Partners
Join the standard.
ioLite Labs is not building another AI tool. We are building the evaluation infrastructure the AI industry is missing. We are looking for partners who understand what that means — and want to help define it.
Partnership Types
Platforms and labs
For: AI platforms, model catalogs and labs that serve the public.
Put the independent standard inside your stack. A third-party measurement is what makes your own safety work believable to regulators, courts, insurers and enterprise buyers.
- —A 90-day pilot, or a two-quarter build with ioLite Gateway
- —Study 02 on your hosted models, with co-published results
- —The certification seal first, and a seat shaping the standard
Benchmark partners
For: Benchmark programs, evaluation infrastructure teams and the labs that run them.
A psychological-safety benchmark that already exists, built the way a benchmark has to be built to be trusted: held-out scenarios, the scorer apart from the system, open traces, repeated runs with the variance reported, and a clinician gold set before any leaderboard.
- —Run it at scale: frontier-model inference and compute for Study 02
- —Host it in the open, on sandboxed and logged infrastructure
- —Co-design the public release, and fund the clinician validation
- —The full method and integrity rules are on the benchmark page, under Research
Research Partners
For: Academic institutions and research organizations.
Contribute to the methodology that will define behavioral AI safety evaluation. Your research will inform the standard at scale — not just one product.
- —Co-develop scenario libraries
- —Collaborate on scoring methodology
- —Co-publish in the emerging field of AI behavioral safety
Clinical Partners
For: Mental health professionals, clinical organizations, crisis intervention experts.
Your domain expertise is what separates a real safety standard from a marketing checklist. Ground our evaluation in clinical reality.
- —Review and validate scenario design
- —Advise on safe-messaging protocol alignment
- —Shape criteria that protect users at scale
Strategic Investors
For: VCs, family offices, and strategic corporate investors.
Independent evaluation of conversational AI is becoming a field. What it lacks is a commercial audit and certification layer, mapped to the law and renewed on every model version. That is what we are building.
- —A field that is forming, with the audit layer still open
- —Regulatory tailwind accelerating timeline
- —Compounding methodology and data moat
Where ioLite fits
Independent evaluation of conversational AI is becoming a field: everyone.AI's AïA scores single responses to adolescents, Oxford and UCL's SIM-VAIL showed risk accumulating across multi-turn conversations, and Common Sense Media rates consumer AI products.
ioLite is the commercial audit and certification layer beside that work: full conversations, every age, nine safety gates, and results mapped to the law.
We welcome interoperability with open evaluation efforts and plan to report alongside open frameworks as they publish their tooling.
The full landscape, with sourcesStandards and research community
We follow and seek to contribute to open efforts on AI and wellbeing, including the iRAISE coalition.
iRAISEThe Investment Case
A field is forming. The audit layer is still missing.
AI systems are deployed in emotionally sensitive conversations at scale, and the laws that govern them are arriving state by state. Researchers and nonprofits are starting to evaluate these systems; what deployers still lack is an independent audit they can hand to counsel, buyers and regulators.
ioLite Labs is building that layer: full multi-turn conversations, every age, nine fail-and-stop safety gates, and results mapped requirement by requirement to the law.
The methodology, the evidence library and the audit infrastructure compound with every engagement, and certification renews on every model version.
69
Clinical sub-constructs in the taxonomy
9
Safety-gate constructs
4
Validated source frameworks
Study 01
Measured; publishing after clinician review
Get in Touch
Partner inquiry.
Tell us about your organization and what kind of partnership you have in mind. We respond to every inquiry within two business days.
— Technology and research partnerships: evaluated on alignment with methodology goals
— Investment inquiries: reviewed by founding team, briefing scheduled within one week
— Clinical advisors: matched to active evaluation projects