Table of Contents
- Explainability Has to Reach the Customer
- Insurance Decisions Carry a Higher Burden
- Communications Cannot Be an Afterthought
- Traceability Starts Upstream
- Reason Codes Are Not Explanations
- Accuracy and Tone Both Matter
- Template Sprawl Creates Risk
- The Communication Record Matters
- Human Oversight Still Matters
- Consistency Builds Trust
In Brief
Explainable AI in insurance doesn’t end with the model — it ends with the message the policyholder receives. Insurers using AI to support claims, underwriting, and pricing decisions need communications infrastructure that connects upstream decision data to approved language, governed templates, and audit trails. Reason codes alone aren’t enough. The letter, email, or portal message a customer receives is both the accountability moment and the final record — and it needs to be accurate, consistent, and defensible across every channel.
Top 10 Webinar Insights on Explainable AI in Insurance
Sharon Malloch 
June 24th, 2026
Some webinar conversations stay high level. This one did not.
In our recent webinar, Closing the AI Explainability Gap in Insurance, Andy Young, founder of TreelinePress, led a practical conversation with Rao Tadepalli, an insurance industry expert and veteran CIO, and Emily Washington, Head of Products at MHC.
At its simplest, explainable AI means being able to understand, trace, and document how an AI-supported decision was made, what information influenced it, and how to communicate that decision to the policyholder in a way that is clear, accurate, and defensible.
After I introduced the speakers, I listened behind the scenes as the discussion moved past whether insurers will use AI and into the harder question now facing the industry: as regulatory expectations rise, what does it actually take to explain, control, and communicate AI-supported decisions when they affect policyholders?
That is where explainable AI in insurance becomes real. Customer communications may be the final step in the process, but they cannot be an afterthought. The resulting letter, notice, email, portal message, claim explanation, or policy document is where an AI-supported decision becomes visible to the policyholder — and where that decision needs to become clear, understandable, and defensible.
Here are 10 insights from the webinar that stood out to me.
In Brief
Explainable AI in insurance doesn’t end with the model — it ends with the message the policyholder receives. Insurers using AI to support claims, underwriting, and pricing decisions need communications infrastructure that connects upstream decision data to approved language, governed templates, and audit trails. Reason codes alone aren’t enough. The letter, email, or portal message a customer receives is both the accountability moment and the final record — and it needs to be accurate, consistent, and defensible across every channel.
Table of Contents
- Explainability Has to Reach the Customer
- Insurance Decisions Carry a Higher Burden
- Communications Cannot Be an Afterthought
- Traceability Starts Upstream
- Reason Codes Are Not Explanations
- Accuracy and Tone Both Matter
- Template Sprawl Creates Risk
- The Communication Record Matters
- Human Oversight Still Matters
- Consistency Builds Trust
01
Explainability has to reach the customer
AI explainability cannot stop with the model, the data, or the internal review process. It has to carry through to the letter, notice, email, portal message, or claim explanation the policyholder actually receives.
As Emily explained during the webinar, every customer communication becomes an accountability moment.
“Every letter, email, portal message that you are communicating to your customer… that’s the accountability moment.”
~ Emily Washington, Head of Products, MHC
If a policyholder challenges a claim decision or questions information in a policy document, the insurer needs to trace that communication back to its source. That means understanding what decision was made, what data supported it, what reason codes or outputs were produced, and how those inputs became the final message sent to the customer.
In other words, explainable AI in insurance is not complete until the policyholder can understand what happened
02
Insurance decisions carry a higher burden
Insurance is built on contracts, obligations, and promises. When AI supports decisions around claims, pricing, coverage, or renewals, insurers still need to explain what happened and why.
Rao made this point clearly. A policyholder buys coverage with the expectation that the insurer will be there when support is needed. That makes clarity especially important in moments like claims, pricing changes, underwriting decisions, renewals, and coverage determinations.
If a claim is denied, the policyholder needs to understand the basis for that decision. If a premium changes, the customer may need to understand why. If AI is involved anywhere in the process, the insurer still owns responsibility for the outcome.
Rao emphasized that regulators have long expected insurers to explain and defend decisions. AI does not remove that responsibility. In many cases, it makes the responsibility more complex.
Insurers cannot simply point to analytics or AI as the reason for a decision. They still need to explain how the decision was made, what information supported it, and how the decision was governed.
“You’ve got to be able to explain how these things are being done.”
~ Rao Tadepalli, insurance industry expert and veteran CIO
03
Communications cannot be an afterthought
Customer communications may be the final step in the process, but they cannot be treated as an afterthought. The resulting letter, notice, email, portal message, claim explanation, or policy document is where an AI-supported decision becomes visible to the policyholder.
That is where explainable AI in insurance becomes real. The communication needs to make the decision clear, understandable, and defensible — not just delivered.
This is why communications teams have a growing role in AI governance. They help turn upstream decision data, reason codes, approved language, workflows, and delivery records into something the customer can actually understand.
A clear communication depends on clear decision data. Insurers need to capture the right inputs, logic, reason codes, and supporting details before the communication is created.
Before a communication can explain a decision, the organization has to know what happened. That means capturing the right data, documenting the decision logic, preserving relevant inputs, and structuring outputs in a way the communications process can actually use.
Emily made this point clearly: communications need structured inputs. Free-text explanations or incomplete decision data can create downstream risk because communications teams still need to generate accurate, approved, consistent messages at scale.
That is especially important in insurance, where one communication may need to reflect policy details, state requirements, reason codes, required disclosures, and customer-specific data. If that information is incomplete or unclear, the communication may be harder to defend.
04
Traceability starts upstream
A clear communication depends on clear decision data. Insurers need to capture the right inputs, logic, reason codes, and supporting details before the communication is created.
Before a communication can explain a decision, the organization has to know what happened. That means capturing the right data, documenting the decision logic, preserving relevant inputs, and structuring outputs in a way the communications process can actually use.
Emily made this point clearly: communications need structured inputs. Free-text explanations or incomplete decision data can create downstream risk because communications teams still need to generate accurate, approved, consistent messages at scale.
That is especially important in insurance, where one communication may need to reflect policy details, state requirements, reason codes, required disclosures, and customer-specific data.
If that information is incomplete or unclear, the communication may be harder to defend.
05
Reason codes are not explanations
Reason codes may help internal teams understand a decision, but customers need plain-language explanations that connect the decision to their policy, claim, coverage, or next step.
AI-supported systems may generate reason codes, decision outputs, scores, classifications, or other structured results. Those outputs matter, but they are not automatically meaningful to the policyholder.
As Emily explained, communications need structured inputs. Insurers cannot simply take a free-text narrative or internal decision output and pass it along to the customer. Those outputs need to be mapped to approved language that reflects the reason codes, the customer’s situation, and the communication being sent.
A customer does not want to decode internal logic. If a claim is denied, they need to understand the basis for the decision, how it relates to their policy, what information was considered, and what options or next steps may be available.
That is where customer communication management becomes critical. The communications layer needs to turn upstream decision outputs into language that is approved, compliant, readable, and appropriate for the situation.
06
Accuracy and tone both matter
An explanation has to be accurate, but it also has to be appropriate for the situation. A claims denial is not the same as a renewal notice.
Emily described two responsibilities at the communications layer: the decision has to be communicated accurately, and the message has to be delivered with the right tone.
That matters because insurance communications often arrive at sensitive moments. When a customer is dealing with a claim, loss, denial, cancellation, or premium change, clarity and care both matter.
The goal is not to make a difficult decision sound better than it is. The goal is to explain the decision clearly, consistently, and responsibly so the customer understands what happened and what they can do next.
07
Template sprawl creates risk
When templates are scattered across teams, systems, products, regions, and channels, consistency gets harder. With AI-supported decisions, inconsistent communications can weaken trust and make decisions harder to defend.
Many insurers already manage hundreds or thousands of templates across different lines of business, products, regions, channels, and teams. When those templates are managed in fragmented systems or updated manually, explainability becomes harder to operationalize.
Emily described this challenge as template sprawl. Different teams may produce different communications for different purposes, using different processes and language. That may have been difficult before AI. With AI-supported decisions, it becomes even riskier.
If a letter says one thing, the portal says another, and the agent cannot see the same explanation the customer received, the organization creates confusion. That inconsistency can weaken trust and make the decision harder to defend.
A governed communications platform helps reduce that risk by centralizing content, templates, rules, approvals, and audit trails so insurers can manage consistency across channels.
08
The communication record matters
If a decision is questioned later, the insurer needs to know what was sent, which template was used, what data was included, who approved it, when it was delivered, and what happened next.
That level of traceability matters because the customer often only has the communication. If a customer disputes a decision, escalates a concern, or takes further action, the communication becomes the record of what was explained and how the process was handled.
The record needs to show not only what was generated, but what was approved, sent, delivered, and available if the decision is questioned later. It should connect the communication back to the data, decision logic, content, workflow, and delivery history behind it.
That same record also helps internal teams stay aligned. If a customer calls an agent, submits an inquiry, or asks for more information, the organization needs the explanation to match what the customer already received.
This is why communications cannot be an afterthought in AI governance. The final communication is part of the governance trail.
09
Human oversight still matters
AI can help accelerate communication workflows, but high-impact insurance communications still need governed content, approval controls, version history, and human review where it matters.
The webinar touched on an important point about AI-generated content and automation. AI can help accelerate content creation, template development, translation, and communication workflows. But that does not mean insurers should remove human oversight from high-impact communications.
Emily explained that MHC’s approach to AI Assist keeps the human in the loop. That is important because insurance communications often need to satisfy multiple requirements at once: regulatory accuracy, jurisdictional language, customer readability, brand standards, tone, and operational timing.
AI can be a powerful accelerator, but it should operate within a governed environment. Approved content still matters. Version control still matters. Auditability still matters. The right people still need to review the right communications.
10
Consistency builds trust
Customers may not know whether AI was involved, but they know when an explanation makes sense. The letter, portal, email, and agent response all need to tell the same clear story.
Most policyholders are not thinking about model governance, audit trails, or reason codes. They are thinking about their claim, their premium, their coverage, or their next step. If they receive a decision they do not understand, trust can erode quickly.
Emily put it simply: as consumers, we do not always know whether AI was involved. What we care about is whether the communication makes sense.
If a claim is denied, the customer wants to understand why. If they call an agent, they want the explanation to match what they received in the letter, email, or portal.
Explainability is not only about satisfying regulators. It is also about helping customers feel that the process was fair, understandable, and accountable.
The Takeaway: The Customer Sees the Message
Explainable AI in insurance cannot live only in the model, the data science team, or the governance committee. It has to reach the customer.
For insurers, that means customer communications are becoming a critical part of AI governance. Notices, letters, disclosures, emails, portal messages, and policy documents must be able to explain AI-supported decisions clearly, consistently, and defensibly.
That requires more than static templates or manual processes. It requires communications infrastructure that can connect upstream decision data to approved language, governed workflows, audit trails, and omnichannel delivery.
Because in the end, the customer does not see the model. They see the message.
Watch the full webinar, Closing the AI Explainability Gap in Insurance, to hear Andy Young, Rao Tadepalli, and Emily Washington discuss how insurers can bring greater visibility, consistency, and control to AI-supported decisions and the communications that follow.
Key Takeaways
- Explainability has to reach the customer. AI governance doesn’t end inside the model or the data team — it ends with the letter, email, or portal message the policyholder actually receives. Every customer communication is an accountability moment.
- Reason codes are not customer explanations. Internal decision outputs need to be mapped to approved, plain-language content that connects the decision to the policyholder’s specific policy, claim, or coverage situation.
- Traceability starts upstream. Before a communication can explain an AI-supported decision, the organization needs structured inputs: captured decision logic, reason codes, policy details, jurisdictional requirements, and customer-specific data — all in a form the communications layer can use.
- Template sprawl creates compliance risk. Inconsistent templates across teams, regions, and channels make AI-supported decisions harder to defend. A centralized, governed communications platform reduces that risk by maintaining consistent content, approvals, and audit trails.
- The communication record is part of the governance trail. Insurers need to know what was sent, which template was used, what data was included, who approved it, and when it was delivered — especially if a decision is later questioned or escalated.
- Human oversight still matters. AI can accelerate communication workflows, but high-impact insurance communications require governed content, version control, approval workflows, and human review where it counts.
- Consistency builds policyholder trust. Customers don’t audit AI models — they read their letters and emails. When the explanation makes sense and matches what the agent says, trust holds. When it doesn’t, it erodes fast.
Sharon Malloch
Sharon leads content marketing at MHC, overseeing strategies that fuel sales and demand generation. With more than a decade of experience in customer communications, Sharon brings deep insight into customer pain points, industry trends, and the critical role of solutions in managing regulatory communications. Before joining MHC, she honed her marketing expertise at Doxim, Messagepoint, and OpenText.
