Overview
Opkit was a New York-based healthcare technology startup, founded in 2021 and backed by Y Combinator's Summer 2021 batch, that built AI-powered phone call automation for the US healthcare industry.The company began as an insurance verification platform for telehealth companies, then pivoted in August 2024 to a broader generative AI call center for medical practices.
Despite strong technical execution, credible investor backing, and at least one beta customer who called the product existentially necessary, Opkit never raised a Series A.It was acqui-hired by AI digital workers startup 11x in late 2024 for its voice AI technology and engineering talent, with the Opkit platform subsequently wound down.
The core failure was structural: the initial insurance verification product was too narrow to build a venture-scale business independently, and the pivot to a broader AI call center—while technically validated—arrived too late and with too little runway to attract growth-stage capital on its own terms.
Image 1 / 2
Founding Story
Sherwood Callaway and Justin Ko met in early 2019 as early employees at Brex, the $12 billion fintech startup.[1] Together they built "Cash," a business bank account product—an experience that gave them direct, hands-on exposure to the engineering challenges of integrating with regulated financial counterparties and building API-driven infrastructure at scale.[1]
The two founders brought complementary backgrounds. Callaway, the son of an orthopedic surgeon, grew up around the business of healthcare and had firsthand familiarity with its administrative inefficiencies.[2] Ko brought prior entrepreneurial experience—he had previously co-founded SpeakGenius, an English learning tool—and had studied entrepreneurship at the University of Southern California.[4] The two were also best friends and former roommates, a personal chemistry that Callaway later cited as a factor in investor conviction.[19]
The founding thesis was built on an explicit analogy between fintech and healthcare. Callaway articulated it directly: "Both are regulated industries. At Opkit, what we're doing is integrating with insurance carriers to facilitate transactions related to determining the patient's eligibility and coverage, which is very similar to the kind of thing that we were [doing at Brex]."[36] In other words, the founders believed that the same developer-first, API-driven approach that had modernized payment rails in fintech could be applied to the administrative plumbing of healthcare.
Ko framed the market opportunity at launch: "Telehealth is an entirely new business model that stands to make healthcare more inclusive, accessible and affordable. But providers need updated tools that are built for a virtual care environment. That's why we built Opkit, to reduce administrative burden for this next generation of healthcare providers."[33]
Opkit was founded in summer 2021 and entered Y Combinator's S21 batch that same year, headquartered in New York.[3] Callaway later acknowledged that YC was essential for navigating the fundraising process as a first-time founder: "I'm a first-time founder and a former engineer, so the fundraising process was daunting. Y Combinator provided a lot of support and guidance. They helped connect us with investors and learn the norms of the venture capital world."[34]
The company's initial vision was focused on telehealth insurance verification, but Callaway signaled from the start that this was a wedge. His stated North Star was broader: "We want to unlock real-time patient payments for covered services."[38]
Timeline
- Early 2019 — Sherwood Callaway and Justin Ko meet as early employees at Brex, co-building "Cash," a business bank account product.[1]
- Summer 2021 — Callaway and Ko found Opkit; the company enters Y Combinator's S21 batch.[3]
- August 31, 2021 — Opkit raises a $125K seed round coinciding with its YC batch.[16]
- March 7, 2023 — Opkit launches publicly out of stealth with its insurance verification platform for telehealth companies, simultaneously announcing a $1M+ seed round led by Global Founders Capital.[5]
- March 13, 2023 — MobiHealthNews covers the launch, detailing the web dashboard, REST API, and automatic re-verification feature.[6]
- March 19, 2023 — AlleyWatch publishes an extensive founder interview; Callaway discusses healthcare investor skepticism and advises founders to keep burn low.[18]
- September 22, 2023 — CB Insights includes Opkit in its Revenue Cycle Management market map.[39]
- August 7, 2024 — Opkit announces a pivot to a generative AI call center for medical practices, with beta clients Fella and Flourish named publicly.[8]
- September 1, 2024 — PitchBook records the Opkit acquisition by 11x (deal date).[20]
- November 11, 2024 — 11x announces its $50M Series B led by Andreessen Horowitz and simultaneously discloses the Opkit acquisition.[21]
- November 11, 2024 — Callaway posts on LinkedIn confirming the Opkit platform will be wound down; both founders join 11x.[24]
What They Built
Product v1: Insurance Verification for Telehealth (2021–2024)
Opkit's first product automated one of the most tedious tasks in healthcare administration: verifying that a patient's insurance will actually cover a visit before the appointment happens. For telehealth companies, this was a persistent operational headache. A clinic scheduling dozens of virtual appointments per day had to manually call insurance carriers, navigate hold times, and confirm coverage details for each patient—a process that was slow, error-prone, and expensive in staff time.
Opkit's platform had two delivery modes, reflecting the founders' developer-first instincts from Brex.[6] The first was a web dashboard for clinic staff who wanted to run verifications manually through a browser interface. The second was a JSON REST API that allowed engineering teams to embed insurance verification directly into their own scheduling or practice management software—no manual steps required.[6]
A notable differentiating feature was automatic re-verification. For returning patients, Opkit re-checked insurance coverage before each new visit rather than relying on a one-time check at intake.[7] This addressed a real pain point: insurance status changes—patients losing coverage, switching plans, or hitting benefit limits—are common, and a stale verification can result in denied claims weeks after a visit. The platform was HIPAA-compliant and SOC 2 Type 2 certified, clearing the baseline trust requirements for healthcare enterprise sales.[12]
At launch, Opkit ranked #5 product of the day and #3 SaaS product of the week on Product Hunt.[29]
Product v2: Generative AI Call Center (2024)
In August 2024, Opkit announced a significant expansion of scope. The new product automated back-office phone calls across multiple use cases: insurance verifications, prescription checks, and medical records requests.[8] This was a meaningful shift—from a single-task SaaS tool to a general-purpose AI calling infrastructure for healthcare.
The technical architecture was built on a stack of modern AI services: OpenAI and Anthropic for language understanding, ElevenLabs for voice synthesis, Deepgram for speech recognition, and Twilio for telephony.[13] The system could navigate Interactive Voice Response (IVR) phone trees, wait on hold, and conduct conversations with insurance carrier representatives. Calls were recorded, transcribed, and made available for staff review after completion.[10]
The product's defining design choice was a human-in-the-loop model. Opkit's algorithm determined whether a given call should be routed to AI or a human agent. If an AI call failed—due to accent recognition errors, a caller refusing to engage with the AI, or call complexity exceeding the system's capability—it was automatically escalated to a human team.[9] This was not a fallback feature but a core architectural decision, reflecting the founders' honest assessment of current AI limitations.
The product also retained a developer API, allowing engineering teams to embed "self-driving" phone calls into their own applications without building the underlying infrastructure.[11] Pricing was usage-based—charged per call initiated and per minute of call duration—positioned as cheaper than offshore call center workers or full-time employees.[29]
According to YC's company profile, Opkit launched five products in total across its life.[14] Only two—the insurance verification platform and the AI call center—are identifiable from public sources.
Market Position
Target Customers
Opkit's initial target was telehealth companies and virtual medical clinics—a deliberate niche within the broader healthcare administrative software market. This focus made sense at founding: the telehealth sector had expanded rapidly during the COVID-19 pandemic, and virtual-first clinics were building their operational infrastructure from scratch, making them more receptive to API-first tools than legacy medical practices with entrenched workflows.
The pivot in August 2024 broadened the target to all medical practices conducting back-office phone calls—a much larger universe that included traditional brick-and-mortar clinics, specialty practices, and health systems. Named beta clients included Fella, a virtual weight-loss clinic, and Flourish, a virtual nutrition clinic.[27] Both were virtual-first, suggesting Opkit's customer base remained concentrated in the telehealth segment even after the pivot.
Market Size
The revenue cycle management (RCM) market—the broader category encompassing insurance verification, billing, and claims processing—is large. CB Insights included Opkit in its RCM market map in September 2023, validating the category positioning.[39]
Callaway identified the structural reason the market remained large and underserved: "For better or worse, today's healthcare system runs on manual phone calls. The industry has been slow to adopt electronic transactions, mainly because payers are not incentivized to streamline these processes. There's also HIPAA, which makes it risky for healthcare companies to share data."[30]
This structural dynamic—payers having no financial incentive to modernize their transaction infrastructure—meant the phone call remained the dominant channel for administrative healthcare communication. That persistence created a durable automation opportunity. However, it also meant that the market was fragmented across thousands of individual medical practices, each with different payer mixes, call volumes, and administrative workflows. Selling into this fragmented market at venture scale required either a very high average contract value or very high volume—both of which are difficult to achieve quickly in healthcare.
Competition
Opkit operated in a crowded but fragmented competitive landscape. In insurance verification specifically, established players included Availity (a large payer-provider data network), Change Healthcare (acquired by Optum), and a range of practice management software vendors that bundled verification as a feature. These incumbents had deep payer integrations and large existing customer bases, but their products were generally built for traditional medical practices rather than telehealth-native companies.
In the broader AI call center space that Opkit entered with its 2024 pivot, competition included general-purpose AI voice platforms (such as Bland AI and Vapi) as well as healthcare-specific automation vendors. The acquirer, 11x, was itself building AI phone representatives—though for sales and business development rather than healthcare back-office operations. The competitive dynamic in AI voice was moving fast: the same technology stack that enabled Opkit's pivot (OpenAI, ElevenLabs, Deepgram, Twilio) was available to any well-funded competitor.[13]
Opkit's differentiation rested on three factors: healthcare-specific domain knowledge (understanding payer IVR systems, HIPAA compliance requirements, and medical terminology), the human-in-the-loop fallback model, and the developer API. None of these were insurmountable moats, but they represented meaningful lead time in a specialized vertical.
Business Model
Opkit operated on a usage-based pricing model for its AI call center product, charging customers per call initiated and per minute of call duration.[29] The company positioned this pricing as cheaper than the alternatives: offshore call center workers or full-time administrative employees. This framing made the value proposition straightforward—Opkit was a cost-reduction tool with a direct labor substitution comparison.
The developer API component suggested a potential platform revenue layer, where software vendors building healthcare applications could embed Opkit's calling infrastructure and pay on a consumption basis. This would have created a B2B2B distribution channel—reaching end-user medical practices through the software they already used—but there is no public evidence this channel was developed before the acquisition.
The initial insurance verification product likely followed a similar usage-based or subscription model, though no pricing details from that era are publicly available. The company's total funding of approximately $1.12–$1.13M across its entire life[15] implies the team operated with an extremely lean cost structure throughout.
Traction
Opkit's publicly available traction data is limited but directionally informative.
At the March 2023 product launch, Opkit ranked #5 product of the day and #3 SaaS product of the week on Product Hunt.[29] This indicated meaningful developer and early-adopter awareness, though Product Hunt rankings reflect community engagement rather than commercial adoption.
For the August 2024 AI call center pivot, two named beta clients were publicly disclosed: Fella (a virtual weight-loss clinic) and Flourish (a virtual nutrition clinic).[27] Flourish's co-founder, Devin Solanki, provided a strong testimonial: "We literally could not run our business without Opkit. I've been really impressed at Opkit's reliability and accuracy."[28] This is a high-quality qualitative signal—a customer describing the product as operationally essential is meaningful—but two named beta clients at the time of a pivot announcement suggests the commercial base was still early.
No revenue figures, customer counts, or growth metrics were disclosed at any point in Opkit's public life. The absence of a Series A, combined with the acqui-hire outcome, is the strongest available proxy for commercial scale: the company did not reach the revenue or growth trajectory required to attract growth-stage capital independently.
Post-Mortem
Opkit operated for approximately three years, raised $1.12–$1.13M in total,[15] pivoted once, and was acqui-hired before reaching Series A scale. The acqui-hire structure—technology and team absorbed, product wound down—is the clearest available signal about the company's commercial state at exit. Callaway's LinkedIn post confirmed the outcome: "As we begin winding-down the Opkit platform over the next few months, we'll be working to ensure a smooth transition for everyone. To my investors: thank you for believing in us and for supporting us through every pivot."[24]
The failure had multiple contributing causes. They are ordered below by importance.
1. The Initial Product Was Too Narrow to Build a Standalone Venture-Scale Business
The insurance verification product was a real solution to a real problem, but it was a feature-sized market within a larger revenue cycle management category. Telehealth companies—Opkit's specific initial target—were a subset of a subset: virtual-first clinics within the broader medical practice universe. This niche focus made early customer conversations tractable, but it constrained the addressable revenue pool.
The product launched publicly in March 2023, approximately 18 months after founding.[5] No customer count or revenue figure was disclosed at launch or at any subsequent point. The company operated for another 17 months after launch before announcing the pivot in August 2024—a period during which no funding was raised and no growth metrics were made public. The absence of disclosed metrics during this period, combined with the eventual pivot, strongly implies the insurance verification product did not achieve the revenue trajectory required to raise a Series A.
The attempted remedy was the pivot itself: expanding from single-task insurance verification to a multi-use-case AI call center serving all medical practices. This was the right strategic direction—it expanded the addressable market significantly—but it came approximately 17 months after the public launch of the original product, leaving limited runway to prove the new thesis before capital ran out.
2. Constrained Capital Created a Narrow Window for Each Bet
Opkit raised a total of approximately $1.12–$1.13M across its entire life.[15] For context, the median seed round for a US healthcare SaaS company in 2023 was substantially larger. The $1M+ seed round closed in March 2023 was the company's last independent financing event. No Series A was raised.
Callaway explicitly identified healthcare-specific investor skepticism as a fundraising obstacle: "Many investors are afraid of backing a healthcare company because they don't understand how the industry works. This made it harder for us to raise money from angels and funds who were otherwise very excited about our team."[18] The investors who did participate—Global Founders Capital, Mischief (Plaid founder Zach Perret's fund), and former a16z partner Rex Salisbury[17]—were backing the team's Brex pedigree as much as the product. Callaway confirmed this directly: investors backed Opkit because of the co-founding team's experience as early engineers at Brex, their experience building software in regulated industries, and their personal chemistry as best friends and former roommates.[19]
The consequence of limited capital was a narrow window for each product bet. When the insurance verification product did not generate the metrics needed for a Series A, the team pivoted—but the pivot was announced in August 2024, and the acquisition deal date was recorded as September 1, 2024.[20] The gap between the pivot announcement and the acquisition was approximately one month, suggesting the two processes overlapped. The company had insufficient runway to prove the new product thesis before needing to find an exit.
Callaway's own advice to founders—"Keep your burn low. Avoid hiring employees until you have complete conviction in your idea—then scale up!"[35]—reflects the lean operating posture the team maintained throughout. This extended runway but also limited growth velocity, creating a slow-burn dynamic that is difficult to convert into venture-scale metrics.
3. The Pivot Came Too Late to Raise Independently
The August 2024 pivot to a generative AI call center was technically well-executed and commercially validated by at least one customer who described the product as operationally essential.[28] The product addressed a genuinely large and persistent market: healthcare's dependence on phone calls is structural, not transitional, because payers have no financial incentive to adopt electronic transactions.[30]
But the timing created an insurmountable problem. A Series A investor evaluating Opkit in August–September 2024 would have seen a company that had just announced a new product direction with two named beta clients and no disclosed revenue metrics. The AI voice space was simultaneously becoming more crowded, with well-funded competitors building similar capabilities. Raising a Series A on a pivot announcement—without months of growth data from the new product—is extremely difficult in any market environment.
The acqui-hire by 11x resolved this problem by providing an exit path that valued what Opkit had demonstrably built: a capable AI voice technology stack and a team with deep healthcare domain expertise. 11x acquired Opkit specifically to accelerate its AI voice capabilities for its "Jordan" AI phone representative product.[22] The acquirer valued the technology and team, not the customer base or revenue—the definitional structure of an acqui-hire.
4. AI Capability Limitations Required Ongoing Human Intervention
The human-in-the-loop model was both a product strength and a unit economics constraint. Callaway was candid about the current limitations: "Some back office phone calls are simple to automate, but others are very complex—too complex for current AI. What happens when the AI doesn't understand a speaker's accent, or when the person on the other end of the call refuses to speak to the AI and hangs up? That's where Opkit's human-in-the-loop approach comes in."[31]
Ko reinforced this: "Conversational AI is incredibly powerful, but it isn't a solution by itself. AI needs to be embedded in a useful product and supported by a team of humans that can monitor performance and step in when it fails. That's where we invested our efforts."[32]
The honest acknowledgment of AI limitations was the right product decision—it made the system reliable enough for healthcare use cases where errors have real consequences. But it also meant Opkit was operating a hybrid AI-plus-human service rather than a pure software business. Human escalation costs are variable and scale with call volume, compressing margins as the business grew. For a company with $1.13M in total funding, staffing a reliable human fallback team while simultaneously building the AI layer was a significant operational burden. This dynamic made the path to the gross margins typical of venture-scale SaaS businesses longer and harder.
Key Lessons
-
Wedge products in regulated industries need a clear path to expansion before the wedge is built. Opkit's insurance verification product was a legitimate solution to a real problem, but the telehealth-specific niche was too narrow to generate venture-scale revenue on its own. The founders had a broader vision—"real-time patient payments for covered services"[38]—but the path from the wedge to that vision required either faster expansion or a larger initial market. Founders entering regulated verticals should validate the expansion path before committing to a narrow initial product.
-
Healthcare investor skepticism is a real fundraising tax that compounds over time. Callaway's observation that many investors avoided healthcare because they didn't understand it[18] reflects a structural disadvantage for healthcare startups relative to consumer or enterprise SaaS. With a smaller pool of willing investors, healthcare founders face higher fundraising friction at every round—which means they need either stronger metrics or more patient capital than their non-healthcare peers to reach the same milestones.
-
Pivots require enough runway to generate new proof points before capital runs out. The gap between Opkit's pivot announcement (August 2024) and its acquisition deal date (September 1, 2024) was approximately one month.[20] A pivot that cannot be validated with growth data before the next fundraise becomes an acqui-hire opportunity rather than a Series A story. Founders considering a pivot should assess whether remaining runway is sufficient to generate the metrics the new thesis requires—and if not, raise before pivoting rather than after.
-
Human-in-the-loop is the right design for AI products in high-stakes domains, but it changes the unit economics. Opkit's decision to build human fallback into the core architecture was technically sound and commercially necessary for healthcare. But hybrid AI-plus-human services have fundamentally different cost structures than pure software. Founders building in domains where AI errors have real consequences should model the human cost explicitly and ensure their pricing reflects it.
-
Team pedigree can open doors but cannot substitute for product metrics at growth stages. Opkit's Brex credentials were the primary driver of investor conviction at the seed stage.[19] This is a legitimate and common dynamic at pre-revenue stages. But growth-stage investors evaluate metrics, not pedigree. The team's credibility bought time and capital to find product-market fit—it could not substitute for the revenue growth data that a Series A requires.
Sources
- Opkit About Page
- Y Combinator — Opkit Company Profile
- Accesswire — Opkit Launch Press Release, March 7, 2023
- MobiHealthNews — Opkit Launch Coverage, March 13, 2023
- CustomerThink — Opkit AI Call Center Pivot, August 7, 2024
- AccessNewsWire — Opkit AI Call Center Press Release, August 7, 2024
- AlleyWatch — Opkit Founder Interview, March 19, 2023
- Tracxn — Opkit Funding Data
- Crunchbase — Opkit Company Profile
- GlobeNewswire — 11x Series B Announcement, November 11, 2024
- Today in AI — 11x and Opkit Acquisition
- Sifted — 11x Series B Coverage, November 11, 2024
- CB Insights — Opkit Company Profile
- Fierce Healthcare — Opkit Launch Coverage, March 15, 2023
- LinkedIn — Sherwood Callaway Post, March 14, 2023
- LinkedIn — Opkit Company Page
- Y Combinator — Opkit Jobs Page (Inactive Status)