AI in Healthcare Today: Customer Obsession, Building Long-Term Moats and Getting to Series A

In Conversation: Trey (CEO and co-founder of Tennr)

By Nisha Dua, Claire Biernacki & Cindy Fan, BBG Ventures

The U.S. spends nearly $5 trillion a year on healthcare — nearly 20% of GDP and almost double the per-person spend of the next highest country. It’s also our biggest employer, with over 22 million healthcare workers. And yet, somehow…we’re still doing manual tasks, still faxing things, and still filling out intake forms and sharing our medical history at every visit.

The system is huge, messy, and deeply broken — making it ripe for disruption with AI. At BBG, we believe there’s a massive opportunity to bring in personalization across the care continuum, from doctor-led care to collaborative (e.g., disease management and routine care) to autonomous and continuous care. We can prove real ROI from both the administrative and clinical side of how care is delivered.

To dive deeper into strategies for building in healthcare AI and delivering on the promise of personalization, we partnered with our friends at Lightspeed, NEA, GV, AlleyCorp, and Next Ventures and supported by Silicon Valley Bank — for a conversation with Trey Holterman, co-founder and CEO of Tennr, which streamlines administrative workflows in healthcare.

Tennr’s product sounds deceptively simple: it processes medical documents (yes, the faxed kind), and streamlines tasks such as patient referrals, insurance checks, and appointment scheduling. But it’s now deeply embedded into workflows — integrating with EHR systems and allowing healthcare providers to adopt the tech without overhauling their current infrastructure. Today, Tennr’s proprietary learning model has been trained on 100M medical documents, 2.3B distinct data fields, and 8000 sets of criteria to interpret complex and unstructured healthcare data, including handwritten notes. This breadth of data ingestion enables Tennr to capture granular patient-level information and institutional nuances, powering highly accurate automation tailored to specific workflows. As a result, customers have reduced insurance claim denials, cut pre-visit patient processing from weeks to hours, and converted more referrals to patients — not only streamlining operations but also enabling more timely, personalized, and accurate care delivery.

Read on for some of the best insights from the discussion — covering everything from customer discovery to building a moat in this rapidly evolving and increasingly competitive market.

From Stand-Up to Start-Up

Trey joined Strava out of Stanford and was doing improv comedy on the side. He didn’t know the first thing about EHRs or prior auth, but he did know how broken the referral system was. He’d seen his mom, who works in family medicine, send patients to Bay Area imaging centers with little response or follow through. This insight led to his hypothesis for Tennr.

Trey’s outsider lens turned out to be a superpower — rare for those building in healthcare. From the beginning, Trey had a focus on building for “real problems”. While others were aiming to replace legacy systems, Tennr took a different strategy: “make it dumb and simple.” Instead of trying to fix the system, which is timely and expensive, “what if we just made the existing system work better”, working within current infrastructure?

Improv, it turns out, was great training for healthcare. You learn to listen obsessively, respond in real time, and adjust your plan based on what’s actually happening. That mindset shaped how Tennr approached the product: working within existing workflows and adapting quickly.

Referrals move through healthcare mostly via…fax. Pre-Tennr, some unlucky staffer has to manually process them: verify insurance, match the patient to an appointment slot, handle errors, etc.

The big unlock: if the receiving provider can make more money by processing faxes better, maybe they’ll actually use this software.

So the Tennr team built a document parser trained on messy, real-world data to do exactly that — starting with a single question centered on ROI: Can we reduce denials and boost conversion?

Takeaway: Don’t underestimate what your unique background brings to the table. Build for real, observed pain, especially the messy and unsexy ones others overlook.

Healthcare sales: Listening to the Customer and Driving ROI

Early on, Tennr falsely assumed that each imaging center would have similar workflows. One of the company’s first deployments failed because the customer had no IT team and highly customized processes, which made Tennr’s initial product impossible to implement.

Similar to our vertical AI conversation with Nick at Topline Pro, the turning point came from customer obsession. When Trey shifted from pure engineering and heads-down coding to obsessive customer listening — “the success of the company and the number of customer meetings on my calendar became directly correlated”. Their customer interviews showed that they needed to build a hyperconfigurable product — one that let them adapt fast to customer needs. Hyperconfigurability also created customer stickiness and high switching costs — ultimately leading to advantage and moat in an increasingly competitive space.

Healthcare is a notoriously hard market to sell-into, with risk-averse buyers, complex workflows, and a tough regulatory landscape. In the beginning, Trey primarily sold the promise and hypothesis of increasing conversion and reducing denials. Their pitch focused not just on cost cutting but the opportunity to improve topline revenue. Tennr’s sales cycle began to accelerate when they were able to lead with examples of ROI and case studies in the pitch to close customers. Perhaps controversially, Tennr has actually stripped AI from their homepage and focused on what matters: fewer denials, more patient conversions, and real revenue impact.

Takeaway: Products with built-in ROI are fundamental to success in B2B sales. Your product is only as good as your understanding of the customer. Build listening into your process, and use ROI as your north star for both product and pitch.

Moat Through Listening to the Customer

Tennr spent two years refining the approach upfront: building their own model, adapting routing logic, and managing edge cases — which for customers, seemed to be every case. This bet panned out, ultimately resulting in a better quality experience, more personalization, higher accuracy, and higher adoption. Tennr’s proprietary learning model has been trained on 100M medical documents to interpret complex and unstructured healthcare data, including handwritten notes that powers automation and personalization to the specific needs of each provider and patient interaction. Looking back, Trey noted that Tennr’s decision to build its own model was partly a reflection of the times: had GPT-3 been out or trained on a comparably large dataset, they might have taken a different path. But the custom model has become a strategic advantage — driving downstream improvements in administrative efficiency and ultimately enabling faster, more personalized care. Though this model would be difficult to replicate today, the team did not seek out a differentiated product as a moat — their moat, in fact, came from building for the needs of their customers.

When asked about moat and defensibility, Trey looks to SaaS giants like Salesforce and ServiceNow for inspiration — companies that won by becoming essential. Tennr’s long-term bet is similar: embed deeply enough into operations that switching becomes nearly impossible.

Takeaway: Sustainable advantage often comes from being deeply useful. Embedding into workflows will mean switching costs grow naturally over time. When building for healthcare, trust is critical and 99.9% accuracy is necessary.

Honing in on the Right ICP (Even if You Have to Fire Customers)

Shortly after hitting $1M in ARR, Trey made a hard call: fire 60% of Tennr’s customers, because they just weren’t aligned with Tennr’s core value prop. Serving the wrong ICP drains time, slows roadmap velocity, and muddles the story.

The decision allowed Tennr to focus. They doubled down on the customers who fit their ICP in terms of price point, use case, and configurability. As a result, they were able to land and expand ACVs dramatically, by continuing to deliver ROI over time.

As a founder, there is a balance between focus and staying informed on the market — Trey continues to talk to non-ICP potential customers to explore future revenue streams. But this is specific to founders — Trey operates his team against the focused KPIs and milestones for the near-term.

Takeaway: Don’t be afraid to hone in — a short-term dip in revenue is well worthwhile for long-term growth.

Raising a Series A (Without Losing Your Mind)

Trey’s advice to founders raising a Series A? Be honest about risks and limitations. And be clear about how you’ll fix it. Even in the earliest days (and even when things weren’t going well), Trey sent an investor update every month — which is a great way of keeping investors informed along the way so that changes to the business don’t come as a surprise.

Here’s what the investors in the room pointed to for Series A:

• Retentive ARR: Not just bookings, but renewals
• Workflow integration: Embedded products drive stickiness and adoption
• Clear ICP: Clear customer love within a more narrow group outweighs “ok” usage across a wider set of users
• Founder Expertise and Resourcefulness: Founders who are always one-step ahead. Ex: How well do you know your company’s growth and profitability levers? Who are you hiring next?

Furthermore, the difference in a successful Series A round can come down to a tight story — so upfront preparation is key.

What We’re Excited About at BBGV

At BBGV, we believe teams that move quickly, deeply understand selling to their customers, and know how to ship products that drive patient and provider adoption will win.

If you’re building in healthcare AI — focusing on using AI to provide more personalized and proactive care both for patients and providers — we want to hear from you. We’ve been thinking about healthcare on across the spectrum of:

• Dependent care: Helping providers deliver better care (provider-led)
• Collaborative care: Disease management and day-to-day care (provider and patient working together)
• Autonomous care: Patient-led, AI-optimized, always-on health management (patient-led)

If you’re excited about the opportunity to provide more personalized, data-driven and better quality care leveraging AI, reach out: hello@bbgventures.com. We’d love to learn more.

About BBGV:

BBG Ventures is a pre-seed and seed fund investing in underestimated founders building solutions for our polycultural future. Our founders are tackling the biggest arenas of the economy that require reinvention: health and wellbeing, work and education, financial security, energy resiliency and overlooked consumers. We look for founders who want to move the world forward, and whose unique combination of lived and learned experiences give them a competitive advantage.