Arizona Built a Chatbot to Solve the Problem Rize Was Founded to Solve

Ask “where can I find treatment near me” into a text box in Arizona now, and a state Medicaid agency’s chatbot answers you back in plain language, in your own language if it isn’t English, with a pin dropped on a map. AHCCCS launched the tool in July with Google Public Sector and Syntasa, built on Gemini and Vertex AI, connecting people to more than 100 vetted opioid-use-disorder providers through the Opioid Service Locator. Since launch it’s logged over 100,000 page views, a 55%-plus engaged-session rate, and reached more than 20,000 unique people across 120-plus Arizona cities.

A state Medicaid agency just spent public money proving the exact thesis Rize was founded on: that “find treatment” is a navigation problem, not just a data problem.

The pitch is almost identical to what any AI-guided treatment finder promises: don’t make someone who’s already struggling parse a static directory built for caseworkers. Let them ask a question the way they’d ask a friend, and get an answer that accounts for where they are, what language they speak, and how they’re trying to reach it — a phone in one hand, maybe standing in a parking lot. Kate Dobler, AHCCCS’s State Opioid Treatment Authority coordinator, put the goal plainly: “Finding treatment should never be complicated.” Google’s Reymund Dumlao framed it as closing “health disparities” — the multilingual, mobile-first design is aimed squarely at people a paper referral sheet or a desktop-only directory was never built to reach.

None of that makes the tool a competitor in the sense that matters most. It’s scoped narrowly to opioid-use-disorder providers, it doesn’t carry insurance matching or a peer-support layer, and it’s a locator, not a navigator that follows someone from crisis through months of aftercare. What it is, though, is confirmation from the inside of the system: the state agency that runs Arizona’s Medicaid program looked at the existing landscape of treatment directories and concluded the gap was real enough to fund a fix. That’s a state validating the market, not filling it.

The number worth sitting with isn’t the page views. It’s the 20,000 people who used a chatbot instead of a phone book, in a state that ranks near the bottom nationally for behavioral health access. Wherever they landed next, they got there because someone finally built the door to look like a question instead of a directory.