Mindgard
AI security platform that maps, attacks and helps fix vulnerabilities in models and agents with continuous automated red teaming
$30M
Series A (Aug 2026)
SOC 2
Type 2
Lancaster
University Spin-out
Overview
Mindgard is a commercial AI security platform that "finds, validates, and fixes high-impact AI vulnerabilities". It works the way an attacker would: Recon maps your models, agents and tools, Attack runs adaptive attacks to prove what is exploitable, Remediate suggests fixes, and Assure re-runs tests whenever the AI system changes. It also offers runtime protection and governance reporting. The company is a spin-out of Lancaster University, raised a £3M seed in 2023 and a $30M Series A led by Album VC in August 2026, and runs offices in London and Boston.
The Verdict
Who Should Use Mindgard?
Best For
- Enterprise security teams that need continuous testing of many AI apps and agents
- AppSec teams that already work in Burp Suite and CI/CD pipelines
- Buyers who need a vendor with SOC 2 Type 2 and support
Not Ideal For
- Small teams or hobby projects (no public self-serve plan)
- Teams happy to run free tools such as garak or PyRIT themselves
- Buyers who need transparent list pricing up front
What's Great
- Covers the full loop: recon, attack, remediation and re-testing on every change
- Tests agents and their tools, not only single model prompts
- Integrates via API, CI/CD and Burp Suite
- Deep academic roots in Lancaster University AI security research
Watch Out For
- No published prices; every plan starts with a demo or sales call
- Enterprise focus; few independent public reviews so far
- Leadership titles differ between older and newer press coverage
Pricing
View all features & details
Key Features
- Recon: attacker-style mapping of models, agents and tools
- Attack: adaptive attacks that validate real exploitability
- Remediate: fix guidance and actions
- Assure: continuous re-testing triggered by AI changes
- AI runtime protection and threat response
- AI governance and compliance reporting
Company
- Seed: £3M (2023), IQ Capital and Lakestar
- Series A: $30M (Aug 2026), led by Album VC
- Offices in London and Boston
Use Cases
- Continuous red teaming of production chatbots and agents
- Security sign-off before an AI release
- Evidence for AI security audits
How It Compares
| Feature | Mindgard | garak | PyRIT | Promptfoo |
|---|---|---|---|---|
| Delivery | Managed platform | Open-source CLI | Open-source library | Open-source CLI |
| Agent and tool testing | Yes | Model and endpoint focus | Via custom targets | Via test configs |
| Runtime protection | Yes | No | No | No |
| Support and SLA | Vendor | Community | Community | Community |
| Price | Contact sales | Free | Free | Free (OSS) |
Loading reviews...
Is this your tool? Add a badge