Claude Dispatch Lets You Text AI From Any Phone

What’s Happening

A new project called Claude Dispatch is making waves this week among AI power users and builders. The concept is deceptively simple: text Anthropic’s Claude from your phone — no app download, no browser tab, no login screen. Just send an SMS to a designated number and get Claude’s response back as a text message. It strips the AI experience down to the most universal mobile interface on Earth: the humble text message.

Dispatch isn’t an official Anthropic product. It appears to be a third-party integration layer, likely built on top of Anthropic’s public API and a telephony service like Twilio or Vonage, that routes SMS messages to Claude and returns the model’s replies in real time. The project has been circulating on developer forums, Product Hunt, and X (formerly Twitter) over the past few days, with early users praising its speed and the almost absurd frictionlessness of the experience. You don’t even need a smartphone — a basic feature phone will do.

The timing matters. We’re at a peculiar inflection point in the AI industry where the models themselves have become remarkably capable, but distribution and accessibility remain bottlenecks. Most people still interact with AI through dedicated apps or web interfaces. Claude Dispatch, and projects like it, are a bet that the next frontier isn’t model intelligence — it’s model access. And that the winners will be the ones who meet users where they already are.

Why Founders Should Care

Let’s talk numbers. According to Pew Research Center, 97% of Americans own a cellphone of some kind, and SMS remains the single most universally accessible communication channel on mobile devices. SMS open rates hover around 98%, compared to roughly 20% for email. People read texts. Almost always. Almost immediately.

Now layer in the economics of AI APIs. Anthropic’s Claude API pricing has dropped steadily over the past year, making it feasible to build lightweight wrapper services that serve thousands of users without enormous compute bills. The cost to send and receive SMS via platforms like Twilio is pennies per message. The margins on an SMS-to-AI bridge — especially one that charges a small subscription fee — are surprisingly healthy once you hit scale.

But the real signal here isn’t about one project. It’s about a distribution paradigm shift. For the past two years, the AI industry has been obsessed with building bigger models and flashier interfaces. Claude Dispatch flips that script. It says: forget the chrome, forget the UI polish, just give people the intelligence through the channel they already use 80 times a day. For founders, this is a massive opportunity space — not just in SMS, but in any “dumb pipe” channel that can suddenly become smart: email, WhatsApp, iMessage, even voice calls.

“The most powerful technology is the one people actually use. And everyone uses text messages.” — A recurring sentiment among SMS-first AI builders on developer forums this week.

If you’re building in AI right now, the lesson from Dispatch is clear: interface innovation is just as important as model innovation. Maybe more so. The next billion AI users won’t come from people who download yet another chatbot app. They’ll come from AI meeting them inside the tools they already have open.

Who’s Already Moving

Claude Dispatch isn’t operating in a vacuum. Several startups and projects have been exploring SMS-first and messaging-first AI interfaces, and the space is heating up fast.

• Dispatch (Claude Dispatch) — The project making news this week. It routes SMS messages directly to Anthropic’s Claude API and returns responses via text. Early reports suggest it supports multi-turn conversations, meaning Claude remembers context within a session. The developer behind it appears to be an independent builder, though details on the team remain sparse. Its simplicity is its strength — and potentially its vulnerability if Anthropic decides to build this natively.
• Buzz (buzzchat.ai) — A Y Combinator-backed startup that launched in late 2025 offering AI assistants via WhatsApp and SMS. Buzz targets emerging markets in Latin America and Southeast Asia where app fatigue is real and data plans are expensive. They’ve reportedly crossed 100,000 monthly active users by leaning into local-language support and ultra-low-bandwidth delivery. As TechCrunch noted in its coverage of the Y Combinator W26 batch, messaging-first AI startups were one of the most represented categories.
• Quill AI — An enterprise-focused startup that lets teams interact with AI models through Slack, Microsoft Teams, and SMS. Quill’s pitch is that knowledge workers shouldn’t have to leave their existing workflows to use AI. They raised a $6M seed round in late 2025 and are targeting mid-market companies with 50-500 employees.
• Anthropic itself — It’s worth watching what Anthropic does here. The company has been expanding Claude’s reach aggressively — through its own consumer app, the API, Amazon Bedrock, and Google Cloud partnerships. An official SMS interface would be a natural extension. If Anthropic decides to launch its own Dispatch-like feature, third-party projects could get squeezed overnight. This is the classic platform risk that every wrapper startup faces.

The Startup Opportunity

If you’re a founder reading this, here’s where it gets actionable. The Claude Dispatch trend reveals several specific startup opportunities worth exploring:

• Vertical SMS-AI assistants. A general-purpose Claude-via-SMS is cool, but the real money is in vertical applications. Imagine a farming assistant that texts crop advice based on local weather data. A medication reminder bot that also answers health questions via Claude. A financial literacy coach for underbanked populations. These are real businesses, not just weekend projects. The channel (SMS) provides distribution; the vertical focus provides defensibility.
• AI-powered SMS for small businesses. Most small businesses already use SMS to communicate with customers — appointment reminders, order confirmations, promotional texts. Now imagine an AI layer on top: a restaurant that lets customers text questions about the menu and get intelligent, context-aware responses powered by Claude or GPT-4. The startup that builds this as a plug-and-play platform for SMBs could own a massive market.
• Multi-model SMS routing. Why limit users to one AI? Build an SMS gateway that intelligently routes queries to the best model — Claude for nuanced analysis, GPT-4 for code generation, Gemini for multimodal tasks. The user just texts a question; the system figures out which AI handles it best. This is a thin but potentially very sticky middleware play.
• Accessibility-first AI. There are approximately 2.5 billion feature phone users worldwide who have zero access to app-based AI. An SMS-first AI service targeting these users — particularly in Sub-Saharan Africa, South Asia, and rural markets globally — could be transformative. The unit economics are challenging at scale, but the social impact and potential for NGO/government partnerships are enormous.

The common thread: don’t build another chatbot app. Build the infrastructure and vertical solutions that deliver AI through channels people already trust and use daily.

Watch Out For

Alright, let’s be honest about the risks — because there are several that could trip up founders in this space.

• Platform risk is the elephant in the room. If you’re building on Anthropic’s API, you’re one policy change or rate-limit adjustment away from a bad day. Anthropic could launch its own SMS product, raise API prices, or restrict third-party use cases that it considers too close to its own consumer offering. This isn’t hypothetical — it’s happened with OpenAI and it’ll happen again.
• Carrier and regulatory friction. SMS isn’t the Wild West it used to be. Carriers increasingly require 10DLC registration (10-digit long code) for application-to-person messaging in the US. Violations can lead to message filtering or outright blocking. TCPA compliance is non-negotiable — one lawsuit from an unsolicited AI text could sink a startup. International regulations vary wildly. If you’re targeting emerging markets, you’ll need to navigate a patchwork of telecom rules.
• Latency and context limits. SMS has a 160-character limit per segment (longer messages get split and can arrive out of order). Claude’s responses tend to be thorough — sometimes too thorough for a text-message format. Managing conversation context across SMS sessions is technically non-trivial, especially without persistent sessions on the carrier side. User experience can degrade quickly if responses feel choppy or truncated.
• Cost at scale isn’t trivial. Twilio charges roughly $0.0079 per SMS segment sent in the US. A single Claude response that spans 5-6 SMS segments, plus the inbound message cost, plus the API call — that’s $0.05-0.10 per interaction before you’ve paid for anything else. Free tiers will burn cash fast. You need a clear monetization path from day one.
• Security and privacy. SMS is not encrypted end-to-end. Sending sensitive queries — medical, financial, legal — over SMS creates real liability. Users may not understand this. If a data breach or interception incident hits the news, the entire SMS-AI category could face a trust crisis.

The Bottom Line

Claude Dispatch is a small project with a big lesson baked into it. The AI industry’s next wave of growth won’t come from better models alone — it’ll come from radically simpler distribution. SMS is just the beginning. Email, voice, WhatsApp, RCS, even USSD codes in emerging markets — every one of these channels is a potential AI interface waiting to be built. The founders who figure out how to deliver genuine intelligence through the world’s simplest communication tools will have a distribution advantage that no amount of VC-funded app marketing can match.

So here’s what to do: stop thinking about AI as a product that lives in a browser tab. Start thinking about it as a utility — like electricity or water — that should flow through whatever pipe is already connected to the user. If you can identify a specific audience, a specific channel they already trust, and a specific problem that AI can solve in that channel, you’ve got the bones of a real business. Just make sure you own enough of the stack to survive the inevitable platform shifts. Build fast, but build smart.