As artificial intelligence (AI) technology becomes more prevalent across the insurance industry, many firms are looking closely at a wide range of AI applications to help streamline operations, enhance customer service, or improve underwriting. In a previous ViewPoint article (Navigating The World of AI for Insurance Brokers) MarshBerry reviewed the basics of AI and how this modern technology can help insurance brokerages and which options are available. But with so many options on the market, how do you know which AI solution is right for your business? The key, obviously, is to ask the right questions before making any commitments. But can you recognize the right answers?
Here are ten essential questions you should ask any prospective AI vendors, along with hypothetical examples of good answers – as well as responses that should raise red flags!
Question 1: What specific insurance-related problems does your AI program solve?
A vendor should be able to clearly articulate how its AI helps with industry-specific challenges.
- Good answer: “Our AI helps predict policyholder risk, automates claims processing and provides insights into customer behavior for cross-selling.”
- Red flag response: “It can do many things,” followed by a list of “things” that are not specific to the insurance industry, such as “save time,” “reduce overhead,” “increase efficiency.”
Question 2: Can your AI solution integrate with my existing systems (e.g., CRM, policy admin systems)?
Seamless integration with your current platforms is essential to ensure efficiency, and this is a feature you need to get clarity on right up front.
- Good answer: “Yes, we have pre-built integrations with major insurance platforms like Applied Epic, or we offer APIs for custom integrations.”
- Red flag response: “There shouldn’t be any problem. You will need to make some modifications to your systems, but we have staff that can help you with that.” Vague promises without clear examples of previous integrations in the insurance space should raise red flags.
Question 3: How do you handle data privacy and security, especially with sensitive customer information?
Given the personal nature of customer information, data privacy is a critical concern for insurance brokerages looking to integrate AI.
- Good answer: “We use encryption, data anonymization, and comply with regulations like HIPAA and GDPR to ensure sensitive data stays secure.”
- Red flag response: “We make data security a top priority and adhere to all the best practices in every industry.” Without mentioning specific privacy protocols or exhibiting any understanding about compliance with regulations, this sort of answer may be glossing over an actual lack of knowledge/capability.
Question 4: Does your AI model retain any of my data for training or other purposes?
Data retention policies can be a dealbreaker, especially in an industry that must operate with high levels of concern over customer privacy.
- Good answer: “No, our models do not retain customer-specific data unless agreed upon with the client. Any data used for training is anonymized and follows strict governance policies.”
- Red flag response: “Yes, we automatically retain data for future training,” without offering an opt-out or clearly defined governance around data use.
Question 5: How does your AI adapt to regulatory changes in the insurance industry?
Insurance is a heavily regulated industry, and your AI solution needs to keep up with changing laws and policies. This is the type of question that can really expose whether a potential vendor has a solution that is truly targeted to the insurance industry or is simply trying to make its application relevant to all industries.
- Good answer: “We track regulatory changes and update our AI accordingly, working closely with legal and insurance experts to ensure compliance.”
- Red flag response: “We can’t track the regulatory aspects of every industry in which we operate. It’s just not feasible. That’s up to the customer to manage.”
Remember, applications built for the insurance industry will address the needs of the insurance industry.
Question 6: What’s the typical implementation timeline for an insurance brokerage like mine?
Understanding how long it will take to get up and running is key both to managing expectations and predicting ROI.
- Good answer: “Implementation usually takes 4-6 weeks, broken into phases with predefined milestones. We’ve worked with agencies like yours before and can provide a detailed plan.”
- Red flag response: “It really depends on a lot of factors. We can give you a more detailed plan once we get started.” Any answer without offering concrete timelines or milestones based on prior experience should be a red flag.
Question 7: How does your solution improve over time?
Not all AI programs learn. But you want yours to! So, a prerequisite question to this one would be “Is your AI model static (non-learning) or dynamic (learning)?” Once you’ve established that you’re dealing with a model that will continuously learn and improve, it’s important to know how that happens.
- Good answer: “Our AI learns from new data and is regularly updated to improve accuracy in assessing risk, underwriting, and customer service.”
- Red flag response: “It works well right out of the box, and we’ve discovered that after the initial training there really is no need for updates. The program is robust to begin with, we haven’t seen the need to do much post-deployment.”
Question 8: What kind of ongoing support do you provide, and do I get a dedicated customer service representative or account manager?
Perhaps more than non-AI programs, the level of ongoing support is crucial to the long-term success of the AI solution.
- Good answer: “Yes, we provide 24/7 support, and you’ll have a dedicated account manager for any issues. We also offer regular check-ins to ensure the AI continues to meet your needs.”
- Red flag response: “We have a fully staffed general support line,” or other answers that indicate you will not be receiving personalized support.
Question 9: Can you share examples of measurable outcomes from your customers, particularly for insurance brokers?
Vendors should be able to back up their claims with real-world examples.
- Good answer: “Firm X reduced claims processing time by 30% after 6 months,” or “Policyholder churn dropped by 15% using our AI insights.”
- Red flag response: “We don’t share client information and we don’t track by industry, but we are happy to share details regarding overall outcomes.”
Given the nascency of many AI applications, this may be more of a “pinkish” flag rather than a red one. But, ideally, you would want a program with a proven record in the insurance arena.
Question 10: What type of ongoing costs should I expect (e.g., maintenance, retraining, support)?
Keep in mind, understanding the true cost of AI involves more than just the initial purchase.
- Good answer: “We charge a subscription fee that includes maintenance, retraining, and support. You can also upgrade to premium support if needed.”
- Red flag response: “There are no ongoing costs after implementation,” which could indicate that ongoing support is sparse – which might lead to unpleasant surprises.
By asking these critical questions, and evaluating the answers, you’ll be better equipped to accurately assess specific AI solutions and determine which one is the right fit for your business. The insurance industry is evolving rapidly, and AI has the potential to give brokers a significant edge, but only when they choose the right solution for their specific needs – and also avoiding potential pitfalls along the way.
