Introducing the AI Model ‘Harness’

Good morning. As AI continues to evolve rapidly, there’s an increasingly important concept for business tech leaders to think about, and it’s called the AI model harness.

The harness around a model is the code that actually runs the AI system, and it does things like provide memory and business data context for the AI model, and lets AI agents take actions, according to David Pan, a director and AI industry practice lead at Moody’s.

In other words, the software harness around an AI model is simply a system wrapped around the brains, which are the AI models, Pan recently told me. A harness allows users to control and direct models, much like a harness allows a rider to guide a horse.

The idea of the model harness started becoming more prominent this year, developing alongside reasoning models. The fusion of a reasoning model with a capable harness allows the model to connect to real systems, execute code and manage workflows, according to Anthropic.

But what makes the harness so important for enterprises? It’s a way for companies to take back control of their AI, Pan says.

Developing their own software around AI models—a practice he calls “harness engineering”—gives businesses a way to decouple their workflows from the models themselves. And that helps them become less reliant on a single AI provider.

“If you bring that harness in-house and control it, you’re baking in a lot more business resilience,” Pan said.

While labs like OpenAI and Anthropic do offer their own model harnesses for enterprise customers, Pan argues that businesses in regulated sectors like banking and government should build their own to keep their workflows private.

But there are benefits to using a vendor-built harness, too. The biopharmaceutical giant Bristol-Myers Squibb chose Anthropic’s Claude as its “standard harness” to avoid rebuilding basic infrastructure tooling, according to Greg Meyers, its chief digital and technology officer.

Another critical component of a model harness is a router—a piece of software that can automatically choose between frontier and cheaper models for various tasks. And that’s an increasingly important tool businesses are relying on to keep their AI token costs down.

Moody’s, the credit-ratings and research company that has been around for over a century, built its own model harness-like tool called the Research Assistant. The assistant, which is an AI agent chatbot, is able to use different Moody’s data sets and can switch between AI models on the back end, Pan said.

More fundamentally, as tools and techniques around AI models continue to develop, the basic need to provide models with business data will always exist, Pan says. Whether the practice is called context engineering, prompt engineering or harness engineering, “what doesn’t change is the ability to supply language models with the right context,” he said.

Have you built your own AI model harness? Or are you using one built by a vendor? Send your feedback to me at belle.lin@wsj.com (if you’re reading this in your inbox, you can just hit reply).

Founders have long put in punishing hours in the name of building the next big thing. But the growing capabilities of AI agents—and the speed at which the models powering them are evolving—give new meaning to working yourself to the bone, The Wall Street Journal reports.

The more work AI agents do, the more founders find themselves working. Add to that the pressure of what many view as a once-in-humanity moment in technology and you get erratic sleep schedules, a struggle to focus on anything but work, and the feeling that even though the pace isn’t healthy or sustainable, you just can’t stop.

The WSJ Tech Council brings together CIOs, CTOs and CISOs advancing innovation and shaping the future. Join this trusted community where tech executives connect with peers to explore emerging trends and gain the perspective they need to stay ahead of disruption.

Request an Invitation

Follow Isabelle Bousquette on LinkedIn, Instagram, X, and TikTok for more behind the scenes on her tech and AI coverage, and lately, her contributions to the WSJ Leadership Institute’s new Executive Resilience series, where she’s profiling America’s top execs about their fitness and wellness habits.

Follow Belle Lin on LinkedIn and X for her latest reporting on enterprise technology and AI.

Steven Rosenbush is chief of the enterprise technology bureau at the WSJ Leadership Institute. He also has a column. You can follow him on LinkedIn.

Tom Loftus is the editor of The Morning Download. He suggests following Isabelle, Belle and Steve on their various social channels. But if you insist, here’s his LinkedIn.