Instead of relying on people to interpret data and connect tools, Ficells creates self-operating "financial cells" that continuously understand, analyze, and act on financial data.
From semantic data ingestion and variance analysis to forecasting, budgeting, simulation, reports, and presentations, Tess 2.5, Ficells' flagship product, handles end-to-end financial planning and analysis while monitoring business performance in real time.
For Ficells Labs, getting one thing right was non-negotiable: trust. But in financial workflows, trust doesn't come from answers alone; it comes from seeing the reasoning, exploring the data, and verifying the details.
Before Thesys Agent Builder, Tess 2.5 could generate answers, but turning those answers into a real product experience required significant engineering effort.
Every new type of output meant building custom UI from scratch.
At the same time, the user experience itself started to break down:
Variance analysis is where this gap became most apparent. In a variance analysis workflow, a user would ask:
"What's driving the increase in marketing spend this month?"
The agent could return something like:
"Marketing spend increased due to higher campaign activity this month."
While accurate, it still stopped at text.
For a finance team, that's only the starting point. Users needed to understand:
Following up with additional questions to piece together the full picture slowed down the analysis and made it harder to verify the answers quickly.
This slowed workflows and, more importantly, eroded user trust, resulting in drop-offs.
Tess 2.5 needed responses that users could interact with, explore, and verify within the product. Not delivering that depth in a finance tool wasn't an option.
"We wanted UI as the output. Normal chat feels like a wrapper around text. Building our own would take too long, and it is hard to get right. Thesys gave us an easy way to ship interactive responses fast."
~ Parag Pakhrani
Ficells chose Thesys because they wanted UI as the output, not just text responses. Traditional chat felt like a wrapper around text, and building a custom solution would have taken too long and been difficult to get right.
With Thesys Agent Builder, Tess 2.5 moved from plain text to structured, interactive outputs inside the product.
Powered by Generative UI, it now generates cards, charts, tables, lists, reports, and more that users can directly interact with. Users can scroll through results, explore the data, and even update inputs within the same flow.
The agent no longer feels like a chatbot; it's a full-fledged workspace.
This solved three problems at once:
With Thesys Agent Builder, they saw immediate improvements across both product experience and development speed.
When users can see the reasoning, explore the breakdown, and check the numbers themselves, they stop questioning the output. And in demos, the difference lands in seconds.
"Now users can get a report that looks and behaves like part of the product. They can interact with it, scroll, and directly update. The agent output feels like a workspace artifact."
~ Parag Pakhrani
As Tess 2.5 expands across more financial workflows, the team can support increasingly complex agent experiences without increasing frontend engineering overhead.
With Generative UI as the foundation, financial analysis is now continuous, explorable, and actionable inside the product.
Key areas this unlocks:
And as Tess 2.5 expands into more complex workflows, the engineering cost stays the same.
For Ficells, Generative UI has become the bridge between financial intelligence and a product experience users can trust and act on.
"If you want your AI to feel like software and not like a normal customer service type chatbot, Thesys helps you get there. It lets you ship interactive UI responses fast and keep iterating without rebuilding everything."
~ Parag Pakhrani