Brown & Brown Selects AI Transformation Partners
Insurance broker Brown & Brown said on July 23 that it enlisted Anthropic, McKinsey and Accenture for an enterprise AI program and will expand Claude Code across its software engineering organisation. The company reported pilot productivity gains of roughly 2x to 8x and 80%–90% faster troubleshooting in some use cases. Those figures are company estimates without a disclosed sample size or independent validation, so they are early signals rather than established benchmarks.
Brown & Brown announced on July 23, 2026, that it has enlisted Anthropic, McKinsey and Accenture to support an “AI-first” enterprise transformation. The insurance brokerage plans to expand Claude Code across its software engineering organisation and create a Value Management Office to track adoption, governance and business outcomes.
The rollout combines local experimentation with central oversight
Brown & Brown says it wants local teams to develop AI uses close to customer and operational needs, then scale successful work across the company. Anthropic, McKinsey and Accenture will collectively provide AI technology, transformation and governance support.
The company did not disclose the value or duration of the engagements, a detailed deployment timetable, or the first business functions beyond software engineering that will move into production. It said the program is intended to improve customer service, operating efficiency, employee productivity and long-term growth.
A new Value Management Office is meant to provide central discipline around the decentralised rollout. Brown & Brown says the office will monitor adoption, measure return on investment and maintain governance as AI use expands.
Early Claude Code results are self-reported
Brown & Brown said participating development teams saw productivity gains ranging from about 2x to 8x during pilots, with some projects that previously took days completed in hours. It also reported that AI-assisted workflows reduced analysis and troubleshooting time by an estimated 80% to 90% in certain use cases and found software vulnerabilities that other tools had not detected.
The company said approximately 80% of participating engineers gave Claude Code the highest available value rating. These are meaningful internal signals, but the announcement does not disclose the number of engineers, the comparison period, the task mix, the quality controls, or whether output was measured through completed work, code volume or another metric. No independent evaluation accompanied the release.
That makes the figures useful as pilot observations, not a general productivity benchmark for insurance or software teams.
What will determine whether the program scales
The important next evidence will come from production outcomes: which workflows are deployed, how Brown & Brown measures quality and security, whether faster development reduces or shifts review work, and whether customer or financial results improve. The Value Management Office could make those outcomes more auditable if it publishes consistent baselines and evaluation methods.
The announcement is notable because it combines a frontier-model provider with two large transformation consultancies and an internal governance function. Its practical significance will depend on whether Brown & Brown can turn early development gains into repeatable business results without weakening privacy, security, regulatory compliance or human accountability.
Key Points
- 1Brown & Brown enlisted Anthropic, McKinsey and Accenture and plans to expand Claude Code across its software engineering organisation.
- 2The company reported pilot gains of roughly 2x to 8x and 80%–90% faster analysis and troubleshooting in some use cases, but did not disclose sample size or evaluation methodology.
- 3A new Value Management Office will track adoption, governance and return on investment as local experiments are considered for wider deployment.
Scoring Rationale
The program is a substantial enterprise deployment signal in regulated insurance and includes concrete pilot estimates plus a governance structure. Its impact remains provisional because the results are self-reported and the announcement omits sample size, evaluation design, contract value and production outcomes.
Sources
Primary source and supporting public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems
