4 tips for building AI agents your business can trust

If your company hasn’t started using agents yet, it will soon—either through off-the-shelf products or in-house tools that draw on large language models and data sources. Joel Hron, CTO at Thomson Reuters Labs, says the firm mixes internal models with commercial tools and focuses on synthesizing human expertise into judgment.

That delivery mechanism is shifting from traditional software toward agents or agents plus software; examples at Thomson Reuters include Westlaw Advantage and the Deep Research agent. Hron’s first lesson is to measure success. Define what good looks like and systematize evaluations.

Thomson Reuters uses public benchmarks, internal benchmarks with clear criteria, and automated evaluations to iterate quickly, but keeps human experts in the loop to provide the final confidence before shipping. The second lesson is to make experts sit together.

ai agents, llms, thomson reuters, joel hron, westlaw advantage, deep research, human experts, benchmarks, automated evaluations, in-house tools