1、The first thing to clarify is what Harper actually is. It is not an “AI insurance company” in the conventional sense of an insurance carrier. It is an AI-native commercial insurance brokerage and agency.

Harper does not primarily put insurance risk on its own balance sheet the way a carrier such as Chubb, Travelers, or The Hartford does. Its role is to collect information about a business, prepare insurance applications, identify relevant carriers and underwriters, obtain and coordinate quotes, place coverage, and continue servicing policies, renewals, and claims-related communication.

Harper's own privacy policy explicitly describes the company as an “AI-native commercial insurance brokerage,” as does Y Combinator. Its innovation is therefore not simply an AI-powered insurance comparison website. It is attempting to redesign a large percentage of the labor performed inside a commercial brokerage as AI-driven workflows.

The contrast is fundamental:

A conventional Insurtech strategy is to build software and sell it to insurers or brokers.

Harper's strategy is to use the software itself and become the broker.

Emergence Capital describes this broader model as AI-Native Services: instead of selling SaaS to a professional-services incumbent, the startup directly delivers the service and captures the industry's service revenue while using AI to improve labor productivity and margins.

That distinction explains most of Harper's strategy.

2、Dakotah Rice's childhood provides much of the explanation for why Harper ultimately focused on commercial insurance rather than becoming another generic AI software company.

Rice says he grew up in rural Alabama. His father operated a nightclub, his uncle ran a trucking business, and Antler's account of his background says his mother worked in home-health services. As a child, Rice helped with the family's businesses, including cleaning ashtrays and assisting with inventory.

Rice has subsequently disclosed that his father died while Rice was young and that his mother largely raised him alone, sometimes working two jobs. He has described a period immediately before ninth grade when the family had no electricity, the refrigerator was empty, and he went to bed hungry.

This is far removed from the stereotypical Silicon Valley path of affluent household, elite technical training, Stanford, and startup.

Two themes repeatedly appear in Rice's autobiographical account:

scarcity and upward socioeconomic mobility;

and

small-business owners being constrained by dysfunctional financial infrastructure.

He frequently tells the story of his uncle's trucking business. In Rice's account, the company was not simply destroyed by poor operations; financing and insurance costs overwhelmed it. Commercial insurance therefore entered his worldview not as an abstract financial product but as infrastructure that could determine whether a family business survived.

Rice has also said that he spent summers around his family's insurance agency and once swore that he would never work in insurance. He ultimately founded an insurance brokerage and named it Harper, his mother's maiden name.

The brand therefore carries a clear family connection rather than being a purely invented startup name.

3、Another formative part of Rice's background was learning to read social environments and manage his own narrative as a Black gay teenager in the American Deep South.

Rice has publicly written that being Black and gay in that environment taught him early to recognize which behaviors would be rewarded, what he needed to soften, what he needed to emphasize, and what might expose him to punishment. He later connected these experiences to his ability to read a room quickly and control the narrative around himself.

That background helps explain his professional role.

Rice is not a technical founder. His recurring strengths are closer to:

sales, storytelling, fundraising, market judgment, relationship building, finance, and corporate strategy.

Those strengths are highly complementary to Tushar Nair's technical profile.

4、Rice's education represents a progression from entering elite institutions to deliberately leaving one.

He attended Brown University, studied political science, and graduated in 2016. Brown records also identify Rice as Finance Chair of the university's Black Student Union during his undergraduate years, indicating that his engagement with Black student organizations was already significant.

His political-science education is notable.

Rice did not enter business as a computer scientist or quantitative-finance specialist. Over time, he instead accumulated knowledge of institutions, organizations, power relationships, capital markets, business narratives, and networks.

Immediately after Brown, he entered Wall Street.

He worked in investment banking at Goldman Sachs around 2016–2017, moved to The Carlyle Group around 2017–2018, and then worked at Coatue Management from roughly 2018 to 2021, investing across technology, internet, media, and telecommunications.

Each institution supplied a different component of his later founder skill set:

Goldman Sachs: transaction execution and capital markets.

Carlyle: large-scale private-equity investing.

Coatue: technology-company analysis, growth investing, and public-market frameworks.

The Coatue period was particularly relevant because it placed Rice at the intersection of how technology businesses operate and how professional investors value them.

5、Harvard Business School became an important turning point, but the important fact is not simply that Rice attended HBS—it is that he chose to leave.

Rice entered Harvard Business School's MBA program around 2021 but did not follow the conventional path to completion. Brown still described him as an HBS MBA candidate in 2022; Rice has subsequently said publicly that he left Harvard, while YC describes him as an HBS dropout.

The identity shift was significant:

professional investor → founder/operator.

Rice later described a growing disconnect between classroom cases and the rapidly changing world outside the classroom. What increasingly attracted him was building a real company rather than continuing to study other companies.

It was the first major occasion on which he abandoned a highly secure elite-career path.

6、Tushar Nair represents the other half of Harper: a long engineering background, experience automating financial workflows, and the capability to turn Rice's business thesis into operating systems.

Public professional records show that Nair studied Computer Engineering at the Illinois Institute of Technology. He then spent a substantial portion of his career in Goldman Sachs' technology organization, holding roles including Technology Analyst and Associate. Public career records also place him at Harry's as a Staff Software Engineer before a later Goldman technology-management role.

Harper- and YC-related materials summarize the most strategically relevant part of that experience: Nair worked on machine-learning, AI, and automation systems associated with Goldman Sachs' commodities operations.

That background maps naturally onto commercial insurance.

Insurance brokerage, investment-banking operations, and institutional trading all contain large amounts of:

semi-structured information + human judgment + email + documents + follow-up + exception handling.

Nair's specialty is turning portions of those workflows into software and machine-learning systems.

Rice and Nair knew each other from their Goldman years. Their relationship deepened through Poolit and its eventual failure, producing a clear division of labor at Harper:

Dakotah Rice: CEO—market, sales, fundraising, strategy, insurance operations.

Tushar Nair: CTO—AI, engineering, automation, product, and internal systems.

The important point is that this is not merely a commercially oriented founder paired with a technical founder for a new idea. They had already operated a startup together and experienced failure together before founding Harper.

English Translation: From Poolit's Failure to the Formation of Harper

7、Rice's first major startup was Poolit, and Harper is best understood partly as his second attempt after Poolit's failure.

Rice began developing Poolit while attending HBS.

The company aimed to reduce barriers to high-quality private-equity and venture-capital funds, allowing qualified individual investors to gain access to alternative investments historically dominated by institutional capital. In 2022, Poolit raised a $5.3 million seed round supported by Harlem Capital and Picus Capital, with other investors including Declaration Capital associated with Carlyle co-founder David Rubenstein, Coatue co-founder Thomas Laffont, and Gilgamesh Ventures.

The financing reveals something important about Rice's first entrepreneurial resource network:

his Wall Street network converted directly into startup capital.

He did not need to construct investor relationships entirely from scratch; professional credibility built at Goldman, Carlyle, and Coatue became fundraising capital.

Forbes subsequently recognized Rice in its Under 30 ecosystem and reported Poolit's approximately $5.3 million financing.

8、Poolit was also the project that transformed Rice and Nair from friends and former colleagues into long-term startup partners.

Nair joined Poolit's technical organization and became one of its key engineering leaders. Harper's launch materials describe Poolit as a fintech company the pair previously built together, while Nair's professional record lists him as Poolit's Director of Engineering.

Poolit reportedly reached roughly 10,000 users. Some Harper/YC-related biographies describe it as having reached roughly $100 million in AUM, while other founder descriptions have used capital-commitment terminology. The precise accounting characterization of that $100 million figure therefore varies across public sources.

The more consequential fact was that:

Poolit failed.

The company closed in 2023. Rice later acknowledged that he never solved the profitability problem and that his reluctance to accept failure kept the company alive longer than it should have been.

That failure matters enormously.

For a founder whose earlier résumé had largely consisted of Brown, Goldman, Carlyle, Coatue, and Harvard, Poolit was the first major entrepreneurial reset.

Nair did not simply disappear when the company struggled.

The experience of operating through Poolit's decline arguably gave the two founders a much stronger basis for their next partnership than founders who meet only because they happen to share enthusiasm for a new startup idea.

9、Between Poolit and Harper sits another often-overlooked project: CaliberX.

Public company records identify Dakotah Rice and Tushar Nair as founders of CaliberX / Caliber Exchange.

Rather than serving retail investors directly, CaliberX was positioned as B2B trade-execution and order-management infrastructure for private markets. Its APIs were designed to connect fund administrators, transfer agents, custodians, law firms, banks, wealth managers, fintech platforms, and fund managers involved in alternative-investment distribution and servicing.

Conceptually, it extended Poolit's experience:

Poolit addressed front-end distribution of alternatives;

CaliberX moved toward back-end infrastructure for alternatives;

Harper left alternatives altogether but preserved the same underlying idea:

Identify a financial-services industry still dominated by people, emails, spreadsheets, documents, and institutional coordination, then systematically automate the workflow.

The true continuity in the Rice–Nair entrepreneurial trajectory is therefore not “investing to insurance.”

It is:

financial services → operationally intensive workflows → software automation → technology-enabled service delivery.

10、Harper was formed in 2024, but its original model was not to become an insurance broker. The founders first tried to sell AI to existing insurance organizations—and that strategy failed.

Rice and Nair participated in Antler's startup ecosystem in 2024 while exploring their next company. Untapped Ventures recalls first backing them when they were only two founders working from an Atlanta apartment with no complete product and no team. Harper later entered Y Combinator's Winter 2025 batch.

The founders initially pursued what appeared to be the obvious strategy:

build AI software → sell it to traditional insurance brokerages.

The response was weak.

Rice later wrote that insurers and brokers were unwilling to entrust important workflows to imperfect AI. Prospective customers did not fully trust the output, struggled to see direct bottom-line impact, and preferred small pilots over restructuring real production workflows.

That failure produced the most consequential strategic decision in Harper's history:

If incumbent brokerages would not buy the AI, Rice and Nair would become the brokerage themselves.

Emergence Capital later highlighted precisely this transition in its investment thesis: Harper stopped trying to sell tools to incumbents and instead owned the customer relationship, performed the work, and captured the service revenue directly.

This was arguably the defining pivot of the company.

11、Consequently, Harper's product is not primarily an independent SaaS product. It is an internal AI operating system for Harper's own insurance brokerage.

YC identifies Harper Hub as a flagship internal tool.

It can assist with:

automatically completing insurance forms;

identifying relevant carriers and underwriters;

managing customer communications;

and automating tasks previously handled manually by broker assistants, producers, and account managers.

Harper has expanded AI use into:

reading insurance applications;

extracting policy and endorsement information;

submission routing;

underwriter follow-up;

quote coordination;

task prioritization;

and staff assistance during calls and servicing.

An important qualification is necessary. Harper's marketing language sometimes emphasizes autonomy and very high levels of AI-led operations, but its privacy policy explicitly states that:

AI does not independently make final insurance-eligibility decisions or bind coverage without review by licensed professionals and carriers.

Final decisions remain subject to human and carrier approval.

The accurate interpretation is therefore:

Harper is pursuing highly automated brokerage operations, not eliminating the legal and professional roles of licensed insurance professionals and underwriters.

12、Harper's first visible product-market fit appeared between the fourth quarter of 2024 and early 2025.

The company says it acquired its first customers in October 2024.

Within several months, around its YC W25 launch, Harper reported more than $6 million in annualized premium across 35 states. Rice subsequently described more than 400 clients and approximately $7 million in annualized premium.

Growth then accelerated.

In a later public post, Rice reported that Harper was approaching:

approximately 3,000 customers;

$40 million in annualized premium;

nearly $5 million in ARR;

roughly 30% monthly growth;

and approximately $8 million in funding at that stage.

Those numbers require an important distinction:

$40 million in annualized premium does not mean Harper generated $40 million in revenue.

Premium is the amount flowing through the insurance transaction in exchange for carriers assuming risk.

Harper's own economic revenue primarily comes from insurance distribution commissions and related brokerage economics.

The simultaneous disclosure of roughly $40 million of annualized premium and nearly $5 million of ARR therefore provides a much more useful picture of the business.

13、By early 2026, Harper had moved from a YC experiment into a more recognizable scale-up phase.

In February 2026, Harper announced approximately $47 million in combined Seed and Series A financing; TechCrunch reported the more precise total as roughly $46.8 million.

The Series A was led by Emergence Capital, while Harper's official announcement also named:

Y Combinator;

Peak XV;

Antler;

10X Founders;

Fellows Fund;

and Outset Capital.

Untapped Ventures says it invested when Harper was still only the two founders and continued investing through later seed and Series A stages.

Harper's capital network can therefore be viewed in three layers:

Startup infrastructure: Antler and Y Combinator.

AI/enterprise-software scale capital: Emergence Capital, Peak XV, and related institutions.

Earlier founder- and network-oriented capital: Untapped, Outset, 10X, and others.

This represents a material evolution from Poolit, whose capital formation was more heavily tied to Rice's Wall Street network.

His financing network has expanded from elite finance into the broader Silicon Valley AI and venture ecosystem.

English Translation: Business Model, Assets, Successes, Risks, and Current Position

14、Harper's core business model is not software subscriptions. It directly participates in the commercial-insurance brokerage commission pool.

The company's Harper Partners program exposes the economics unusually clearly.

Licensed insurance agents can refer commercial clients they cannot place to Harper.

Harper then handles:

quoting;

placement;

binding;

servicing;

renewals;

and claims-related servicing.

When coverage is successfully bound, the referring agent can receive 50% of Harper's commission. Crucially, Harper's terms state that the referred client account and corresponding book of business belong to Harper.

That is essential to understanding Harper's long-term value.

A SaaS company's core asset is recurring subscription contracts.

An insurance broker's core assets include:

customer relationships + the book of business + renewal commissions.

Because businesses generally need to renew insurance repeatedly, acquiring one client can potentially create a multiyear revenue stream.

Harper's economic objective is therefore not merely to use AI to process a transaction cheaply. It is to use AI to reduce acquisition, quoting, and servicing costs while accumulating a recurring book of commercial-insurance relationships.

15、Harper's real company-level assets can be understood in five layers.

The first is the book of business.

Harper's partner terms explicitly state that referred commercial accounts belong to Harper rather than the referring agent.

The second is its carrier and underwriter network.

Harper said in its 2026 financing announcement that it already worked with 165+ underwriters and intended to use new capital to expand direct appointments, programs, and capacity for complex commercial risks.

The third is proprietary AI and workflow infrastructure.

That includes Harper Hub, document extraction, submission routing, carrier matching, quote coordination, and communications automation.

The fourth is proprietary operational data.

Harper's privacy policy says its proprietary and fine-tuned models may be trained on data generated through Harper's own operations, including:

customer emails, texts, and call transcripts;

insurance applications;

quotes;

policies;

endorsements;

carrier appetite guides;

and internally labeled operational data.

That creates the possibility of a data flywheel:

more customers → more submissions → more carrier responses → better understanding of carrier appetite → better matching → higher conversion → more customers.

The fifth layer is brand and founder influence.

Rice in particular has increasingly become a visible case study for AI-native services, second-time founders recovering from failure, and Black-founder venture financing.

These should not be confused with personal financial assets. Reliable public information does not establish the founders' personal net worth or precise ownership percentages.

16、Harper's efficiency thesis is fundamentally about increasing the number of customers each insurance professional can handle—not simply reducing insurance prices.

The problem in complex commercial insurance is often not that no carrier exists.

The challenge is the fragmentation of information and coordination.

A broker may need to:

understand the customer's operations;

collect loss runs;

read financial records;

determine required coverages;

submit the account to multiple carriers, MGAs, or wholesalers;

follow up repeatedly with underwriters;

compare exclusions, limits, deductibles, and premiums;

and communicate the tradeoffs back to the customer.

Those are precisely the operational tasks Harper is attempting to automate.

The company says processes that traditionally require several days or weeks can sometimes be compressed to approximately 24–48 hours.

Untapped Ventures additionally claims that Harper is producing roughly 30 times more revenue per sales representative than conventional brokerages. That figure comes from an investor and should be treated as company/investor-reported operating data rather than an independently audited metric.

Harper's bet is therefore not:

“AI will make insurance disappear.”

It is:

“AI will allow a comparable number of insurance professionals to manage dramatically more business.”

17、A second major moat may come from distribution rather than AI itself.

AI models are becoming increasingly available.

What is harder to replicate is:

customers;

carrier appointments;

underwriter relationships;

insurance licenses;

historical submission data;

renewal relationships;

and expertise in complex industries.

That makes Harper Partners strategically important.

The program effectively converts commercial-insurance customers that smaller personal-lines agencies cannot or do not want to handle into an external acquisition channel for Harper.

The referring agent does not need to service the customer after handoff. Harper takes responsibility from quotation through ongoing servicing and shares the resulting commission.

At scale, that could make Harper more than a direct-to-business digital brokerage. It could become commercial-insurance infrastructure behind other agencies.

Such a distribution network could be considerably harder to copy than merely having a marginally better language model.

18、Harper's investors appear to be underwriting the broader “AI-native services” thesis rather than only an Insurtech thesis.

Emergence Capital's logic is explicit.

Historically, technology startups sold SaaS to accountants, lawyers, brokers, consultants, and other professional-services firms.

Large language models and agents may now allow startups to become the accounting firm, law firm, brokerage, or consulting company themselves.

Harper is a clean example:

It does not primarily sell insurance software.

It becomes the insurance broker.

AI performs an increasing share of service delivery.

The company captures brokerage commissions rather than only software subscriptions.

In theory, the accessible revenue pool is much larger than that of a narrow software vendor.

But the company is also operationally heavier than pure SaaS.

It needs:

licenses;

insurance expertise;

customer support;

carrier relationships;

renewal operations;

claims support;

and regulatory compliance.

Harper is therefore better understood as a financial professional-services company seeking software-like productivity than as a conventional lightweight SaaS company.

19、Harper's most impressive result so far should not be described as “disrupting the insurance industry.” A more defensible claim is that it has rapidly validated a new organizational model for an insurance brokerage.

As of the company's February 2026 disclosures:

Harper had served more than 5,000 businesses in roughly 13 months of active operations;

customers included manufacturers, healthcare providers, hospitality companies, transportation fleets, construction companies, bars, and restaurants;

the company worked with more than 165 underwriters;

and it had announced approximately $47 million in combined Seed and Series A financing.

A current YC job posting says Harper is adding approximately 1,000 customers per month and has grown roughly 100-fold over the prior year. Those are company-reported recruiting claims rather than audited performance statistics.

Harper and its investors also describe the financing as the largest publicly disclosed Series A raised by a Black founder. The “publicly disclosed” qualification is essential; it should not be converted into an unqualified historical record claim.

The strategic achievement is narrower but still significant:

Harper has demonstrated that even a relationship-heavy, document-heavy, judgment-heavy industry like commercial insurance can shift a substantial amount of brokerage work to AI while remaining organized as a licensed brokerage.

20、Rice's most important career decisions form a remarkably clear sequence of identity changes.

Brown → Goldman:
Entering elite finance and completing a major socioeconomic transition.

Goldman → Carlyle → Coatue:
Moving from transaction execution to capital allocation and then technology investing, building a framework for evaluating businesses.

Coatue → HBS → dropout:
Giving up the lower-risk investor/MBA trajectory to become an operator.

Founding Poolit:
His first attempt to convert Wall Street knowledge and networks into a product, company, and financing platform.

Closing Poolit:
His first direct experience that prestigious credentials and strong investors do not guarantee a viable business model.

Continuing with Nair:
Transforming a relationship tested by failure into a core organizational asset for the next company.

Abandoning the strategy of selling AI to insurers:
Harper's most important product and go-to-market pivot.

Becoming the brokerage itself:
Moving from the SaaS revenue pool into the much larger brokerage commission pool.

That final decision is what most clearly separates Harper from many AI startups.

21、Tushar Nair's role should likewise not be reduced to “the CTO who writes the software.” He is closer to the architect of Harper's operating model.

Public discussions of Nair's agent architecture suggest a preference for first making AI reliable at narrowly defined actions, then connecting specialized agents, and using human feedback to build systems that can identify and correct failures rather than immediately pursuing complete autonomy.

That approach maps closely onto Harper's actual operating structure:

first automate extraction;

then forms;

then routing;

then follow-ups;

then decision support;

then progressively increase the percentage of the workflow handled by AI.

This is considerably more realistic than the promotional image of a single super-agent independently selling commercial insurance.

Harper's own privacy policy explicitly preserves review by licensed professionals and carriers for final decisions, reinforcing the interpretation that Nair is building a human-in-the-loop automation architecture rather than simply removing humans.

22、The clearest documented negative event in the founders' history is Poolit's failure rather than a legal scandal surrounding Harper.

Poolit closed in 2023, and Rice acknowledged that he never established a sustainable path to profitability.

The lesson is significant:

prestigious résumés;

strong venture investors;

famous backers;

user growth;

and an attractive market narrative

cannot substitute for unit economics and a sustainable business model.

Rice has even acknowledged that ego delayed his acceptance of failure.

From that perspective, Harper's unusually strong emphasis on revenue, ARR, paying customers, and operational productivity can plausibly be read as a reaction to what happened at Poolit.

Rice's more recent public commentary repeatedly emphasizes actual revenue rather than pitch decks or valuation.

23、The second important failure was Harper's original B2B AI-software go-to-market model.

That experience exposed one of the most common weaknesses of enterprise AI:

an AI demo working does not mean a company will hand over a production workflow.

The issue becomes even more acute in a regulated industry such as insurance.

Rice described several obstacles:

imperfect outputs;

lack of trust;

unclear immediate ROI;

customers wanting only pilots;

and reluctance to turn over core workflows.

Harper's strategic response was unusually direct:

stop persuading incumbent brokers to change.

Create a new brokerage without legacy workflows.

That eliminates one of the central problems facing enterprise-software vendors: organizational inertia inside the customer.

24、Harper's more important current risk is whether hypergrowth can remain compatible with insurance regulation and service quality.

This is not a software API that can scale indefinitely without friction.

Commercial brokerage involves:

state licensing requirements;

customer and business data;

regulatory recordkeeping;

carrier relationships;

errors-and-omissions exposure;

complex coverage judgments;

claims;

renewals;

and human accountability.

Harper's privacy policy notes that certain insurance records generally need to be retained for five to seven years or longer and explicitly states that AI cannot bypass licensed professionals and carriers for final decisions.

As Harper grows from thousands of customers toward tens of thousands or more, the difficult questions may move beyond whether an AI model can read a PDF:

Can quality control scale with volume?

How are quoting mistakes and coverage gaps controlled?

Will carriers continue providing capacity?

Can renewal and claims experiences remain strong?

Who ultimately bears accountability in increasingly automated workflows?

Those questions will determine whether Harper can evolve from an impressive startup into a major national brokerage.

25、Harper's data flywheel is also a data-governance obligation.

The company says its internal models may use customer communications, insurance applications, call transcripts, policies, quotes, endorsements, and internally generated operational information for training. It also says that when using external AI vendors, it employs enterprise channels under which customer data is not used to train the vendors' foundation models.

This data is valuable because real insurance workflow data is far more specialized than generic public-web text.

But Harper therefore handles substantially more sensitive business information than an ordinary SaaS company storing names and email addresses.

Its systems may process:

business revenue;

locations;

vehicle and driver information;

loss runs;

claims reports;

financial information;

contracts;

insurance applications;

and detailed risk exposures.

Information security, model leakage, inappropriate data use, and AI bias therefore become increasingly important as the company scales.

These are best understood today as structural risks rather than evidence of a Harper-specific scandal.

26、On competition, Harper is not merely competing with other AI startups. It is competing with the existing commercial-insurance brokerage system.

Rice has told TechCrunch that he effectively views most traditional brokerages as competitors. The same report identifies AI-native brokerage Gyde and insurance AI-tool companies such as FurtherAI and Vantel as other emerging participants.

The larger challenge, however, may come from incumbent brokers themselves.

Large brokerages already possess:

enormous customer books;

longstanding carrier appointments;

deep industry expertise;

claims organizations;

multistate and global licensing infrastructure;

and high renewal retention.

Harper's advantage is the absence of legacy systems.

Incumbents' advantage is that they already own distribution and trust.

The competitive outcome may therefore depend less on who has the “best LLM” and more on:

whether Harper can accumulate enough customers, proprietary data, and carrier relationships before established brokers successfully modernize themselves with AI.

27、As of 2026, Rice, Nair, and Harper occupy a distinct position: they have moved beyond the concept-stage AI startup, but they remain far from the scale of the insurance industry's major institutions.

Harper now has:

real paying customers;

brokerage commission revenue;

meaningful premium volume;

a carrier/underwriter network;

a proprietary AI stack;

Y Combinator backing;

institutional venture capital led by Emergence and others;

an increasingly explicit distribution flywheel;

and a rapidly expanding book of commercial business.

But it is still a company founded in 2024.

Calling it an industry disruptor in the accomplished past tense would therefore be premature.

A more accurate interpretation is:

Harper is one of the clearest current examples of the thesis that AI startups may stop selling software to professional-services companies and instead become the professional-services company themselves.

Rice's primary contribution is not AI research.

Nair's primary contribution is not insurance-brand influence.

Their value is in the combination:

Rice converts his upbringing, financial networks, sales ability, and market narrative into demand and capital;

Nair converts complex financial workflows into scalable AI/software architecture;

Harper then locks those capabilities into a brokerage that accumulates customer relationships, renewal economics, proprietary insurance data, and carrier relationships.

That combination is the central reason Harper is worth watching.

28、The entire story becomes especially clear when compressed into a timeline.

2012–2016: Brown University.
Rice studies Political Science, participates in the Black Student Union, and makes the first major transition from rural Alabama into an elite U.S. university environment.

2016–2017: Goldman Sachs.
Rice enters investment banking, while the professional networks of Rice and Nair begin to overlap.

2017–2018: Carlyle.
Rice moves into private equity.

2018–2021: Coatue.
Rice develops his technology-investing experience.

2021: Harvard Business School / formation of the Poolit idea.
Rice begins moving from investor to founder.

2022: Poolit expands.
The company raises approximately $5.3 million; Rice leaves HBS to work on it full time, while Nair becomes a central technical leader.

2023: Poolit closes; CaliberX period.
Rather than ending the partnership, Rice and Nair continue exploring infrastructure for alternative investments.

2024: Antler → Harper.
The pair first attempt to sell AI into insurance, then pivot after weak adoption and decide to become the brokerage themselves.

October 2024: Harper acquires its first insurance customers.

2025: Y Combinator W25.
At public launch, Harper reports operations across roughly 35 states and approximately $6–7 million in annualized premium.

Later 2025:
Rice reports approximately 3,000 customers, $40 million in annualized premium, and nearly $5 million ARR.

February 2026:
Harper announces approximately $47 million of combined Seed and Series A financing, more than 5,000 business customers, and relationships with 165+ underwriters.

Current 2026 stage:
The company continues aggressive hiring, and a YC recruiting page says it is adding approximately 1,000 customers per month. Harper's central challenge has shifted from proving that an AI-native brokerage can work at all to proving that the model can become a genuinely large, compliant, durable national insurance brokerage.