How to Build an App Like Lemonade: AI Claims Processing, Insurance Architecture, and What Insurtech Requires

Lemonade did something unusual: it became a licensed insurance carrier.

Most digital insurance products do not do this. They sell policies issued by someone else, collect a fee, and call themselves InsurTech. Lemonade went further. They applied for carrier licenses in 43 US states, built their own underwriting models, and put their own capital at risk.

You cannot replicate this path quickly. Carrier licensing takes 18 to 36 months and requires $5M or more in capital per state. The good news: you do not need to. The MGA (Managing General Agent) path gets you to market in 18 to 22 weeks at a fraction of the cost.

Running costs after launch: $5K to $15K per month for actuarial data feeds, claims processing APIs, compliance tooling, and fraud screening. If you are still selecting a fronting carrier when development starts, add 4 to 8 weeks.

How Lemonade makes money, and what your options are

Lemonade's business model is premium income minus claims and reinsurance costs, plus "giveback" marketing. They charge policyholders a flat fee (roughly 25% of premium) and use the rest to pay claims. Reinsurance covers catastrophic losses. Any leftover is donated to charities selected by policyholders, which drives retention and PR.

When you build your own platform, three monetization models are available to you.

MGA commission model. Your fronting carrier writes the policies. You earn a percentage of the gross written premium, typically 15% to 25%. The carrier holds the regulatory capital and absorbs the risk. You keep the customer relationship and the technology. This is the fastest model to launch and the most common among InsurTech startups.

Subscription model for embedded insurance. Some products are sold at point of purchase (device coverage at checkout, travel insurance at booking). You charge a flat monthly or per-transaction fee to the platform distributing you. This works well when volume is predictable and you do not want per-policy revenue complexity.

Carrier model. You hold the risk yourself. You keep the full premium minus claims and reinsurance. The margin is higher if your underwriting is accurate. The capital requirement is enormous. Every dollar of premium you write requires a percentage in reserve. Lemonade raised over $400M before they were writing meaningful premium volume at scale.

Unit economics reality check: at 5,000 active renters policies at $200/year average premium, an MGA earning 20% commission generates $200K in annual revenue. At 50,000 policies, that is $2M. Those are realistic numbers for a focused vertical in year two or three, not year one.

Who builds this instead of buying Lemonade

Three specific buyer types drive most insurance platform builds.

Vertical-specific InsurTech founders. Lemonade covers renters, homeowners, pet, and car insurance for consumers. They do not serve short-term rental hosts, rideshare drivers, independent contractors with liability exposure, or gig workers who need income protection. Founders with underwriting expertise in one of these gaps build their own product because Lemonade's rate structure and coverage design does not fit their vertical.

Property management companies and platforms. A company managing 10,000 rental units has leverage over Lemonade that a consumer does not. They can build or white-label a renters insurance product, offer it at lease signing, and earn the MGA commission themselves instead of paying Lemonade to distribute it. The distribution is already built into their existing customer relationship. The software investment pays back in two to three years.

Legacy insurers building a digital product line. Traditional carriers have the licenses, the rate filings, and the claims infrastructure. What they lack is a mobile-first quote and bind experience that competes with Lemonade on speed and simplicity. They hire teams to build the digital layer separately, connecting to their existing policy administration system via API.

Embedded insurance platforms. Travel booking platforms, device retailers, and subscription companies want to sell insurance at the moment of purchase. They need a quote engine that produces a premium in under two seconds and a bind flow that completes in one tap. Lemonade's consumer app is not designed for this. A custom embedded product is.

The regulatory choice that decides everything

This is the first decision. It shapes every other one.

Become a licensed carrier. You hold the risk on your balance sheet. You set your own rates. You keep the full premium. The catch: 18 to 36 months for licensing, $5M or more in capital requirements per state, an appointed actuary on staff, and annual reserve audits. This is where Lemonade sits. Plan for 3 to 5 years and $20M or more before you are operational at scale.

Become an MGA. You program and distribute policies, but a carrier partner holds the risk. You earn a percentage of premium. The carrier takes the rest and holds the regulatory capital. Partner with a fronting carrier: State National, Markel, and Employers are the most MGA-friendly in the US market.

The MGA path lets you launch with $3M to $5M in total capital instead of $50M or more. Your lawyers file the MGA paperwork in parallel with development. Within 6 to 9 months of founding, you can be writing policies.

McKinsey's 2023 InsurTech survey found that 65% of insurance startups that launched between 2018 and 2022 used the MGA model. Only 12% pursued carrier licensing from day one. The numbers confirm what the regulatory timeline makes obvious.

Most InsurTech startups that matter today (Hippo, Branch, Openly) launched as MGAs and built carrier licenses later, once they had data and traction.

Build vs. Lemonade: when does custom win?

Keep using Lemonade (or buy a white-label platform) when:

• You are entering a vertical Lemonade already serves well (basic renters, standard homeowners) and do not have underwriting expertise to price it differently

• Your total addressable market is under $10M gross written premium -- the build cost is hard to justify below that threshold

• You need to be in market in under six months and do not have a fronting carrier relationship in place

Build your own when:

• You control the distribution channel and can embed insurance at the point of purchase -- the MGA commission you earn becomes a new revenue line, not a cost

• Your vertical requires underwriting logic Lemonade does not support (gig worker income protection, short-term rental liability, B2B professional liability)

• You are a property manager, HR platform, or marketplace that already has 10,000+ customers who could be policyholders -- the build cost is recoverable in two to three years of MGA commission

• You are a legacy carrier that needs a modern digital front-end without replacing your existing policy administration system

The payback math is straightforward for distribution-first builds. A property management company with 10,000 units converting 40% of tenants to renters policies at $200/year average premium earns $800K in annual gross written premium. At 20% MGA commission, that is $160K/year. The $90K to $160K build cost pays back in 7 to 12 months.

What to build first, second, and third

V1 -- launch (18 to 22 weeks, $90K to $160K)

The product that needs to exist on day one: quote accurately, bind cleanly, handle basic claims, and not lose customer data.

Cross-platform mobile (React Native or Flutter) saves $15K to $25K versus native iOS and Android. Insurance apps do not require performance-intensive animations. Clarity and reliability matter more. The quote flow, policy management screens, and claim filing interface need to work on a five-year-old phone with a slow connection.

The policy data needs a relational database. Insurance data is relational by nature: policies have endorsements, endorsements have effective dates, claims have coverage checks against active policies at the date of loss. A document database makes these queries brittle. This architecture choice adds one to two weeks upfront but prevents a painful rewrite as volume grows.

V2 -- growth (months 6 to 12, add $30K to $50K)

Once you have your first 500 to 1,000 policyholders and understand the loss patterns:

• AI damage assessment integration: vendor APIs (Tractable, CCC Intelligent Solutions) analyze claim photos and return estimated repair costs. Small claims below a threshold get approved automatically. This is how Lemonade handles 30% of claims in under three seconds. Adding it post-launch costs $12K to $20K.

• Automated renewal engine: manual renewal at 500 policies is manageable. At 5,000, it is not. Automated renewal quotes, notices, and collection add $8K to $12K.

• Agent portal: if you are distributing through insurance agents or brokers, they need their own view. This is separate from the consumer app and typically costs $10K to $18K.

V3 -- scale (above 10,000 policies)

• Proprietary underwriting model: most MGA founders start with vendor APIs for actuarial rating and move to proprietary models once they have enough loss history. This requires an actuary on staff or retainer and a data science investment.

• Reinsurance reporting: your reinsurance partners require detailed loss reports. The reporting layer is straightforward but time-consuming to get right.

• Multi-state expansion automation: each new state requires rate filings, regulatory approvals, and compliance checks. Automating the state-expansion process becomes a meaningful operational lever above 20 active states.

How the quote engine actually works, and why it matters

"The quote engine is where most InsurTech companies lose. They underestimate actuarial complexity and ship rates that are either too high to be competitive or too low to be solvent." -- Caribou Honig, co-founder of InsurTech Connect, in a 2022 Forbes interview

A quote engine takes risk inputs and produces a premium. For renters insurance, those inputs might include ZIP code, apartment size, construction type, prior claims history, and coverage limits selected. Each factor adjusts the base rate. The math behind that adjustment is actuarial, built from loss history data spanning years and geographies.

Three build options exist, each with different cost and timeline trade-offs.

Build from scratch with an actuary. Your fronting carrier requires actuarial sign-off on your rate filings anyway. The actuary defines the rating algorithm, your team implements it, and the carrier files the rates with each state insurance department. Most flexible option. Also the slowest. Plan for 10 to 16 weeks just for the actuarial work.

Use a rating engine vendor. Majesco, Duck Creek, and Instec offer configurable rating platforms where your actuary sets the rules and the platform handles the calculation. You integrate via API. Faster to reach a filed rate, but you add a vendor dependency and a monthly licensing fee of $3K to $8K.

Use the fronting carrier's rate book. Some fronting carriers let you use their existing approved rates at launch, then transition to your own filed rates once you have loss history. Fastest path to launch. You give up pricing differentiation until your own rates are filed.

The failure mode we see most often: founders who underestimate the time required for rate filings. State insurance departments move slowly. Plan for 60 to 90 days per state for rate approval. Build your launch timeline backward from your target go-live date and work out how many states you need approved before launch.

How claims processing works, and what it costs to get it wrong

Lemonade's 2023 annual report shows they process 30% of claims in under 3 seconds using their AI claims bot. The other 70% go to human adjusters. For a new platform, that 30% straight-through rate is a meaningful cost reduction and a genuine customer experience differentiator.

NAIC's 2022 data shows mid-term policy changes (endorsements, reinstatements, and cancellations) account for 40% of policy administration workload. Claims processing is the operational load that surprises most founders. Manual adjudication at volume becomes expensive fast.

FNOL (first notice of loss) is the customer's initial report. Date of incident, type of loss, brief description, estimated value. For property damage, photo upload is required. For theft, a police report number. Your FNOL intake must be fast and mobile-first. Most customers file from their phones immediately after an incident.

AI damage assessment analyzes the photos. Vendors like Tractable and CCC Intelligent Solutions return an estimated repair or replacement cost with a confidence score. For clear-cut damage below a set threshold, the claim is approved and paid automatically. Set the threshold conservatively at launch ($250 to $500) and raise it as accuracy data accumulates.

Human adjuster routing handles everything else: high-value claims, ambiguous damage, fraud flags, disputed coverage. Every decision needs a reason code. Your reinsurance partners will ask for it.

Premium collection runs through Stripe or Braintree. Claim payouts go out via ACH transfer (faster and cheaper) or physical check (some policyholders expect it). Most states require payment within 30 days of claim approval or the carrier faces penalties.

What we have seen go wrong in insurance builds

The most common failure mode in insurance platform builds is the policy data model. Founders treat it as a simple database schema and discover the edge cases mid-build: what happens when a policy lapses and is reinstated, and a claim was filed during the lapse period? What happens when an endorsement changes coverage limits, and a claim references an incident that occurred before the endorsement? Getting the effective-date logic wrong creates claims disputes that require manual intervention on every case. Teams that design the data model with an insurance domain expert upfront save four to eight weeks of rework.

The second failure mode is the fronting carrier relationship. Developers can build the platform in 18 weeks. The carrier diligence process -- reviewing your technology, your compliance infrastructure, your rate filings, and your reinsurance arrangements -- takes its own time. Founders who start carrier conversations after development begins end up with a finished product waiting on a carrier for 60 to 90 days.

How RaftLabs fits

We have built digital insurance platforms and MGA tools, including quote engines, FNOL intake flows, and AI claims integrations. If you have a vertical in mind and a fronting carrier conversation underway, the first useful conversation is a 30-minute scoping call to map your quote engine requirements against your filed rate timeline. That conversation determines whether your go-live date is realistic.

Book that call here.

Related reading: AI Agents for Insurance covers how AI fits into underwriting, fraud detection, and claims automation once the platform is live.