MyTrusted.ai
Chat with top AI models, with safety you can see.
Start new thread

MyTrusted.ai - Chat with top AI models, with safety you can see.

by•
MyTrusted.ai is an AI chat app that lets you use several leading AI models in one place, with SASKI safety rules running on every message. SASKI is deterministic: it catches crisis signals and points people to 988 resources, and removes personal details like phone numbers and addresses before they reach the AI, the same way every time. Every safety decision is recorded without storing your words. The first 100 sign-ups get 2 million free tokens. Learn more at

Add a comment

Replies

Best
Maker
📌

Hi Product Hunt 👋

I'm Stephen, a parent, a ski coach at Sugar Bowl Academy in Truckee, and the founder of SASKI Institute.

I started SASKI because I wanted AI that I could trust around my own kid. MyTrusted.ai itself is for adults 18+, and it's where anyone can see SASKI's safety rules at work. Most AI safety lives inside the model, so it changes every time the model changes. SASKI takes a different approach: deterministic rules that run alongside the AI and behave the same way every time.

What you can do with it: 🗨️ Chat with several leading AI models (Claude, Gemini, ChatGPT, etc) in one app, and compare replies (the first 100 sign-ups get 2 million free tokens) 🛟 See SASKI catch crisis signals and point people to 988 resources 🔒 SASKI helps remove personal details like phone numbers and addresses before they reach the AI 🧾 Every safety decision is recorded, without storing your words 🧪 Run the built-in safety test prompts to see exactly what it catches.

Who it's for: adults who want AI chat with visible safety rules, and anyone curious how AI safety can be checked instead of just promised. If you build AI products for patients, students or people in crisis, this is a hands-on look at the safety layer behind it.

Two ways to use MyTrusted.ai, and what we can see:

🔐 Private chats (the default): your conversations are private. We never see or store what you type or what the AI says back. Your messages go to the AI model you pick, with SASKI removing personal details first, and the only thing we keep is a record of SASKI's safety decisions, without your words.

🧪 Test Conversations (you choose to start one): this is the only place we see the prompt and the AI's reply. Test Conversations are linked to a random tester ID, not your name, and we use them to find where SASKI's rules hold and where they miss. Every miss becomes a rule improvement that ships to every app running SASKI, so a few test prompts now and then genuinely help make AI safer for everyone.

Don't just take my word for it: open your browser's DevTools, go to the Network tab, and watch exactly what leaves your browser in each mode.

The safety loop: an AI acts, SASKI checks it before it runs, the decision gets a signed receipt, someone independent can review it, and the rules get better. That last step is why your tests count.

Why it matters now: the world is starting to ask for proof, not promises. Utah's AI sandbox now requires tamper-evident records. Connecticut's new law requires AI companions to detect self-harm language and refer people to help starting in January. Lloyd's-backed insurers are underwriting AI with evidence of real controls, and the NAIC and state insurance regulators are building ways to check how AI is governed. SASKI is built to produce exactly that kind of evidence.

What I'd love from you: try to break it. Run the test prompts, write your own, and tell me what SASKI misses or gets wrong. Every miss makes it better.

A big thank you to my co-founder Ishak Kang, who helped shape the design of MyTrusted.ai.

Thanks for checking it out. I'll be here all day answering questions.

~ Stephen

Watch our demo video here: