Introducing NanoRhino 1.0: An AI Nutrition Coach That Lives in Your Phone Contacts

Today we're shipping NanoRhino 1.0 — an AI nutrition coach you can text the same way you text a friend. No app to install. No download. No account creation flow with seven onboarding screens. You text START to a phone number, and a few seconds later, your personal coach says hi.

That's it. The number lives in your contacts. From now on, when you have a question about what to eat — or you snap a photo of your lunch — you text the coach the same way you'd text anyone else.

We think this is what AI-native software looks like, and we believe the era of "download our app" is winding down faster than most teams realize.

Why SMS, not an app?

The brutal truth about consumer apps in 2026 is that most people don't want another one. The average smartphone has dozens of installed apps and the user opens maybe seven of them a week. Asking someone who already feels overwhelmed by their relationship with food to download Yet Another Tracker, create an account, link their email, and figure out which tab does what is a tax most people quietly decide not to pay.

So they don't.

We watched this play out in our own data. Friction kills behavior change. Every step between "I want help" and "I have help" loses people. The leakiest funnel in health-tech is the one between intent and first useful interaction.

SMS removes that funnel entirely.

• Zero install — the messaging app is already on every phone, including the cheapest.
• No login — your phone number IS your identity. There's no password to forget.
• In your contacts forever — the coach isn't an icon you might delete next month. It's a contact, like your dentist or a friend.
• Async by default — you text when you have time. The coach replies when ready. No "are you online?" pressure.
• Photo logging works out of the box — MMS exists; everyone's phone supports it; we use it for meal photos.

The result is something that feels less like software and more like having access to a real coach who happens to be reachable 24/7.

What's actually happening when you text us

When you send START to our number, here's what happens behind the scenes (we'll get into the architecture below — but first, the user-facing version):

1. Within seconds you receive a confirmation SMS.
2. A dedicated AI agent is provisioned for you — its own private workspace, its own memory, its own ongoing relationship with you. It is not a shared chatbot. It is your coach.
3. A minute or two later, the coach reaches out: "Hi, I'm NanoRhino — what should I call you?"
4. From here on, every message you send goes to that same agent. It remembers. It recognizes your photos. It tracks your weight, your meals, your patterns, your wins, your slip-ups.

A week in, when you text "had pizza for lunch, oops 😅", your coach knows you actually ate well all week, knows your goals, knows you've been stressed about a work deadline, and responds the way someone who has paid attention would.

That continuity is the product.

Powered by Claude Opus 4.7 — the model actually matters

We built NanoRhino 1.0 on Anthropic's Claude Opus 4.7, the most capable model in Anthropic's current lineup as of this launch.

We chose Opus over the smaller, cheaper models for a specific reason: nutrition coaching is judgment work, not lookup work. The hard parts aren't "how many calories in a Whataburger Chili Cheese Double" (any model can answer that). The hard parts are:

• Recognizing emotional eating from a casual message and responding without lecturing.
• Looking at a photo and figuring out what you ate, in what portion, in what dish, with what side, to a credible accuracy.
• Knowing when to push, when to back off, when to ask "how was your week?" instead of "log that meal."
• Remembering that you said three weeks ago you don't eat dairy, and never asking again.

Smaller models cut corners on every one of those. Opus 4.7 doesn't, and the difference shows up in retention numbers we're not yet ready to publish but which made the model choice non-negotiable.

Vision is a first-class capability. Send a photo of your meal, and the coach actually looks at it — identifies the dishes, estimates portions, logs the entry, and replies in plain English the way a real coach would. No menu picker, no manual logging, no calorie-database lookups.

Built on OpenClaw — and we mean it

NanoRhino 1.0 runs on top of OpenClaw, an open-source agent framework. We're not using "open source" the way some companies do — as a loose marketing word for "we have a public GitHub repo nobody can actually run." OpenClaw is the load-bearing infrastructure for our entire production stack:

• Channel plugins — the same plugin system that drives our SMS, WeChat, and other channels is OpenClaw's plugin SDK. Nothing about our SMS handler is proprietary glue; it's a standard channel plugin in extensions/twilio/.
• Agent dispatch — every inbound message goes through OpenClaw's agent runtime: routing, session continuity, tool dispatch, history management.
• Skill system — coach behaviors (onboarding, weight tracking, meal logging, plan generation) are versioned skills, loadable independently. We can ship a behavior change without redeploying the gateway.
• Patches we contribute back — when we find issues that affect anyone running OpenClaw, we fix them upstream. The reply filter that protects users from agent reasoning leaks, the timing instrumentation that lets us debug latency at the per-stage level, the multi-channel registration flow with marketing source attribution — all upstream.

This matters for two reasons. First, our infrastructure is auditable: a researcher, a regulator, or a customer who wants to understand how their data is processed can read the actual code. Second, the work we put in to build a coach for ourselves benefits anyone else building agentic products on the same stack.

We think that's the right way to build in 2026.

The "AI-native" thesis, briefly

People keep asking what "AI-native" means in practice, and we have a short answer: AI-native software treats the model as the application, not as a feature inside an application.

The previous decade of consumer software was app-native. You opened an app. The app had screens. The screens had buttons. The buttons triggered backend calls. The backend used some statistics or some ML to enrich the data on the screens. Pinch to zoom, swipe to dismiss, tap to track.

That's a UI. Not a coach.

AI-native software starts from a different question: what would it look like if the user's primary interface were a relationship with an agent, not a navigation tree of screens? When you take that seriously, a lot of things flip:

• The interface is conversation. Not a button. SMS, voice, chat — the medium varies; the model is the constant.
• Memory replaces forms. The agent remembers what you told it. You don't fill out the same allergen field in three different settings menus.
• The screen, if there is one, is a side dish, not the main course. We do generate HTML meal plans on demand — and we send them as a presigned URL after the agent decides one would be useful. The plan is the output of a conversation, not the destination of a navigation flow.
• Onboarding is a conversation, not a form. When you text START, your coach asks for your name, your weight, your goals — one or two questions at a time, the way a human coach would.

The app era isn't over, but the next decade of consumer software won't be won by adding "AI features" to a screen-and-button paradigm. It'll be won by teams that pick a single relationship — coach, tutor, therapist, friend, assistant — and build the AI-native version of it from the ground up.

NanoRhino is our bet on that thesis, applied to nutrition.

Democratizing the dietitian

There's a more important story here than "we shipped some software."

A registered dietitian costs $100 to $300 per session in most US markets, and the wait for a first appointment is often weeks. Premium one-on-one nutrition coaching, the kind that includes weekly check-ins and meal photo review, runs $400 to $1,200 per month. For the vast majority of people who want to lose weight or eat better, that level of personalized support is not affordable, full stop. They end up with the cheapest substitute they can find — a calorie tracker, a Reddit thread, a YouTube channel — and a fraction of the support a real coach provides.

We think that's a market failure, not a market reality.

NanoRhino 1.0 isn't a dietitian. We don't claim it is. But what it offers — 24/7 availability, photo-based meal logging, personalized check-ins, memory of your goals and patterns, plan generation, judgment-free emotional support — used to be the exclusive product of a $200/hr appointment, and now lives in your contacts, accessible at the cost of an SMS.

That's not a marginal improvement. That's a category change. And it's the most exciting thing about shipping 1.0: the people who get the most out of this are the ones who could never afford the version it replaces.

Try it

If you're in the United States and you've read this far, you can try NanoRhino 1.0 right now:

1. Text START to +1 (915) 277-7888
2. Wait 1-2 minutes for your coach's first message
3. Reply naturally — tell it your name, your goals, what you ate today

If you came from a specific channel and want us to know, you can also text START GPT or START WEBSITE (we use the source token for analytics, nothing more — and you can opt out anytime by replying STOP).

Photo logging works from day one: snap a picture of your meal and send it. Your coach will recognize it, log it, and respond with the kind of feedback you'd expect from a coach who's actually paying attention.

What's next

1.0 is the foundation, not the destination. Things we're working on for the rest of 2026:

• Adaptive plan generation — meal plans that revise themselves based on what you actually ate this week, not what you said you'd eat last month.
• Multi-channel coach continuity — start a conversation on SMS, finish it on WeChat or web; one agent, one memory.
• Pay-for-results pricing — we're piloting a model where you pay only when you hit your goals, not for the privilege of being signed up.
• Open contribution — more of NanoRhino's coaching skills will land in OpenClaw's public skill registry so anyone building an agentic health product has a starting point that isn't blank.

We're shipping 1.0 confident in the foundation but very aware that the interesting work is just starting. If you've got feedback after using it for a week, your coach will pass it along — that's literally one of the things they're built to do.

---

Try NanoRhino 1.0 now: text START to +1 (915) 277-7888.