www.thoropass.com/platform/compliance-ai) |
| [Vanta](/platforms/vanta/) | [7](/platforms/vanta/#c-ai) | Vanta says its agent drafts policies, completes questionnaires and flags issues, and questionnaire answers are suggested for the user to "review and submit". Its principles add human-in-the-loop review "where necessary". Vendor-only evidence caps the total of 10 at 7. | [Vanta: AI](https://www.vanta.com/products/ai) [Vanta: Cyber Essentials](https://www.vanta.com/products/cyber-essentials) [Vanta: AI principles](https://help.vanta.com/en/articles/11345368-vanta-ai-principles) |
| [Scytale](/platforms/scytale/) | [7](/platforms/scytale/#c-ai) | Scytale documents AI agents that detect gaps such as "MFA off on 2 root users", one-click policy generation and AI-drafted questionnaire answers that users approve, edit or reject. Vendor-only evidence caps the total of 10 at 7. | [Scytale: agentic compliance](https://scytale.ai/agentic-compliance/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) [Scytale: what is new in Q2 2026](https://scytale.ai/resources/whats-new-in-scytale-q2-2026/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) [Scytale: AI security questionnaires](https://scytale.ai/ai-security-questionnaires/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) |
| [Drata](/platforms/drata/) | [7](/platforms/drata/#c-ai) | Drata documents Agentic TPRM, AI test failure insights and AI policy-to-control mapping, and AI answers security questionnaires that users approve, edit or reject. AI policy drafting: none found (read 18 Nov 2026) (AI suggests policy controls only). Vendor-only evidence caps the total of 8 at 7. | [Drata: AI](https://drata.com/products/ai) [Drata: AI questionnaire assistance](https://drata.com/products/ai-questionnaire-assistance) |
| [Scrut Automation](/platforms/scrut/) | [7](/platforms/scrut/#c-ai) | Scrut Automation Teammates are named AI agents that collect evidence, draft policies and auto-fill security questionnaires, and the homepage says "Your team reviews and approves." The third-party CREST release covers pentesting, not AI, so vendor-only evidence caps the total of 10 at 7. | [Scrut Automation: home page](https://www.scrut.io/) [Scrut Automation: Teammates](https://www.scrut.io/platform/scrut-teammates) |
| [Delve](/platforms/delve/) not ranked | [7](/platforms/delve/#c-ai) | Delve says autonomous AI agents take screenshots, write reports and validate evidence, AI autofills vendor questionnaires, and a dedicated team member reviews everything by hand before the audit. AI policy drafting: none found (read 18 Nov 2026) (templates are starting points). All AI evidence is vendor-only, so the total of 8 is capped at 7. | [Delve: home page](https://delve.co/) [Delve: SOC 2 framework](https://delve.co/product/framework/soc-2) [Delve: response to misleading claims](https://delve.co/blog/response-to-misleading-claims) [Delve: inside its compliance process](https://delve.co/blog/inside-delve-trusted-compliance-process) |
| [Hicomply](/platforms/hicomply/) not ranked | [7](/platforms/hicomply/#c-ai) | Hicomply AI Evidence Discovery surfaces evidence from integrations and suggests control mapping, AI accelerates policy drafting, and every suggestion is for the user to review, accept, reject or amend. AI questionnaire drafting: none found (read 18 Nov 2026). Vendor-only evidence caps the total of 8 at 7. | [Hicomply: AI evidence discovery product](https://www.hicomply.com/product/ai-evidence-discovery) [Hicomply: Hicomply AI](https://www.hicomply.com/platform/hicomply-ai) |
| [DataGuard](/platforms/dataguard/) not ranked | [7](/platforms/dataguard/#c-ai) | DataGuard says its AI Co-Pilot automates policy drafting, evidence collection and audit preparation, generates policies from templates and auto-fills questionnaires. Policies go through a review workflow with approval; questionnaire answers are only editable, with no approval step found. Vendor-only evidence caps the total of 10 at 7. | [DataGuard: AI automation](https://www.dataguard.com/platform/ai-automation/) [DataGuard: AI security questionnaire update](https://www.dataguard.com/product-updates/ai-security-questionnaire/) [DataGuard: frameworks and policies](https://www.dataguard.com/product/frameworks-policies/) |
| [IT Governance](/platforms/it-governance/) | [0](/platforms/it-governance/#c-ai) | IT Governance mentions plain-English AI insights in a threat dashboard and sells AI policy templates, which are policies about AI use, not AI that drafts policies. No AI agent, AI policy drafting, AI questionnaire drafting or review step: none found (read 18 Nov 2026). | [IT Governance: partnership with Defense.com](https://grcsolutions.io/grc-solutions-partners-with-defense-com-to-help-businesses-take-control-of-cyber-risk/) [IT Governance: AI governance](https://grcsolutions.io/ai-governance/) [IT Governance: Cyber Essentials](https://grcsolutions.io/cyber-essentials/) |
| [CyberSmart](/platforms/cybersmart/) | [0](/platforms/cybersmart/#c-ai) | Questionnaire answers fill in from CyberSmart device data, not AI. No AI agent, AI policy drafting, AI questionnaire drafting or AI review step: none found (read 18 Nov 2026). | [CyberSmart: Active Protect](https://cybersmart.co.uk/products/active-protect/) [CyberSmart: security](https://cybersmart.co.uk/security/) [CyberSmart: Cyber Essentials](https://cybersmart.co.uk/frameworks/cyber-essentials/) |
| [ISMS.online](/platforms/isms-online/) | [0](/platforms/isms-online/#c-ai) | Virtual Coach is videos, checklists and guides, and policy templates are pre-written. No AI agent, AI policy drafting, AI questionnaire drafting or AI review step: none found (read 18 Nov 2026). | [ISMS.online: virtual coach](https://www.isms.online/platform/features/virtual-coach/) [ISMS.online: plans](https://www.isms.online/plans/) [ISMS.online: self-assessment](https://www.isms.online/cyber-essentials/self-assessment/) |

## What earns points on AI that does the work under control?

Add: (a) AI or an agent that collects evidence or carries out compliance tasks, documented 3; (b) AI that drafts or tailors policies 2; (c) AI that drafts answers to questionnaires (security questionnaires or the scheme's own questions) 2; (d) documented human review or approval before AI output is submitted or used, or an audit trail of AI actions 3. An AI feature with no documented review step scores 0 on (d). Cap 10. Vendor-only evidence caps the total at 7.

The weight is 10 percent in the main ranking, and the five situations weight it at 5 percent (plus for a contract), 10 percent (first certificate for a bid), 10 percent (iso 27001 next), 10 percent (renewals across a growing estate), 15 percent (no it team).

## What would move the ranking on AI that does the work under control?

Scytale's score of 7 rests on 3 linked pages; Comp AI's 10 rests on 4. A change to a sourced fact on any of those pages is what would change a score; nothing else does. The top score here is worth 1.00 points of a vendor's total at the main weight.

## Sources

- [Comp AI: home page](https://www.trycomp.ai/) trycomp.ai, read 18 Nov 2026
- [Comp AI: security questionnaire](https://www.trycomp.ai/docs/security-questionnaire) trycomp.ai/docs/security-questionnaire, read 18 Nov 2026
- [Comp AI: AI policy editor](https://www.trycomp.ai/docs/ai-policy-editor) trycomp.ai/docs/ai-policy-editor, read 18 Nov 2026
- [TechCrunch: Comp AI and an agentic future for compliance](https://techcrunch.com/2026/09/17/comp-ai-sets-eyes-on-a-continiously-agentic-future-for-security-and-complaince/) techcrunch.com/2026/09/17/comp-ai-sets-eyes-on-a-contin..., read 18 Nov 2026
- [Sprinto: Sprinto AI](https://sprinto.com/sprinto-ai/) sprinto.com/sprinto-ai/, read 18 Nov 2026
- [Sprinto: pricing](https://sprinto.com/pricing/) sprinto.com/pricing/, read 18 Nov 2026
- [Sprinto: product updates, July and August 2026](https://sprinto.com/blog/sprinto-product-updates-july-august-2026/) sprinto.com/blog/sprinto-product-updates-july-august-20..., read 18 Nov 2026
- [Sprinto: security](https://sprinto.com/security/) sprinto.com/security/, read 18 Nov 2026
- [Secureframe: AI features](https://secureframe.com/features/ai) secureframe.com/features/ai, read 18 Nov 2026
- [Secureframe: MCP server announcement](https://secureframe.com/newsroom/secureframe-mcp-server) secureframe.com/newsroom/secureframe-mcp-server, read 18 Nov 2026
- [AIFN: Shrav Mehta of Secureframe on building AI in production](https://aifn.co/shrav-mehta-of-secureframe-on-building-ai-in-production/) aifn.co/shrav-mehta-of-secureframe-on-building-ai-in-pr..., read 18 Nov 2026
- [Secureframe: overview of AI capabilities](https://support.secureframe.com/en/articles/15111072-overview-of-ai-capabilities-in-secureframe) support.secureframe.com/en/articles/15111072-overview-o..., read 18 Nov 2026
- [Thoropass: MCP and AI](https://www.thoropass.com/solutions/mcp-ai) thoropass.com/solutions/mcp-ai, read 18 Nov 2026
- [Thoropass: compliance AI](https://www.thoropass.com/platform/compliance-ai) thoropass.com/platform/compliance-ai, read 18 Nov 2026
- [Vanta: AI](https://www.vanta.com/products/ai) vanta.com/products/ai, read 18 Nov 2026
- [Vanta: Cyber Essentials](https://www.vanta.com/products/cyber-essentials) vanta.com/products/cyber-essentials, read 18 Nov 2026
- [Vanta: AI principles](https://help.vanta.com/en/articles/11345368-vanta-ai-principles) help.vanta.com/en/articles/11345368-vanta-ai-principles, read 18 Nov 2026
- [Scytale: agentic compliance](https://scytale.ai/agentic-compliance/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) scytale.ai/agentic-compliance/, read 18 Nov 2026
- [Scytale: what is new in Q2 2026](https://scytale.ai/resources/whats-new-in-scytale-q2-2026/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) scytale.ai/resources/whats-new-in-scytale-q2-2026/, read 18 Nov 2026
- [Scytale: AI security questionnaires](https://scytale.ai/ai-security-questionnaires/?utm_source=cyberessentialssoftware.com&utm_medium=referral&utm_campaign=method) scytale.ai/ai-security-questionnaires/, read 18 Nov 2026
- [Drata: AI](https://drata.com/products/ai) drata.com/products/ai, read 18 Nov 2026
- [Drata: AI questionnaire assistance](https://drata.com/products/ai-questionnaire-assistance) drata.com/products/ai-questionnaire-assistance, read 18 Nov 2026
- [Scrut Automation: home page](https://www.scrut.io/) scrut.io, read 18 Nov 2026
- [Scrut Automation: Teammates](https://www.scrut.io/platform/scrut-teammates) scrut.io/platform/scrut-teammates, read 18 Nov 2026
- [Delve: home page](https://delve.co/) delve.co, read 18 Nov 2026
- [Delve: SOC 2 framework](https://delve.co/product/framework/soc-2) delve.co/product/framework/soc-2, read 18 Nov 2026
- [Delve: response to misleading claims](https://delve.co/blog/response-to-misleading-claims) delve.co/blog/response-to-misleading-claims, read 18 Nov 2026
- [Delve: inside its compliance process](https://delve.co/blog/inside-delve-trusted-compliance-process) delve.co/blog/inside-delve-trusted-compliance-process, read 18 Nov 2026
- [Hicomply: AI evidence discovery product](https://www.hicomply.com/product/ai-evidence-discovery) hicomply.com/product/ai-evidence-discovery, read 18 Nov 2026
- [Hicomply: Hicomply AI](https://www.hicomply.com/platform/hicomply-ai) hicomply.com/platform/hicomply-ai, read 18 Nov 2026
- [DataGuard: AI automation](https://www.dataguard.com/platform/ai-automation/) dataguard.com/platform/ai-automation/, read 18 Nov 2026
- [DataGuard: AI security questionnaire update](https://www.dataguard.com/product-updates/ai-security-questionnaire/) dataguard.com/product-updates/ai-security-questionnaire/, read 18 Nov 2026
- [DataGuard: frameworks and policies](https://www.dataguard.com/product/frameworks-policies/) dataguard.com/product/frameworks-policies/, read 18 Nov 2026
- [IT Governance: partnership with Defense.com](https://grcsolutions.io/grc-solutions-partners-with-defense-com-to-help-businesses-take-control-of-cyber-risk/) grcsolutions.io/grc-solutions-partners-with-defense-com..., read 18 Nov 2026
- [IT Governance: AI governance](https://grcsolutions.io/ai-governance/) grcsolutions.io/ai-governance/, read 18 Nov 2026
- [IT Governance: Cyber Essentials](https://grcsolutions.io/cyber-essentials/) grcsolutions.io/cyber-essentials/, read 18 Nov 2026
- [CyberSmart: Active Protect](https://cybersmart.co.uk/products/active-protect/) cybersmart.co.uk/products/active-protect/, read 18 Nov 2026
- [CyberSmart: security](https://cybersmart.co.uk/security/) cybersmart.co.uk/security/, read 18 Nov 2026
- [CyberSmart: Cyber Essentials](https://cybersmart.co.uk/frameworks/cyber-essentials/) cybersmart.co.uk/frameworks/cyber-essentials/, read 18 Nov 2026
- [ISMS.online: virtual coach](https://www.isms.online/platform/features/virtual-coach/) isms.online/platform/features/virtual-coach/, read 18 Nov 2026
- [ISMS.online: plans](https://www.isms.online/plans/) isms.online/plans/, read 18 Nov 2026
- [ISMS.online: self-assessment](https://www.isms.online/cyber-essentials/self-assessment/) isms.online/cyber-essentials/self-assessment/, read 18 Nov 2026

Last reviewed 18 Nov 2026.

## Keep reading

- [All seven criteria and weights](/method/)
- [Pricing transparency: all scores](/method/pricing/)
- [Continuous monitoring of the five control themes: all scores](/method/monitoring/)
- [Comp AI, highest on this criterion](/platforms/comp-ai/)
- [Scytale's profile](/platforms/scytale/)
