Quick Answer: 3 AI Programs You Can Use for Claims Adjusting
Insurance adjusters do not need to learn every artificial intelligence (AI) platform available.
Claims professional of 35 years, Jim Cresse, recommends starting with the task you want AI to help you complete. Then choose two or three tools that fit those needs and learn how to use them well.
The three tools he currently prefers are:
- Perplexity for research and finding original sources
- Claude for analysis, report work, and technical tasks
- Midjourney for diagrams, images, videos, and training materials
The best tool is not necessarily the most popular one. It is the one that best fits the work in front of you.
| About these recommendations: This article reflects Subject Matter Expert Jim Cresse’s experience testing AI tools for claims-related work. Neither Cresse nor AdjusterPro was paid by any AI company mentioned in this article. Editor’s note: AI tools, features, pricing, and security practices change quickly. Adjusters should continue testing which platforms best fit their needs and align with their employer’s privacy, security, and claims-handling requirements. |
Which AI Tools Are Worth an Adjuster’s Time?
New AI tools seem to be appearing every week. Each one promises to help you research faster, write better reports, or create stronger visuals.
That can make it hard to know where to start. Which tools are worth your time? Which ones can actually help with claims work?
To answer those questions, AdjusterPro spoke with Jim Cresse, a National General Adjuster with more than 35 years of experience and nine AI-related certifications. He has spent significant time testing AI tools and exploring how they may support adjusting work.
AdjusterPro provides training and education for insurance adjusters, but we do not sell or officially promote the tools covered in this article. Our goal is to share Cresse’s recommendations so you can decide what may work best for you.
His advice is simple: Do not start with the tool. Start with the task.
Are you researching an unfamiliar repair method? Comparing estimates? Reviewing a report? Creating training materials? Once you know what you need to accomplish, you can choose one or two tools that support that task and learn them well, rather than trying to learn 15 platforms at once.
By the end of this article, you’ll know why Cresse recommends Perplexity, Claude, and Midjourney, how he uses each one, and whether they may fit your work.
Table of Contents
- Perplexity Helps With Research
- Claude Helps With Analysis, Reports, and Technical Tasks
- Midjourney Supports Visual Training
- Testing AI-Generated Estimates
- Which Tool Should You Learn First?
- A Note About Confidential Claim Information
- Build Claims Knowledge Alongside Your AI Skills
Perplexity Can Help With Research
Perplexity is one of Cresse’s preferred tools for research.
What stands out to him is its ability to provide information while also showing where that information came from. Instead of receiving only a written response, the user can examine the supporting sources.
That may be useful when an adjuster needs to research:
- An unfamiliar construction method
- A type of repair
- Industry terminology
- A technical process
- A general insurance or claims topic
For example, an adjuster reviewing damage involving an unfamiliar roofing system could use Perplexity to identify relevant terminology and locate sources explaining the materials and repair process.
Perplexity can help you find a starting point. You still need to open the source and confirm that it supports the answer.
The source must also apply to the specific claim, policy, location, or situation you are handling.
Claude Can Help With Analysis, Reports, and Technical Tasks
Claude is another tool Cresse prefers. He has used it to write computer programs in Python and sees value in its ability to support complex analysis and report work.
An adjuster could use a tool like Claude to compare two repair estimates.
The response might identify that:
- One estimate includes line items missing from the other
- The quantities are different
- The scopes call for different repairs
- One estimate contains work that needs closer review
For example, a contractor’s estimate may include drywall replacement, additional labor, or more material than the carrier’s estimate. AI can help identify those differences so the adjuster knows where to begin the review.
Cresse has also used AI to examine report language and organization. A model may suggest clearer headings, improve phrasing, or point out information that appears to be missing.
AI can organize and analyze information. The adjuster remains responsible for determining whether the result is accurate and appropriate for the claim.
Midjourney Can Support Visual Training
The third tool on Cresse’s list is Midjourney.
He prefers it for creating photographs, videos, diagrams, and other visual materials. He also sees potential for it to support training programs that combine visuals with instruction.
Visual content can be especially useful in adjuster education.
For example:
- A diagram could identify the parts of a roofing system
- An image could support a lesson about building materials
- A sequence of visuals could explain an inspection process
- A video could reinforce workplace safety training
These materials still require someone who understands the subject, reviews the output, and explains how the lesson applies in the field.
AI can help create the training. It cannot replace the knowledge behind it.
Generated visuals must also remain separate from real claim evidence. An AI-created image can illustrate a concept, but it does not document what happened during an actual loss.
What Cresse Learned From Testing AI-Generated Estimates
ChatGPT is not one of Cresse’s three current recommendations, but he has tested it alongside Claude for claims-related work.
In one experiment, Cresse gave the same photo narrative to ChatGPT and Claude and asked each tool to create a detailed repair estimate.
One estimate came back about 15% lower than the estimate he had prepared. The other came back about 15% higher. Each tool produced a basic estimate in about 20 to 30 minutes.
Cresse saw possible value in using an AI-generated estimate as a starting point for reviewing the scope or thinking through an initial reserve. But he emphasized that you should not treat it as the final repair estimate or settlement decision.
| This experiment highlights an important distinction: An AI tool does not need to produce a perfect final answer to be useful. It may save time by giving the adjuster an organized starting point. But the adjuster must still review the scope, correct any mistakes, and determine whether the estimate accurately reflects the actual loss. |
Which AI Tool Should You Learn First?
Choose the tool according to the work you need to complete.
| Tool | Best Starting Point If You Need To |
| Perplexity | Research unfamiliar topics and check sources |
| Claude | Compare documents, review reports, analyze information, or complete technical tasks |
| Midjourney | Create visuals, diagrams, images, videos, or training materials |
After choosing a tool, test it on one narrow task. Learn where it helps, where it struggles, and when you need to verify the answer.
Choose the task first. Choose the tool second.
A Note About Confidential Claim Information
Before uploading photographs, estimates, reports, or other claim materials, follow your employer’s security policy and confirm that the platform is approved.
Cresse removes insured names, locations, and other identifying information before submitting materials to an outside model. He may also create a separate file containing only the photographs needed for the task.
Do not submit confidential claim information to a public AI tool without authorization.
Build Your Claims Knowledge Alongside Your AI Skills
AI tools are most valuable when the person using them understands the claims process.
An adjuster still needs knowledge of:
- Inspections
- Estimating
- Documentation
- Communication
- Claims handling
- Professional responsibility
- State licensing requirements
Without that foundation, it is difficult to recognize when an AI answer is incomplete, inaccurate, or inappropriate for the claim.
The strongest career strategy is not choosing between claims knowledge and AI skills. It is developing both.
If you are preparing to enter the profession, start by understanding your state’s licensing requirements and completing the education needed to qualify.
If you are already licensed, certifications and continuing education can help deepen your knowledge and prepare you for more specialized work.
Then add the AI tools that support the work you are trained to perform.
