At this year’s Insurtech Insights Conference in New York, held June 3–4, (where Evolve hosted our annual private, invite-only rooftop event following the main programming, with photos below) we spent the two days in back-to-back meetings with carriers, MGAs, software and services providers, and capital partners. Last year, we left writing that true disruption still felt some distance away. Twelve months later, the industry gave us its answer, and it largely proved that read right.

ITI USA 2026: A Lot of AI, Not Much Movement

What Evolve saw at Insurtech Insights in New York, June 3 to 4, 2026

Evolve was back at Insurtech Insights (ITI) USA in New York this June, our second consecutive year at the conference and part of a long-running presence across technology-enabled insurance services, specialty MGAs, distribution, TPAs and claims, and software. Last year, we left writing that true disruption still felt some distance away. Twelve months later, the industry gave us its answer, and it largely proved that read right. We spent the two days in back-to-back meetings with carriers, MGAs, software and services providers, and capital partners, then kept the conversations going at our annual invite-only mixer that evening. What follows is our read.

Our read this year can be summed up in one line, and it came from the main stage:

there is more AI than ever, and it has barely moved the numbers that actually matter.

Key Themes Evolve Observed

1) AI is everywhere, but the results are not in the financials yet

Takeaway: AI funding has decoupled from AI outcomes, and the industry’s core ratios are still waiting on a payoff.

• “AI-first” positioning was higher than at any prior event, yet operating proof was thinner. Most pitches recycled the same use cases behind an LLM layer: data collection, faster underwriting, claims intake.
• Per Gallagher Re, AI-focused companies captured more than 95% of the ~$1.6B in global Insurtech funding in Q1 2026, even as core ratios sit still. Bain sizes the GenAI opportunity in P&C claims at over $100B, but straight-through processing still averages below 10%.
• Hippo, on the main stage, has now strung together four straight profitable quarters at a break-even combined ratio. It took roughly a decade to get there, a useful calibration on how long these gains actually take to show up in the financials.

2) The language moved from “wrappers” to “agents”

Takeaway: Agentic AI is real for a few, marketing for many, and regulated deployment is where the gap is widest.

• A year ago the critique was thin GPT wrappers on legacy stacks. This year the vocabulary advanced to agentic systems, with agents collapsing quoting timelines the clearest example, and Celent finds 22% of insurers plan an agentic solution in production by year-end 2026.
• The credible end of the spectrum exists. The opening Allianz-Anthropic keynote reflects a January 2026 partnership deploying agentic workflows in claims with human-in-the-loop oversight and decision logging built for insurance regulation.
• Carriers furthest along made a related point: these tools raise the premium on judgment rather than removing it. And the constraint is often the client, not the technology.
• A claims provider we spoke with described carrier customers that bar AI and agents from their systems outright, compressing margins on work that should be getting cheaper.

3) Private equity is cooling on software, and the reason is AI itself

Takeaway: The software premium is compressing because AI is shortening the build-versus-buy decision, and public and private markets are both adjusting.

• This was the sharpest shift from prior events. The PE view is that software is becoming less attractive because AI itself threatens to disrupt it: foundation models lower the cost of building internally, while asking multiples have not adjusted.
• Software-focused buyout value fell to ~$50B in the first five months of 2026, down from ~$88B a year earlier and the weakest start since 2020. Listed insurance distributors felt it as well, with the largest brokers selling off sharply in a single February session on AI disruption concerns, a markdown that has largely held.
• An ITI M&A panel of investors and bankers added the discipline point directly, warning that successful Insurtech raises are pricing at just 1 to 2.5x premium and forecasting a fresh crop of zombie companies as a result.

4) MGAs and specialty distribution remain the steadier story

Takeaway: Specialty distribution is the most consistent growth corner of the market, and the volume of capital following it suggests that won’t change soon.

• Specialty MGAs continued to show up in force, still searching for carrier capacity and increasingly focused on a defined risk type. Conning counts $114.1B of US MGA premium for 2024, up 16% YoY and roughly a tenth of the US P&C market, with fronting carriers supporting more than $18B of it.
• The deal tape backs it. Weeks before the conference, Sands Point Risk, a multi-program MGA platform barely two years old, acquired Launch Environmental Underwriters, its largest deal to date, pushing combined gross written premium past $250M while launching new programs in political violence and terrorism, aviation, and D&O.
• We expect this end of the market to keep performing for years, with AI a tool layered onto sound specialty underwriting rather than a substitute for it. Embedded distribution had its own proof point: PayPay, with 70M+ users, agreed to acquire ~70% of T&D Financial Life for ~$840M, a wager that owning the moment of need at scale is worth paying for.

Who We Met, and Why It Was Interesting

Highlights below are generalized and anonymized:

• A cloud-native policy administration platform.
  Near-zero churn over the past decade and high switching costs, proof the unglamorous core-systems layer still commands loyalty and pricing power flashier AI entrants cannot replicate.
• An AI-native document processing platform.
  Around $10 million in ARR converting unstructured files into validated data; a test case for the cycle, since its defensibility depends on staying ahead of general-purpose foundation models.
• A travel-insurance BPO reaching profitability.
  AI voice automation streamlining claims intake, with one new client large enough to carry it to profitability; AI applied to a real workflow rather than sold as the product.
• A growth-stage specialty carrier.
  A workers’ compensation underwriter expanding into new states and openly weighing M&A, evidence that traditional risk-bearing paired with sensible technology still attracts capital.

The Evolve Team Takeaways

• Even earlier, even smaller.
  The barbell of large incumbents and very early teams was more pronounced than ever, with many firms under a year old and at or below $500K in revenue; Series B, C, and D companies were notably thin.
• AI vendors are becoming targets, not just sellers.
  Established services players are acquiring AI capability rather than building it. The same week, Admiral completed its £80 million acquisition of AI-native telematics firm Flock, folding the models and data into its own underwriting, and advisors increasingly call technology the cost of entry rather than an advantage.
• “AI” is losing precision as a descriptor.
  The companies that held attention could name the specific process they improve and the metric it moves. The next gap is similar: models can flag a risky exposure instantly, but little of the stack makes that exposure smaller, ground where durable franchises may be built.
• Desire for capital/partnerships from U.S.
  A meaningful share of the firms we met were based outside the US, from Europe and Singapore to India, Canada, and Southeast Asia, most seeking a US path through capital or a partner.

How Our Read Aligns With Public Recaps

Public coverage of ITI leans on attendance figures and product announcements, and the official closing narrative declared AI’s place in insurance settled, turning the conversation to the data foundation beneath it. Our read is that AI is settled as a topic, not yet as a result, and the data side supports it: WTW finds insurers furthest along in analytics ran combined ratios six points better than slower adopters, yet 42% of P&C insurers still cite data quality as their top barrier. The most important signal this year was not a launch. It was the candid acknowledgment, on the main stage and in the hallways, that the industry’s core metrics have not yet responded to the AI build-out. For founders and investors, that is the more useful truth to plan around.

Evolve’s Merge & Mingle Mixer Photos:

Contact us at Evolve:

If you have questions about the evolving landscape of technology-enabled insurance services or software reshaping the industry, we invite you to connect with the Evolve team. Whether you’re a founder or investor considering a transaction now or planning for the future, we would welcome the opportunity to speak with you. Evolve continues to be an active and growing presence in the insurance space, having advised on more than half a dozen insurance transactions over the past two years.