Josh Allen, CEO of TRDA (a $30M SaaS company), spent two months building a custom dashboard using AI coding tools that consolidates all his business data into one place. He walks through the dashboard, the tools he built on top of it, and how any founder can replicate the approach in a single weekend with zero coding experience.
The Revenue Tracker Supreme 10,000
Josh started by pulling 2.5 years of historical invoice and revenue data from spreadsheets and feeding it into Claude Code to build a unified dashboard. The tool standardizes data from multiple sources and formats into a single view.
- Revenue consolidation — aligns payments that arrive at different times across months into an accurate, real-time picture
- Key metrics at a glance — NRR, GRR, goals, trends, concentration risk, and alerts all on one screen
- Multiple ARR views — toggle between recognized ARR, 3-month rolling, and T12 average with built-in explainers for each
Executive Memo & Exit Readiness
Even without plans to sell, Josh used AI to generate a board-ready executive memo by prompting it to act as a 20-year M&A veteran preparing the company for a potential sale.
- Triple pattern confirmation — T12 average < 3-month rolling < contracted ARR, showing accelerating growth across every timeframe
- Concentration risk analysis — surfaced that one client had grown 3% over a year but declined 21% in the last 3 months
- A real business adviser called the AI-generated memo "phenomenal" and asked to share it with his VC partners
Something like this makes me feel a little bit more reassurance that me continuing on my own path is the best path. It's a healthy business. It looks great.
Automating Real Work: Bonuses & Invoicing
Team Bonus Plan
Using the historical data already in the system, Josh built a tiered bonus plan in a day or two — something that had been on his to-do list for years.
- Eligibility and multipliers based on seniority
- Previews of payout at each tier, driving team engagement as a remote company
- Transparent goals everyone can track together
One-Click Invoicing
What used to involve checking four or five systems and running invoices one at a time now takes about 30 seconds: upload the file, hit run, done.
This is a 4-year process that has now been taken to like 30 seconds.
How to Build Your Own Version
- Get Claude — subscribe and install the CLI on your machine
- Start with existing data — pick a repetitive task you already have spreadsheets for
- Seed it with history — pull 2+ years of reports, drop them in a project folder, and ask AI to standardize them
- Iterate from a simple dashboard — start with a table view, then layer on visualizations and forecasting
- Always verify — have the AI check its own work, and check it yourself
Multi-Model Consensus for Reliability
For critical calculations, Josh runs six AI models simultaneously (three from Claude, three from OpenAI) and requires them all to agree before accepting a result.
- Models argue until consensus is reached — one session ran for two hours of "deliberation"
- A 5-of-6 threshold can be set for less critical decisions
- Result: zero problems after consensus is reached