Financial services
Accelerate financial due diligence, modeling, and analysis with enterprise-grade AI built for compliance and security.
Q&A with Strib Walker, Head of Product
How Rogo evaluates frontier models for institutional finance
Industry:
Software
Company size:
Startup
Product:
Claude Platform
Location:
North America
50,000+ queries per day
across research, analysis, and artifact generation
35,000+ financial professionals
at major banks, investment firms, and advisory practices
Financial services
Accelerate financial due diligence, modeling, and analysis with enterprise-grade AI built for compliance and security.
Video caption
Financial services
Accelerate financial due diligence, modeling, and analysis with enterprise-grade AI built for compliance and security.
Rogo is an AI platform built specifically for finance. More than 35,000 financial professionals at major banks, investment firms, and advisory practices run over 50,000 queries a day on the platform. We spoke with Strib Walker, Head of Product at Rogo, about what makes financial AI different, how Rogo evaluates models, and where the work is headed.
Anthropic: What do most people get wrong about applying AI to finance?
Strib Walker, Rogo: Finance demands very bespoke tools, and the difference between "looks right" and "is right" is enormous. A lot of what makes something “right” in finance is highly specific to the context. It varies by specialty, firm, industry, team, and even individual senior preferences. General models may understand standard financial line items from a 10-K, but they often miss nuances like which ratios matter most, which peers are appropriate, or how information should be presented. Those details are critical to professionals and can change the interpretation entirely.
Where does Claude fit into what you're building?
Walker: What we do in plain terms is help institutions operationalize AI inside their firms. Claude Opus 4.7 and Sonnet 4.6 are integrated across the Rogo platform, where they power parts of the core research, analysis, and artifact-generation experience our customers rely on every day. Anthropic models are available through Felix, our agent orchestrator for end-to-end financial workflows. Felix takes on the kinds of multi-step assignments that define the work in finance: building decks, financial models, and documents from start to finish with the structure, formatting, and rigor the work demands. Claude is one of several frontier models Rogo draws on inside Felix.
What specifically drew you to Claude?
Walker: Two things. The first was long-context reasoning over the documents that finance lives in: data rooms, multi-year filing sets, transcript archives, internal research libraries. Reasoning effectively over a fully utilized context window is exactly what diligence and deep research demand.
The second, and the one that most directly tipped the decision, was artifact creation. In our most recent benchmarks, we focused specifically on artifact generation in PowerPoint and Excel, reflecting the primary formats used in real-world financial analysis and reporting. Claude showed strong improvements in PowerPoint generation.
Success for us had three components: measurable improvement on our internal finance benchmarks; production-grade quality on artifact generation as graded by our finance evaluators; and a deployment that met the security, privacy, and compliance bar our customers require, at the speed they expect us to move.
You have former bankers working alongside AI researchers. What does that collaboration actually look like?
Walker: To close that gap between looking right and actually right that I mentioned, we have an entire team of former bankers, investors, and research analysts embedded alongside our AI researchers and engineers. They work together on model evaluation, prompting, artifact pipelines, and workflow design, mapping how the work is actually done on the desk onto what frontier models like Claude can do.
"Claude's strengths in long-context reasoning and artifact generation are a part of what makes that possible."
Head of Product, Rogo
How has the work evolved since the early days?
Walker: Early on, the work was proving frontier models could do serious finance work at all. As the underlying models matured, our focus moved up the stack: orchestrating multiple models, building deeper firm-specific intelligence, and turning point capabilities into end-to-end agentic workflows. That trajectory is what led us to Felix, and to the deeper integration of Claude into the platform.
Can you walk us through a specific workflow in Felix?
Walker: A user would typically start with some guidance and raw inputs: financials, documents, a banker brief, or other data room materials. From there, research is conducted within the platform, and financial data is pulled and substantiated directly from the inputs. Supporting materials and backups are built alongside the core narrative. Then Rogo shells out the deck and starts populating it with the appropriate inputs. The output of this work is almost always a deck or a financial model, and we needed a system that could produce structured PowerPoint and Excel output at institutional quality, not a generic approximation of it.
How do you evaluate which model to use for which task?
Walker: We're a model-agnostic business. We run dedicated testing to evaluate each model's performance across core financial workflows, and we dynamically route tasks to the systems that perform best for a given job.
Our evaluation framework goes beyond measuring raw financial intelligence. We place a strong emphasis on artifact creation, which is central to how financial work is actually consumed. Our internal testing assesses not only correctness but also the structure, clarity, and usability of generated outputs.
What can your customers do now that wasn’t possible before?
Walker: This is really about enabling workflows that didn’t exist before, not just speeding things up. Users can now create and iterate on full outputs, decks, financial models, much earlier in the process. They can explore multiple directions in parallel without the usual time constraints. That changes how work gets scoped and executed. It opens up a more iterative and exploratory way of working. Users often go from skepticism to excitement once they see something usable emerge, and that shift tends to stick once they validate the results.
If you were advising another financial institution just getting started with AI, what would you tell them?
Walker: A big part of getting started with AI is operationalizing it effectively. Having strong underlying intelligence is important, but making it usable in real workflows takes time and effort. That means building processes, integrations, and validation layers around it.
What's next for Rogo?
Walker: We're extending Felix and the broader platform deeper into the work financial institutions actually do: not just material creation, but the full arc of a deal process, from origination through execution. Claude's strengths in long-context reasoning and artifact generation are a part of what makes that possible, and we expect the workflows we take on to get longer and more complex.
Choosing the right Claude model
Learn when to use Haiku, Sonnet, or Opus to get better results and stay inside your rate limit. A practical guide to picking the right Claude model.
Choosing the right Claude model
Learn when to use Haiku, Sonnet, or Opus to get better results and stay inside your rate limit. A practical guide to picking the right Claude model.
Video caption
Choosing the right Claude model
Learn when to use Haiku, Sonnet, or Opus to get better results and stay inside your rate limit. A practical guide to picking the right Claude model.
"We needed a system that could produce structured PowerPoint and Excel output at institutional quality, not a generic approximation of it."
Head of Product, Rogo
Video caption