Every Company Has a Digital Profile. Insurance Just Hasn’t Been Using It.

Tanner Hackett’s entry into insurance was, by his own account, entirely accidental — he met a former Chief Digital Officer of Farmers Insurance in Vegas, who told him he should join “this revolution,” and that was his introduction, in late 2014. Before that, he’d built and run Button, a New York-based mobile marketing company, where the thing he considers most responsible for its success wasn’t the marketing technology itself but the deliberate investment in people operations — hiring infrastructure, training, communication, and collaboration systems that let the company attract and retain genuinely strong talent.

When family circumstances brought him back to Los Angeles around 2018-2019, he tried to package that same people-operations discipline into a product called “the culture stack,” aimed at small businesses that lacked the compliance and HR infrastructure larger companies take for granted. It failed to land with the entrepreneurs he pitched — the investment required felt too uncomfortable, too much of an admission that their own operations weren’t as buttoned-up as they assumed. But in having those conversations, he kept encountering real, expensive examples of what happens when small businesses don’t have that infrastructure: harassment claims, wrongful termination suits, governance failures — precisely the risk category management liability insurance exists to cover. That observation became Counterpart.

In Episode 87 of InsurTechTalk, Tanner and I covered how Counterpart’s “Digital Risk Profile” actually works, why he’s skeptical of “AI is the future” claims from other insurtech founders, and why he deliberately built on surplus lines rather than pursuing admitted filings.

About Tanner Hackett

Tanner Hackett is CEO and founder of Counterpart, an MGU focused on management liability insurance — D&O, employment practices liability (EPLI), fiduciary liability, and crime insurance — for small businesses under 250 employees. Before Counterpart, he founded and ran Button, a mobile marketing company, and an earlier e-commerce company. Counterpart is backed by Markel and Aspen as capacity partners, operates primarily on surplus lines, and is headquartered remotely with Tanner based in Venice Beach, Los Angeles.

Why Small Business Is Genuinely Underserved Here

Tanner’s framing of the target customer is specific and deliberately narrow: businesses under 250 employees, which structurally lack the compliance functions larger companies build as a matter of course — no dedicated HR team managing terminations properly, no formal training infrastructure, no legal review before a difficult personnel decision. His pointed observation: these are people who’ve dedicated their lives to their business, and very few institutions are actually looking out for them — if anything, the regulatory and litigation environment (he specifically called out California as unusually difficult) has gotten harder to navigate as a small operator over time, not easier.

The Insurance Analogy From E-Commerce

Tanner’s underwriting philosophy comes directly from his e-commerce background, and the parallel is genuinely clean: e-commerce companies predict a consumer’s probability of purchase based on behavioral attributes — browsing patterns, purchase history, time-of-day activity. Counterpart applies the same first-principles logic in reverse: instead of predicting purchase probability from consumer attributes, it predicts claim probability from business attributes.

Culture Compliance, Defined Precisely

We spent real time on what “culture compliance” actually means, since it’s easy to treat as a soft, undefinable concept. Tanner’s working definition: culture is simply a representation of a company’s values, and the problem isn’t inherently “good” or “bad” culture — it’s dissonance between a company’s stated culture and how leadership actually behaves (his example, delivered carefully: Elon Musk’s approach isn’t necessarily “bad culture,” the issue is dissonance between existing organizational culture and an abruptly different imposed one). On the compliance side specifically, the exposure is concrete and often invisible to small operators: state-by-state variation in layoff notification requirements, OSHA mandates that differ California to Massachusetts, Family Medical Leave Act application shifting between remote and in-office work — a level of regulatory complexity a small business owner focused on simply selling their product rarely has the bandwidth to track correctly.

The Digital Risk Profile: What Actually Goes Into It

This is Counterpart’s core underwriting innovation, and Tanner was specific about both what it is and — deliberately — what it isn’t.

The starting point: a typical management liability application runs 20-50 questions — location, employee count, revenue, prior litigation — genuinely validated data that actuaries have relied on for decades. But Tanner’s argument is that what arrives in that PDF represents only a sliver of the actually relevant information about a business. Every company has a broader digital footprint — much like the behavioral data an e-commerce company already tracks about individual consumers — and Counterpart’s Digital Risk Profile pulls from that broader footprint to build a claim-probability score, presented to underwriters with an explicit confidence score indicating how statistically reliable that specific assessment is, rather than presenting every output with false uniform certainty.

The general categories feeding the model (Tanner was candid he couldn’t disclose the specific formula): signals related to company culture (what employees are saying publicly), prior litigation history, compliance and governance practices (how terminations and layoffs are actually handled), and broader business operations health — his sharpest specific insight being that financial distress correlates directly with elevated claim probability, because companies under real financial pressure tend to make more rushed decisions and cut corners in ways that produce exactly the kind of litigation management liability coverage exists to address.

Why He’s Skeptical of “AI Is the Future” as a Pitch

This was the sharpest and most technically grounded critique in the conversation, and it’s worth taking seriously as a framework for evaluating any insurtech’s AI claims.

Tanner’s explicit disagreement with founders who present AI as the core differentiator: in a complex risk category like management liability, genuinely good outcomes require pairing data with actual underwriting expertise, not replacing it. Counterpart employs underwriters it calls “risk engineers” specifically to evaluate what the Digital Risk Profile surfaces alongside the application data — deciding both whether to write a given risk and how the digital signal should adjust coverage and pricing, rather than letting a model’s output stand alone as a black-box decision.

His more general critique of the “AI” label across insurtech broadly: most of what actually matters happens before the AI/ML step. His breakdown of the real sequence — determining what data is actually relevant, figuring out how to acquire it, structuring it usably, and only then applying machine learning to it. His pointed jab at competitors: plenty of companies run a linear regression and call it AI, and plenty of traditional insurers are still training models on data that’s two, five, or ten years old — largely irrelevant in a post-COVID operating environment that fundamentally reshaped how small businesses function.

Counterpart’s actual technical bench, per Tanner: three actuaries working alongside a team including a PhD in machine learning, with continuous back-testing of model assumptions specifically to catch when a single data source is disproportionately skewing results — a genuine risk with small businesses specifically, since a small operation (his example: a small restaurant or an independent contractor) often has a genuinely thin digital footprint, making any single overweighted signal dangerous if it happens to be missing or misleading for a given business.

Explainability as a First-Order Design Requirement

Tanner explicitly named explainability as something too many insurance companies neglect when adopting AI-driven underwriting — the discipline of being able to trace why a model produced a given output, not just accept the output itself. His framing directly echoes a broader theme worth flagging across the insurtech AI conversation generally: an underwriting decision that can’t be explained and audited after the fact is a genuine governance and regulatory liability, not just a technical nicety.

Why Surplus Lines, Not Admitted

I asked Tanner directly why Counterpart built on surplus lines with Markel and Aspen as capacity rather than pursuing admitted filings from the start. His answer centered entirely on speed and iteration velocity: admitted filings can take 6-18 months of regulatory process per state, while Counterpart’s entire underwriting thesis depends on accumulating real claims data quickly to refine its models — every additional rep (his phrase) improves statistical confidence. Markel and Aspen were specifically attractive partners because they already understood management liability deeply and had existing surplus paper ready to support the product, letting Counterpart get to market and start generating real underwriting data far faster than an admitted-filing path would have allowed. He noted admitted filings remain a longer-term consideration, but weren’t the right starting point given how much the model depends on fast iteration.

MGU, Not MGA

Tanner drew the distinction precisely: an MGA is primarily focused on distribution. Counterpart positions itself as an MGU — its platform is built around underwriting more thoughtfully, providing genuine analytical value to carrier and broker partners, not just moving policies through a channel. Depending on the specific program, Counterpart writes on a quota-share basis alongside Markel and Aspen or uses their paper to write up to 100%, with the structure varying based on each carrier’s specific appetite for what they want retained on their own balance sheet.

Claims as the Next Data Frontier

Claims handling currently sits with Aspen and Markel directly, but Tanner flagged it as a genuine near-term opportunity: Counterpart has built data infrastructure specifically aimed at reaching claims resolutions faster, benefiting all stakeholders — his framing being that a defendant facing litigation benefits directly from faster resolution, independent of the insurer’s own loss ratio interest, making this a genuinely aligned-incentive opportunity rather than a purely cost-driven one.

Advice: Find a Monastery

Asked for closing advice, Tanner pointed to something genuinely personal rather than tactical: Deer Park Monastery, a Buddhist monastery near Carlsbad (between Orange County and San Diego), founded by the late Thich Nhat Hanh, with sister locations in New York, France, and Thailand. He visits for anywhere from a single Sunday to a full week, describing the effect afterward as making “everything look a little more vivid” — a specific, deliberate counterbalance to how distorting a constant news cycle can be, and a discipline he considers directly relevant to sustaining founder resilience over the long run, not a separate hobby disconnected from the work.

Key Takeaways

• Small businesses under 250 employees are structurally underserved in management liability coverage because they lack the compliance infrastructure larger companies take for granted, and the regulatory burden they face keeps growing regardless
• Counterpart’s Digital Risk Profile applies e-commerce-style behavioral attribute modeling to businesses instead of consumers, predicting claim probability rather than purchase probability from a broader footprint than a standard 20-50 question application captures
• Effective AI-driven underwriting in complex risk categories requires pairing data models with genuine human underwriting expertise, not replacing judgment with a black-box output — a direct rebuttal to “AI is the future” positioning common elsewhere in insurtech
• Most of what determines whether an “AI” claim is real happens in data engineering — sourcing, structuring, and validating relevant data — well before any machine learning model is applied; skipping that step is where most buzzword-driven claims fall apart
• Explainability (the ability to trace why a model produced a specific output) is a first-order underwriting requirement, not an optional governance nicety, especially in a regulated, high-stakes coverage category
• Choosing surplus lines over admitted filings was a deliberate speed-to-data tradeoff — Counterpart’s entire model depends on rapid claims data accumulation, which a 6-18-month admitted filing process per state would have significantly slowed
• MGU (underwriting-focused) and MGA (distribution-focused) are meaningfully different positions in the value chain, and Counterpart’s differentiation depends specifically on the underwriting value it brings to carrier partners, not just moving policy volume