Insurance Costliest Bias

New Report Finds 91.5% of IT Projects Fail Budget or Timeline; 95% of Enterprise GenAI Initiatives Produce No P&L Impact

Press Release

Insurance

Written by

Published on

April 17, 2026

MIAMI & OSLO, Norway (April 17, 2026) — Artem Gonchakov, CEO of Simplifai, the leading provider of Agentic AI for P&C insurance claims, has released a comprehensive analysis examining one of enterprise technology’s most consequential and consistently mishandled decisions: the choice between building custom software and purchasing purpose-built commercial solutions.

The report, titled “The Paradox of Build vs. Buy: Why Smart Companies Keep Making the Same Expensive Mistake,” draws on three decades of published research, landmark behavioral science, and documented case histories to make the case that the desire to build software is not a strategy but rather a cognitive and organizational bias with nine-figure consequences.

Rooted in peer-reviewed research from Harvard, Oxford, MIT, and Wharton, the analysis identifies psychological mechanisms including the IKEA Effect, the planning fallacy, optimism bias, and the illusion of control. Each of these systematically distort build-vs-buy decisions in favor of internal development, regardless of the financial or operational evidence.

Key Findings

Among the report’s most striking data points:

• 91.5% of IT projects go over budget, over schedule, or both (Flyvbjerg, Oxford University)

• 78% of lifetime software Total Cost of Ownership occurs after launch — making the day of release the beginning of the expense, not the end

• 95% of enterprise generative AI initiatives fail to reach P&L impact (MIT, 2025)

• Over 40% of agentic AI projects will be cancelled by end of 2027 due to escalating costs, unclear business value, or inadequate risk controls (Gartner)

• The average cost overrun for large IT projects is 447% for projects that go over 50% — not a project management problem, but a documented human failure to forecast complex work

The AI Paradox: Faster to Build, Slower to Value

The report dedicates particular attention to the generative and agentic AI era, arguing that the rapid acceleration of AI coding tools has made the build bias worse, not better. The speed at which organizations can now generate working prototypes has created a widespread illusion that the hard part of software development has been solved. It has not.

Gonchakov distinguishes between a working demo and a production-grade enterprise system — secure, observable, compliant, integrated, and maintainable at scale. “AI tools accelerate the first day,” the report states. “They do not accelerate the next five years.”

Implications for the Insurance Industry

The analysis includes a dedicated chapter on the insurance sector, an industry Gonchakov identifies as a concentrated mirror of every dynamic described in the broader research. With insurance technology spending forecast to reach $420 billion by 2033, and AI now consuming 36% of insurance IT budgets, the report argues the build-vs-buy decision is being made in real time at every major carrier. This is being done often under the same structural biases that have produced decades of legacy debt.

The report finds that carriers who purchased modern platforms rather than building them are already outperforming legacy-burdened competitors on combined ratios, claims cycle times, and customer satisfaction. The competitive gap, Gonchakov argues, will widen materially over the next three years.

“I have never once heard a logically compelling argument for building software that a purpose-built vendor already offers at scale. What I have heard, repeatedly, is a variation of the same words: ‘We want to do it ourselves.’ That is not a strategy. That is a preference. And preferences, when they carry nine-figure consequences, deserve a great deal more scrutiny than they typically receive,” said Artem.

A Framework for Removing the Bias

The report concludes with a practical decision framework built around three questions every organization should answer before approving a build initiative: What does the historical reference class for similar projects actually show? What is the honest five-year Total Cost of Ownership comparison? And who benefits from the recommendation being made?

Gonchakov also identifies the four legitimate cases in which building custom software is justified: when software is the core product and competitive differentiator, when no adequate commercial solution exists, when deep proprietary data creates a genuinely defensible advantage, or when regulatory requirements preclude commercial options. The report sites these exceptions are real but far narrower than most organizations acknowledge.

Availability

"The Paradox of Build vs. Buy" is available in full on the Simplifai website at https://www.simplifai.ai/blogs-and-news/paradox-of-build-vs-buy.

About Simplifai

Simplifai is transforming insurance claims through Agentic AI with AI Agents that execute the complete claims lifecycle under human supervision. Simplifai’s Agentic AI allows insurers, TPAs, and MGAs to automate the full claims lifecycle, from intake and processing to payment and closure, with human oversight built in by design. The company currently serves leading insurance organizations across Europe and North America and is expanding rapidly through its 2026 growth strategy. Visit Simplifai for more information.

Media Contact:

Stacia Kirby, 206-478-5841

stacia@kirbycomm.com