Last Thursday, I ran a webinar for AI Finance Club on Claude's advanced features. I showed several features and use cases. One topic got real audience engagement: the interactive dashboard and commentary generation app.

I do these presentations regularly and seldom see this level of response. Honestly, I was amazed by how the app turned out and by the numerous use cases it opens up.

So this week, we're breaking down exactly how to build these tools. The free section covers what I built, why only Claude can do this, the two-window process, and how sharing works.

The paid section gives you the exact prompt I used, the refinement process that makes it CFO-grade rather than generic, and three additional use cases with full prompts: cash runway calculator, headcount planning tool, and pricing sensitivity analyzer.

Fractional CFO Corner – AI Finance Club. This month’s session dives into using AI for Financial Analysis. If you’re already a member, join me on January 29th for the live discussion.

Not a member yet? Now’s a great time to join the club and see how other CFOs are applying AI to real workflows.

For a behind-the-scenes look, join the free webinar How to Become an AI CFO.

An interactive HTML app that allows the user to upload an Excel file with quarterly data.

The app generates two outputs.

Tab 1 is a visual dashboard. Tab 2 is AI-generated CFO commentary.

Built in 30 minutes. Reusable every quarter. Same structure, different data.

You can see here how it works.

If you want to test it, use this link to access the app.

Note, it is built for a specific (fake) company with a specific report structure, so to better test it, use the same report that I used. Just download it and then upload to the app.

Why It Resonated

Three reasons, I think.

First, it's easy to build. I'll walk you through the process in a moment, but the short version: two Claude windows, three rounds of prompt refinement, one execution. No coding background required. Just clear thinking about what you want.

Second, it replicates YOUR exact process. Your dashboard structure. Your variance thresholds (>5% is material, not >10%). Your target ranges (EBITDA 18-20%, not "industry standard"). Your commentary logic (what you look at first, what triggers a red flag, how you separate signal from noise). This isn't a generic template. It's your financial lens, codified.

Third, and this matters: normally, a Claude artifact doesn't allow you to control who has access. Once it's published, everyone with the link can access. This creates data security concerns for these types of applications. But with this app, we're not sharing any data. Only the user who uploads their file can see their dashboard and commentary. This is very important and opens numerous other use cases.

Here's where this becomes more than automation. You're not just building tools for yourself. You're building tools for other departments with your financial logic embedded.

Sales team needs to model pricing scenarios. You could join every pricing call. Or you could build them an app with your margin requirements, your volume assumptions, your discount approval thresholds already built in. They run scenarios in their meetings. You're not in the room, but your expertise is.

HR planning headcount for next quarter. You could review every hiring plan. Or you could build them a calculator with your fully-loaded cost assumptions, your ramp timelines, your budget impact models. They explore options using your framework. They know immediately when they're hitting a threshold that requires your approval.

Marketing analyzing campaign ROI. You could sit in every campaign review. Or you could build them a dashboard with your attribution model, your payback period requirements, your efficiency targets. They evaluate campaigns using your standards.

You control the logic. They use the tool when they need it. You're not blocking their workflow. They're not guessing at your requirements. This is business partner support that actually scales.

Why build this in two separate Claude windows? Two reasons: token efficiency and output quality.

App generation burns a lot of tokens. These artifacts are complex. If you're designing and building in the same conversation, you're using tokens on back-and-forth that doesn't need to be there. Separate the design thinking from execution.

Also, cleaner output. No conversation clutter in the generation window. Just prompt, generate, done.

Here's the process:

Window 1: prompt design. Start a new Claude (or any other LLM) chat. Describe what you want to build in plain language: "I need a finance dashboard generation app where I upload a source file, and the app generates a dashboard according to my instructions. Help me write a prompt for this task." Ask the LLM to create the prompt for you. Then refine it. Add your specific metric definitions, variance thresholds, target ranges, color coding rules, error handling requirements, and commentary structure. Read and refine carefully, then save the final prompt.

Window 2: execution. Open a new Claude chat. Make sure Artifacts is enabled in Settings→ Capabilities. Paste your refined prompt. Let Claude generate the app. If your prompt is solid, you'll need minimal iteration here.

The first time takes 30 minutes. After that, you have a template. Adapt it for other use cases in 10-15 minutes.

What You Need

Claude Pro/Teams/Max subscription. The Artifacts feature is only available on Pro and higher plans. This is also required for the Claude-in-Claude API call that generates the commentary section.

Clear understanding of your own analysis process. What do you look at first? What's your materiality threshold? What target ranges do you use? What triggers a red flag? If you can't articulate your process, Claude can't codify it.

Structured data approach. Your data doesn't have to be perfect, but it needs consistency. Same column names quarter to quarter. Same tab structure. Consistency matters more than perfection.

Other Applications

This same two-window approach works for other finance use cases. Cash runway calculators where the sales and exec team can model burn scenarios without you. Pricing sensitivity analyzers for deal structures. Headcount planning tools for department heads. AR aging analysis with collection priority. Budget variance dashboards. SaaS metrics monitoring with your exact definitions. Marketing ROI analysis using your attribution model. Investor update generators.

That's the concept and the basic process.

But if you want to actually build one of these tools this week, you need the exact prompt I used and clear guidance on what must be specified to make it CFO-grade instead of generic.

That's what I'm sharing in the paid section.

Plus three complete use cases with full prompts: cash runway calculator, headcount planning tool, and pricing sensitivity analyzer.

When I present use cases in webinars, I never know what will resonate most. Sometimes it's the technical capability. Sometimes it's the time savings. This time, it was something different: the recognition that this is what business partnership actually looks like when you combine CFO expertise with the right technology.

I'm using the cash runway app right now with one of my clients. They're in a tight cash position and need frequent updates on runway, literally weekly. Before, that meant pulling data, rebuilding scenarios, running calculations, scheduling calls. Now, they upload their latest numbers, adjust assumptions themselves, and see immediately where they stand. I review their conclusions, add context, adjust strategy. My expertise scales. Their decision-making speeds up. Its a win-win!

If you build one of these tools this week, I'd love to hear about it. What use case did you pick? What surprised you about the process? Reply to this email or tag me on LinkedIn.

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina
AI-Powered CFO