I’ll close with the case I’d make to an investor, because I think specialty pharmacy is one of the most compelling vertical AI opportunities hiding in plain sight — and “hiding” is the operative word. Let me lay out why this is a category and not just a feature.
The market is big, growing, and structurally phone-dependent
Start with size. Specialty pharmaceuticals dispensed roughly $293 billion in 2025, up nearly 10% year over year, with the broader market around $285 billion and forecast to keep compounding at a steep rate into the 2030s. More than 80 new specialty drugs are slated to launch from 2025 to 2027, many of them cell and gene therapies with per-patient costs into the millions. This is not a mature, flat market you’re trying to squeeze. It’s growing double digits, and every new high-cost therapy adds more of exactly the operational and communication burden that voice AI addresses. The TAM expands as the drug pipeline expands.
And critically, this industry runs on the phone. Outbound and inbound calls are the connective tissue of specialty pharmacy — verification, adherence, refills, delivery, monitoring, after-hours triage. That makes voice AI not a peripheral tool but a wedge into the core operational workflow. The closer your product sits to the core workflow, the more durable the business.
The pain is quantified and acute
I’ve spent this whole collection putting numbers to the pain because investors should demand them. Non-adherence: $528 billion a year, 100,000-plus deaths. Prior auth: 13 hours per physician per week, 94% reporting harm to outcomes. Cold chain: $35 billion in annual failures, six figures per excursion. Time-to-therapy: a critical 48-hour window gated by week-long verification. Call centers: 30% abandonment after a minute on hold. And the kicker — one independent pharmacy doing 80,000 calls a month by hand. This isn’t a vitamin. It’s a painkiller for problems that are measured in lives and billions of dollars.
The market structure is favorable
Specialty pharmacies are largely independent and fragmented today. For a startup, fragmentation is a gift: there’s no single incumbent gatekeeper to displace, you can land individual pharmacies on their own merits, and — because they all share the same operational pain — each successful deployment becomes a reference that travels through a tight-knit industry. I’ve watched how fast word moves at gatherings like Asembia AXS. In a market like this, proof compounds.
The moat is the part that scares generalists off
Here’s why I don’t lose sleep about the crowded “AI call center” landscape. The general voice AI market is brutally crowded — lots of companies that all sound the same, fighting over restaurants and generic CX. Specialty pharmacy is the opposite, and it’s defended by exactly the things that make it hard: deep integration with niche pharmacy management systems, the clinical and emotional bar for talking to vulnerable patients, and a serious regulatory surface — accreditation, REMS, adverse-event reporting, privacy. A generalist vendor can’t just point their restaurant agent at a specialty pharmacy and win. The difficulty is the durability. Whoever does the hard, unglamorous work of building for this vertical earns a position that’s genuinely defensible.
The business model is sound
This is recurring, usage-based revenue tied to mission-critical workflows, with pricing models — concurrency-based, in particular — that scale predictably with the pharmacy’s volume. Specialty prescriptions are high-value, so the ROI story to the buyer is strong, which supports healthy pricing and low churn. And once an agent is integrated into a pharmacy’s system of record and trusted with its nights and its patients, switching costs are real. High value to the customer, strong retention, expanding TAM — that’s the profile you want.
The honest risks
I’d be doing you a disservice not to name them. Sales cycles in healthcare are slow; pharmacies move deliberately, and you have to be patient with the pipeline. Integration work is real engineering, not a weekend. Regulation means you have to build carefully and conservatively from day one. And reputational risk is asymmetric — one bad clinical interaction can set the whole category back, so safety-first design isn’t optional. These are real, but notice that every one of them is also a barrier to entry. The risks and the moat are the same wall, seen from different sides.
The thesis, one more time
A massive, double-digit-growing, phone-dependent industry, with quantified and acute operational pain, a fragmented and winnable market structure, a regulatory-and-integration moat that repels generalists, and a recurring revenue model tied to mission-critical work. That’s not a feature looking for a market. That’s a category. I bet my company on it after one conversation about 80,000 calls — and the more I’ve learned, the more convinced I’ve become that vertical voice AI for specialty pharmacy is one of the clearest opportunities in applied AI right now. Most people still don’t know it’s there. That’s exactly why it’s worth backing.