2024: The Year of the Copilot

When we started Kalepa 6 years ago, predictions about the impact of AI on insurance were stuck in two extremes: AI would have minimal impact given the complexities of a highly-regulated and nuanced industry, or that it would completely replace human underwriters and deliver 100% straight-thru processing of all risks.

We set out to chart a different path for insurers. That’s why one of our first decisions was to name our product Copilot.

In many ways 2024 has been the year of the Copilot. I often reference my conversation with an industry friend during the WSIA conference shortly after Microsoft Copilot’s Superbowl ad, when I was (jokingly) asked when we were sending them a cease-and-desist letter. 

This rapid emergence of AI Copilots for nearly any type of knowledge work continues to be a strong validation of our founding belief – that AI would ultimately deliver the most value supporting and enhancing expert humans. That includes people like Cassie Paul from Paragon Insurance, whose team was able to double the number of quotes and binders last year with the help of Kalepa’s Copilot.

I am still frequently asked if Kalepa’s and Microsoft’s Copilots are related (they are not), and I always use those opportunities to highlight how the concept of a copilot is not about a new chat interface or the re-emergence of Clippy. A copilot is best understood as a set of foundational product and design principles that deliver real, measurable value in the enterprise:

1. Adaptability: Providing users with precisely what they need at any given step in their workflow while taking care of tasks or tools that don’t require human judgement
2. Simplicity: Seamlessly taking advantage of all data, models, and tools in the background to provide a user with easy-to-understand, actionable insights 
3. Transparency: Clearly showing the provenance and evidence for insights to allow users to make thoughtful decisions, maintain a high bar for compliance and auditability, and collaborate
4. Learning: Carefully paying attention to what is most important to help the “pilot” succeed and continuously improving against that objective

As we roll into 2025, the natural evolution of AI copilots is setting up a “new paradigm” for AI software: AI agents. These agents can iterate through tasks to take on more and more actions, freeing underwriters up to focus on the parts of their job where their skillsets are truly needed.

AI luminary Dr. Andrew Ng outlined four design patterns for so-called “agentic workflows” earlier this year: reflection, tool use, planning, and multi-agent collaboration. They have interesting parallels to the design principles we helped establish for copilots:

1. Reflection: Agents that can analyze their own outputs and identify areas for improvement. That is, agents that can learn.
2. Tool Use: Agents that can access and utilize tools to enhance their capabilities. That is, agents that can adapt.
3. Planning: Agents that can think ahead, consider multiple options, and make informed decisions. That is, agents that can simplify complex workflows. 
4. Multi-Agent Collaboration: Multiple agents working together to solve complex problems. That is, agents that are designed for interacting with each other (and with humans, the ultimate agent!).

So even as the paradigm changes, the principles stay the same. AI in the enterprise should center on adapting, learning, simplifying, and providing clarity and transparency to combat hallucinations and promote collaboration.

These principles are critical for AI to deliver true value to insurers: significant lifts to quote rates, bind rates, and premium bound, and significant reductions in loss ratios and time spent by underwriters on each risk. 

Just as we have for the past 6 years, we stayed true to these principles during a tremendously successful 2024:

• We shipped new capabilities to seamlessly integrate portfolio-level and account-level underwriting, including the ability to review and automatically optimize existing books of business during key points in the insurer’s life cycle. Agentic planning in action.
• We launched a brand-new AI-powered triage engine that automatically reflects and learns, helping underwriters identify which submissions they’re most likely to bind and get quick declines out on the submissions that aren’t worth their time.
• We released SafeInit, Kalepa’s first open source contribution, designed for the development of more robust and resilient systems – such as the ones that power all of Kalepa’s infrastructure.
• We published the go-to guide for COOs on using AI to drive meaningful change in underwriting operations.
• We were recognized for the third year in a row by Fintech Global as one of the 100 most innovative insurance technology companies in the world

What’s Ahead in 2025

Most of the AI headlines for the past two years have been dominated by the foundational model companies: OpenAI, Google, Anthropic, Meta, etc.

I am a firm believer that 2025 will see the application layer - i.e., the products that real users utilize to improve their performance - become the true difference-maker when it comes to generating value from AI. 

Foundational models will start to plateau in quality, thanks to scaling laws on the data they are trained on. This means the spotlight will shift to the applications that seamlessly capture the improvements from different models and turn them into tangible and demonstrable enterprise value. Insurers that miss this transition will find themselves amidst the thousands of “POCs to nowhere”. 

At Kalepa, we will continue our unwavering focus on helping many of the leading insurers drive massive value from AI so they can continue to protect and support individuals and businesses across the world. We are grateful to our customers and partners for their trust and support. 

As we kick off the new year, we look forward to sharing exciting announcements about our newest AI agents, our much-anticipated integration workstation, and how we’re helping underwriters everywhere make decisions with confidence. 

Wish you a happy and prosperous 2025!

-Paul Monasterio, CEO and Co-Founder at Kalepa