AI 2030 Accelerator · Founder assessment
Startup AI Safety
Readiness Assessment
Are you ready for enterprise buyers, investors and responsible AI scrutiny? Get a score, a readiness level, three audience scores and five priority actions.
Three audiences, one assessment
See how you look to the people who decide your future.
Can you pass procurement, security and AI diligence and stand behind your contracts?
Can you show defensible, well-governed AI and a credible risk story?
Would your practices hold up to regulators, press and civil society?
First, tell us what is in scope
We’ll tailor the assessment to your AI environment.
Choose “Not applicable” on any question that does not fit your company; it is left out of your score.
We develop or fine-tune AI systems or models.
We use third-party AI products or APIs.
AI materially affects customers, employees or other people.
We use autonomous or agentic AI.
Seven domains
What the startup assessment covers.
01Product Purpose & Risk3 questions
- We can clearly explain our AI product’s intended use, users, benefits, limitations and material risks.
- We have identified foreseeable misuse and high-consequence failure scenarios.
- We know which customer or regulatory contexts could make our product higher risk.
02Data & Model Foundations3 questions
- We can explain where material training, tuning, retrieval and evaluation data comes from and how it may be used.
- We have controls for sensitive data, privacy and customer data separation.
- We understand important model dependencies, third-party providers and associated risks.
03Evaluation & Safety3 questions
- We test the product against defined performance and safety criteria before major releases.
- We evaluate relevant failure modes such as hallucination, bias, harmful outputs, security abuse or agentic misuse.
- We can show enterprise buyers evidence of our testing rather than only describe it.
04Governance & Ownership3 questions
- A named leader is accountable for responsible AI and product-risk decisions.
- We document significant risk decisions, exceptions and mitigations.
- We have a practical process for deciding when a feature or use case should not ship.
05Enterprise Buyer Readiness3 questions
- We can answer common enterprise diligence questions about security, privacy, governance, data and model risk.
- Our customer documentation clearly states intended use, limitations, human oversight expectations and support processes.
- We can support contractual commitments without promising controls we do not actually operate.
06Monitoring & Incidents3 questions
- We monitor deployed systems for quality, abuse, drift and emerging risks.
- Customers have a clear path to report serious AI issues.
- We can investigate incidents, remediate them and communicate material changes.
07Team & Market Trust3 questions
- The team receives role-appropriate responsible AI and security training.
- We can explain our responsible AI approach credibly to customers, investors and partners.
- Responsible AI work is tied to product quality and market access, not treated only as compliance.
How to answer
Answer for what is operating today.
Choose the option that best reflects what is operating today, not what is intended. “Measured / evidenced” means the practice is consistently operating and you can produce evidence. This is a maturity benchmark, not a certification or legal-compliance determination.
Built for founders
Become buyer-ready and investor-ready.
Use your results to prepare for enterprise diligence, then go further with the AI 2030 Accelerator.