5 Things We Heard At WCI 2026

Four days, countless conversations, and a lot to unpack. The SiftMed team left the Workers’ Compensation Institute’s (WCI) Annual Conference in Orlando with a clear view of where workers’ comp insurance is headed - and plenty to think about.

The conversations covered a lot of ground, and several themes stayed with us. Here are five takeaways that stood out.

1. People Are Over The AI Pitch

Over the years, the conversation around AI has shifted - from hesitancy and job concerns, to curiosity about what it can do and how to implement it safely. At WCI, we saw the conversation shift again: claims professionals aren’t asking for “more AI,” they want fewer steps in the process. 

The question is no longer “what can AI do for me?” but “what can AI meaningfully take off my plate?” 

When vendors stop pitching and start listening, the ask is clear: fewer portals, less manual searching, and faster turnaround. The appetite for innovation is there, but so is the expectation that technology should make work simpler, not add another layer of complexity.

The gap between what AI makes possible and what’s actually being used in claims is still significant. Closing that gap means moving beyond adopting AI for AI’s sake and strategically pointing AI at the specific challenges that are slowing teams down.

2. Defensibility Is Table Stakes

As AI moves deeper into the claims process, being able to prove where an answer came from is becoming a baseline expectation. 

Claims professionals need to know:

• Can you trace a conclusion back to the specific page it came from? 

• Is there an audit trail? 

• Does someone on your team own this decision, or does it belong to AI?

The fundamentals haven’t changed - claims decisions need to be supported by evidence and traceable. What’s changing is the scale and complexity of the work - AI can assist in many parts of the process, but an AI-generated insight is only as good as the adjusters’ ability to validate and support it. As AI integrates deeper into claims processes, teams need clear visibility into where the data came from, what supports it, and how it was generated.

An output that can’t be tied back to its source document creates a new point of risk, that’s why human oversight from the experts reviewing claims is especially important.  

3. Fraud Hides In Disconnected Data

One of the most compelling examples we heard was a reminder that better fraud detection doesn’t always start with a smarter model or new process - it starts with connected data.

When FedEx consolidated its operating companies in 2023, claims data that had previously lived in separately could finally be reviewed in the same systems. The result was broader visibility that they hadn’t had before, and new patterns began to emerge - the same vehicles, doctors, addresses, and treatment progressions were turning up across hundreds of litigated files.

What started with one person manually flagging suspected claims evolved into a structured program with Gallagher Bassett, mapping those connections across addresses, phone numbers, and emails (with some linked back to hundreds of claims). 

Those patterns weren’t immediately visible because someone bought a smarter tool, it became visible because the underlying records stopped being scattered across systems that didn't talk to each other.

4. Nobody Names The Adjuster Tenure Problem

The adjuster tenure problem came up in multiple sessions and nearly every conversation, yet nobody seems to be drawing enough attention to it.

Adjusters are leaving faster than they can be replaced, while claim volume and complexity continue to grow; leaving newer adjusters with less institutional knowledge to draw from. They’re being asked to make decisions on increasingly complex claims without the years of experience that help an adjuster recognize patterns, spot red flags, and know what to look for.

The U.S. Chamber of Commerce projects that half the current insurance workforce will leave within fifteen years, opening more than 400,000 positions the industry cannot fill at its current pace. Post: The Claim Is Decided In The Medical Record And Nobody Has Time To Read It

The challenge isn't only to get new adjusters up to speed. It's finding ways to retain the expertise that comes with experience and make that knowledge accessible across the entire claims team.

5. "AI Can't Be Cross-Examined"

One of the most memorable quotes from the conference was “AI can’t be cross-examined.” It’s a useful reminder that while AI can analyze records, summarize data, and surface patterns, it shouldn’t be the one drawing conclusions.

An expert can be cross-examined on inconsistencies between what they said last time and what they're saying now. AI doesn’t have the same accountability; it can’t explain its reasoning, provide context, or take accountability for decisions as it lacks the professional judgment and experience required for claims.

That doesn’t make AI less valuable. The strongest use of AI is to strengthen the expert’s work, not replace the expertise. When a report or claims decision needs to withstand scrutiny, a qualified human still needs to understand the records, apply professional judgment, and stand behind decisions.

What We’re Taking Away

If there was a common thread across these conversations, it’s that the next phase of AI in workers’ compensation claims won’t be defined by how many impressive features or new tools we can add. It will be defined by how effectively we can remove friction, preserve and enhance expertise, connect information, and give people the confidence to work with what AI can help with. 

Looking at it from this lens creates a more practical, and more exciting, opportunity for the industry. The tools are evolving quickly, now the focus needs to be using AI to make the work easier, make the answers defensible, and help us do more with the information and expertise we already have.