This is Part 2 of our WEX series. In this blog, we explore how the company scaled self-service analytics by embedding AI—read Part 1 on their people-first approach.
You’ve got AI pressure from every angle: execs, customers, and competitors. But legacy analytics doesn’t just slow down development—it frustrates users and undermines the value your product is supposed to deliver. When analytics depend on engineering, product teams can’t move fast, users get stuck waiting, and innovation takes a back seat.
For WEX, that breaking point turned into a breakthrough. In our recent webinar, WEX Product Manager Zach Holm joined ThoughtSpot Product Marketing & GTM Leader Ivan Seow to share how they transformed a clunky, request-heavy reporting system into a fully embedded, AI-powered experience.
Whether you’re drowning in ad hoc requests, struggling to scale insights to customers, or just tired of half-baked BI integrations, this is how you build analytics that actually work for devs and users alike.
How WEX Went From Homegrown Chaos to Customer Delight
WEX Field Service Management (WEX FSM) serves 35,000+ contractors in HVAC, plumbing, and electrical services who spend an average of five hours daily on their platform. These field service professionals need insights on efficiency, growth opportunities, and operational gaps, but they want answers fast because they have businesses to run.
The team's breaking point came when they realized their homegrown PHP-based reports couldn't keep up with customer expectations. Even worse, users were hitting five-minute timeout windows without getting any data back.
Zach's team was drowning in hundreds of requests for single field additions or filters, creating what he called "this mountain of backlog."
📺 Hear WEX’s embedded analytics story firsthand—watch the webinar on demand.
How to Build the Foundation: Data Architecture for Embedded Analytics
Before any embedding could happen, WEX had to overhaul their data architecture. Their original reports were pulled directly from production databases, causing performance issues and limiting analytical capabilities.
At the 2025 Snowflake World Tour in NYC, ThoughtSpot CDAO Steffin Harris and WEX VP of Engineering Mohamed Battisha broke down how WEX’s team migrated to a Snowflake-powered data warehouse, layered on a semantic model, and connected everything to ThoughtSpot for blazing-fast, customer-facing analytics.
"The data is the foundation. You cannot just build this framework without the real data behind it," Battisha emphasized.
This architectural change was crucial for enabling the semantic layer that powers natural language queries. Without clean, well-modeled data, even the most sophisticated AI can't deliver accurate insights.
The WEX team prioritized:
Data warehouse migration: Moved from production database queries to an optimized analytical data store
Semantic layer implementation: Enabled natural language understanding of business concepts
Quality validation: Used ThoughtSpot worksheets to verify data accuracy before customer exposure
Performance optimization: Architectural changes enabled sub-three-second query responses
Want to ship self-service analytics? Read how WEX reimagined their health and benefits platform in Part 1.
How WEX Chose the Right Embedded Analytics Platform
When WEX evaluated embedded analytics platforms, they needed something that felt native to their product, not bolted on. Their previous attempt with a traditional BI vendor resulted in what Holm described as "a single iframe box" with developer tools that meant nothing to field service technicians.
ThoughtSpot Embedded's architecture allowed WEX to take individual UI components and customize them with their own branding, as Zach explained:
The team implemented a three-layer approach:
Maintaining familiar stock reports for business continuity
Adding interactive Liveboards with real insights and KPIs
Embedding Spotter for natural language queries
The impact:
Seamless integration: Components blend naturally with WEX FSM's existing interface
Progressive feature rollout: Hidden advanced features initially to avoid overwhelming users
Drill-anywhere capability: Users can explore data in any direction without predefined paths
AI-powered customization: Natural language queries replace traditional drag-and-drop report building
This flexible approach delivered enterprise-grade analytics while maintaining complete control over user experience.
📖 Want your own comprehensive playbook for embedded analytics? Download your copy of the Buyer’s Guide.
Spotter: Transforming How Field Service Teams Ask Questions
The most dramatic adoption came from Spotter, ThoughtSpot's AI agent that WEX rebranded as "AssistIQ." Rather than forcing users to learn complex BI tools, Spotter enables conversational analytics where users can ask questions with natural language.
The AI agent solves the cold start problem by suggesting relevant questions based on available data and user context. When users ask follow-up questions, Spotter retains conversation history and context, creating a natural dialogue with data rather than starting from scratch each time.
"Our view is that Spotter and AI, these capabilities where you can have a conversation, that is gonna be the future of custom reports," Holm explained during the webinar:
The Impact of ThoughtSpot Embedded:
30x faster report delivery: From five-minute timeouts to under three seconds for critical reports
65% AI adoption rate: Within 90 days of rollout, most user interactions involved AI-powered queries
Three-month implementation: From data foundation to production-ready embedded analytics
Eliminated backlog bottlenecks: Engineering resources freed from maintenance to focus on core innovation
Transparency and trust: Search tokens show exactly how questions are interpreted and executed
Continuous improvement: Built-in thumbs up/down feedback loop trains the model daily
Contextual suggestions: AI recommends relevant follow-up questions based on current data exploration
One user told the team, "It's weird. I never expected to use AI within WEX FSM," but now uses it daily to run reports they never had access to before. This shift from skepticism to daily usage demonstrates how intuitive AI can drive adoption among non-technical users.
Fast Feedback and Real Change: How WEX Managed Transformation at Speed
WEX’s success wasn’t just about choosing the right technology—it was about reinventing how they worked. With just three to four engineers and product managers, they adopted a crawl-walk-run framework to introduce embedded AI capabilities at speed, without sacrificing trust or usability.
But speed alone isn’t transformation. To help with change management, WEX adopted a crawl-walk-run framework to guide teams and users through AI adoption:
Crawl: Focused on identifying key problems and building trust through validated data and clear communication.
Walk: Introduced AI augmentation, where users stayed in control while AI helped accelerate their workflows.
Run: Full automation for the right use cases—once the foundation, comfort, and confidence are in place.
The team used bi-weekly delivery cycles and deployed changes twice daily during development, allowing them to involve customers early, gather continuous feedback, and iterate rapidly.
Customers participated in the design process from the worksheet stage onward, helping to ensure the first analytics experiences were both high-quality and immediately useful, as Zach explained:
To build that trust, the team used ThoughtSpot's worksheets to validate data quality before exposing analytics to customers—ensuring credibility from the very first insight.
“People will believe you if you have tangible results,” said WEX VP of Engineering Mohamed Battisha. “You need that happy medium to drive transformation.”
Quick wins that made it work:
Two-week delivery cycles for quick pivots and fast learning
Twice-daily releases to accelerate experimentation
Human-in-the-loop by design, so users stayed in control
Gradual rollout of features, based on comfort and trust
Time saved, so 10-hour manual days streamlined to AI-assisted workflows instead
Early customer involvement: Users participated in the design process from the get-go
Small team efficiency: Three to four people accomplished what larger teams struggle with
It's Time to Rethink Embedded Analytics
WEX's journey demonstrates that moving from legacy, homegrown analytics to embedded, AI-powered solutions isn't just about technology. It's about freeing you and your team to innovate, delighting users with intuitive experiences, and driving real business results.
The most powerful insight from their transformation? Small, focused teams with the right tools and processes can accomplish what large, resource-heavy initiatives often struggle with.
Ready to eliminate your backlog with AI? Start your free trial of ThoughtSpot today.
