Please note: This conversation was based on peer-to-peer engagement and shared experiences, not formal research. The insights shared reflect personal perspectives from within the community.
AI is quickly becoming one of the fastest-growing line items in insurance technology budgets, and the old playbooks for managing cloud and infrastructure spend do not always apply. As usage increases, costs can shift quickly and are often harder to track, especially when a single automated task can trigger many model interactions.
Recently, the Technology Business Management (TBM) Council hosted an interactive coffee chat specifically for technology, finance, and business leaders within the insurance sector to exchange perspectives, ask questions, and learn from peers navigating similar challenges. Event hosts Nathan Batchelder (Director, Tech Portfolio & Business Management with Liberty Mutual) and Dave Roush (Manager, Global Technology Business Management with QBE Insurance) led an open floor where attendees from various organizations shared valuable perspectives on this topic.
Whether still defining where AI fits into their cost models or working through more detailed cost tracking, the community came together to decode the real-world operational challenges of AI execution.
Key Insights from the Discussion
The conversation sparked an incredible exchange of operational strategies, architectural ideas, and budgeting hurdles unique to the current state of artificial intelligence deployment:
-
Widespread Daily Adoption: Polling during the chat revealed that over 80% of attendees leverage generative AI tools in their daily activities, ranging from qualitative tasks like executive summaries to quantitative data analysis.
-
AI for TBM Automation: Practitioners are actively using AI to automate historically tedious tasks, such as mapping complex vendor line items, operational data, and job titles directly into the standard TBM taxonomy. The Fluidity of Billing Models: A major shared pain point is the shift from stable, seat-based SaaS pricing to volatile, usage-based token and consumption metrics, complicating traditional corporate forecasting cycles. Re-Architecting for Multi-Model Efficiency: Organizations are exploring advanced architectural frameworks, using lower-cost foundational models to filter basic queries and escalating complex requests to heavyweight frontier models only when necessary.
-
Bridging the Cost-to-Value Gap: While tracking immediate consumptive cloud infrastructure costs is maturing, connecting those operational expenses directly to realized business capabilities remains a primary objective.
-
Polling metrics confirmed that 0% of the participants felt they had an entirely finalized, static data and AI cost model, providing validation that the entire sector is collectively navigating these fluid commercial parameters.
Detailed Discussion Summary
AI for Technology Finance & TBM
The first major focus of the session explored how technology finance teams use AI tools to optimize their own internal workflows. Rather than relying strictly on manual efforts, standard practice is rapidly incorporating localized versions of internal GPT instances and generative text pilots to execute massive data categorization efforts.
For instance, one practitioner shared a case where thousands of disparate vendor products and employee roles needed to be mapped to specialized framework categories. Executing this manually threatened to bog down operations for days. By loading the static documentation into an internal enterprise GPT program, the system outputted an accurate baseline categorization in a fraction of a day. Crucially, the AI did not just execute the mapping; it provided logical drivers and reasoning for each decision. This transparency dramatically reduced internal debate among cyber and infrastructure experts during validation reviews.
Other participants confirmed similar multi-million-row taxonomy mapping initiatives have shifted from tedious multi-month burdens into rapid, high-productivity automated workflows. Beyond heavy numbers, everyday productivity callouts included using tools to distill massive, slide-heavy executive decks into concise summaries for corporate leadership.
Managing the Fluidity of Token Invoicing & Budgeting
The discussion shifted toward corporate governance and the macro challenges of preparing multi-year technology budgets. A collective frustration emerged regarding how quickly model providers modify their billing frameworks. Moving from predictable, seat-based subscriptions to tokenized, volume-dependent pricing creates massive budgeting visibility issues. When vendors change infrastructure parameters overnight, providing clear forecasting guidance to business units becomes an immense challenge. Polling metrics confirmed that 0% of the participants felt they had an entirely finalized, static data and AI cost model, providing validation that the entire sector is collectively navigating these fluid commercial parameters.
Combining FinOps Tactics with Long-Term TBM Total Cost of Ownership (TCO)
To tackle this dynamic environment, practitioners explained how they combine real-time FinOps playbooks with long-term TBM principles. FinOps frameworks are deployed transactionally to provide granular, day-to-day cost attribution and anomaly alerting on incoming public cloud billing data.
Once this data is captured at the transactional level, it is categorized using the standard TBM framework to establish an asset lifecycle view. Leaders noted that looking at cloud infrastructure bills in isolation hides the true cost of enterprise AI. A holistic TBM strategy incorporates human labor costs, specialized software licenses, and on-premises infrastructure alongside direct model processing bills to create a true Total Cost of Ownership (TCO) dashboard. These robust TCO views are proving to be powerful tools for driving corporate executives back to the TBM table, as business leaders demand granular visibility into the multi-million dollar investments supporting advanced automation.
Advanced Optimization & Architectural Cost Control
The final segment of the chat focused on direct engineering tactics used to maximize efficiency. Experienced infrastructure practitioners pointed out that computing utilization routinely experiences dramatic drop-offs during off-peak hours. To avoid paying for idle capacity, engineering teams are partnering with corporate finance to establish batch-job scheduling. Non-time-sensitive compute workloads, such as processing deep archives of medical records or background system transformations, are shifted to nighttime slots when providers offer steep processing discounts.
Furthermore, instead of routing every prompt directly to top-tier models, organizations are designing multi-agent environments. Lower-tier, highly cost-effective models are placed on the front lines to evaluate incoming queries. A secondary validation agent automatically evaluates the response quality, escalating to premium frontier models only when the lower-tier output is flagged as insufficient. This multi-tiered strategy enables businesses to manage massive volumes of transactions while keeping compute costs strictly aligned with complexity.
Keep the Conversation Going!
The Insurance Industry Strategy Community brings together technology, finance, and business leaders to advance how Technology Business Management (TBM) is applied across the insurance sector. Our goal is to make technology investments more transparent, effective, and collaborative—driving innovation and greater business value across the industry.
In 2025, the team produced the How Insurance Companies are Leveraging TBM eBook and launched the TBM in Insurance Survey, which measured adoption, maturity, and the impact of TBM practices on technology and business outcomes. Together, these efforts strengthen industry benchmarks and guide the next phase of TBM growth in insurance.
Join us to contribute to these and other exciting areas, expand your professional connections, and play a vital role in advancing TBM and technology value management across the Insurance industry.
Next Steps: Engage and Learn More!
To continue your journey with Technology Business Management, we invite you to explore our website, log into TBM Connect here, dive deeper into the TBM framework, and learn the fundamentals of TBM modeling. Consider enrolling in a TBM course, joining the TBM Council to connect with a global community of peers, or attending the next TBM Conference to engage with thought leaders driving innovation at the intersection of technology and business. You can also explore our network of partners for additional resources and expertise, and subscribe to our newsletter (login required) to stay informed about the latest insights, events, and tools in the world of TBM.
Ready to connect, learn, and grow? Join one of our TBM Strategy Communities on TBM Connect to participate in open forums, attend upcoming Coffee Chats, and contribute to our growing body of shared knowledge!
What is a TBM Council Coffee Chat?
TBM Council Coffee Chats are informal, peer-to-peer engagements that create a relaxed space for community members to share real-world insights, ask questions, and learn from one another. It’s all about connecting and growing together. These are not webinars or open forums; there are no slides or demos, just real conversations led by community members. The chats occur every 4 to 6 weeks for about 30 to 45 minutes. They are hosted by fellow TBM community members who are passionate about sharing their experiences and lessons learned.
