Needle, a startup based out of Salt Lake City, helps patients find hard-to-locate medications that are out of stock in their area. Instead of forcing people to spend hours calling pharmacies themselves, Needle does the work for them.

Behind the scenes, this means calling a lot of pharmacies. A single medication search often requires reaching out to 30 to 50 locations, sometimes more. At scale, that level of calling creates both an operational and economic challenge for their business.

The Challenge

Before automation, Needle’s calling operation was entirely manual.

“We started doing everything manually at first,” said Brennan Staheli, CEO of Needle. “That obviously wasn’t sustainable.”

Even after moving to offshore labor, the numbers still did not work. Each human call cost Needle roughly $2 to complete. When multiplied across tens of thousands of searches a month, the business quickly became cost-prohibitive.

“Our product wouldn’t exist if we didn’t have AI to do the calls,” Brennan said. “Full stop.”

Cost was only part of the problem. Pharmacies use inconsistent IVR systems that frequently change. Calls often require navigating multiple menus just to reach the right department.

Needle needed a solution that could reliably handle IVR navigation, collect structured data, and integrate directly into their custom backend without adding operational overhead.

The Solution

After researching several voice AI providers, Needle chose Bland to handle these phone calls. “One of the big factors for choosing Bland was how easy it was to use the API,” Brennan said. 

Because of this, implementation was fast. “I got Bland up and running in production within just 48 hours,” Brennan said. “It’s really, really easy to place a phone call out of Bland. For an engineering minded team like ourselves, we love the Bland platform.”

Needle built a fully automated outbound calling pipeline on top of Bland’s API. When a user submits a request in the Needle platform, the system pairs that request with a pharmacy-specific prompt and dispatches calls automatically.

“We maintain a database of custom prompts,” Brennan explained. “When we dispatch, we look at which pharmacy we’re calling and pair it with the right prompt.”

Those prompts are designed to navigate IVRs and collect specific data points including medication dosage, form, and quantity. Each call returns structured results that flow directly back into Needle’s system.

“Our ops folks are the ones listening to calls most often,” Brennan said. “They’re empowered to make edits to the prompts when IVRs change. The people closest to the problem own it, and we love that.”

The Outcome

Today, Needle handles about 2,000 calls per day, totaling roughly 60,000 calls per month. Since launch, the company has completed over 800,000 calls on Bland.

“About 81 percent of every call we make to check pharmacy inventory gets fully resolved with Bland,” Brennan said. That’s 81 percent of all calls that get completed end to end without any human involvement.

The cost difference for this is significant. Human calls typically cost Needle about $2 each. AI calls through Bland cost them 92% less than that.

At their current volume, handling all 60,000 monthly calls with human agents would cost roughly $1.2M per year. With Bland handling 81% of those monthly calls, that number drops significantly, resulting in nearly $1M in savings every single year.

Beyond savings, Bland helps deliver consistency. IVRs are navigated reliably. Data is collected the same way every time, meaning patients get clear answers faster without spending hours on hold.

The Future

Needle remains focused on delivering excellent medication access, but the infrastructure they’ve built with Bland gives them room to expand.

For now, the priority is scaling what already works. With a developer-friendly API, a reliable IVR navigation system, and AI handling over 80 percent of all calls, Needle has built a scalable engine for outbound calling.

“Our product doesn’t exist without this,” Brennan said. “Bland makes the entire business possible.”