I do not download invoices anymore. My folder and Claude do it.
The speaker demonstrates how he uses AI (Claude) integrated with his local folder system to automate invoice collection for his accountant, downloading 36 invoices in 5 minutes with minimal manual intervention. He explains his AI team methodology, work streams, and how persistent local knowledge bases enable AI to handle recurring tasks independently without constant re-explanation.
Summary
The speaker begins by discussing the pain of manually collecting invoices as a freelancer—a task that consumed several hours monthly across 36-126 invoices scattered across various subscription platforms and email systems. He showcases his solution: Claude AI integrated with ClickUp (project management), Dropbox (file storage), and email, accessing these through a local folder knowledge system that provides persistent context.
The workflow operates through a single prompt where the speaker provides: (1) a ClickUp link showing missing invoices needed by his accountant, (2) a Dropbox folder destination, and (3) reminders of available resources (email access, Chrome access for direct portal downloads). Claude's orchestrator agent (named Larry) then autonomously searches emails for invoice PDFs and navigates portal websites to download invoices directly when emails don't contain them.
A critical element is the 'work stream'—documentation files stored in the local folder that instruct AI how to execute recurring tasks. Rather than re-explaining the entire process each time, the speaker has AI write down procedures as work streams so future requests require only minimal prompts. He contrasts this with how most people repeatedly restart chats and re-explain tasks from scratch.
The speaker emphasizes that his system uses plain text files, API connections to ClickUp, and email/Chrome automation capabilities available in Claude Desktop. He shows the actual execution where Claude found 36 subtasks, searched through Gmail, discovered some invoices weren't in email, logged into various portals (LinkedIn, Bitly, Heptabase, Zapier, Todoist), and downloaded missing invoices. When blockers appeared (login required), he manually logged in and Claude resumed downloading. The entire process completed with 31 of 36 found automatically, then 5 more after manual login assistance.
He concludes by explaining why AI is more valuable than traditional automation: traditional automations are rigid and break when email subjects change, while AI understands context and adapts. He advocates for identifying annoying, recurring workflows first, then implementing AI solutions rather than forcing AI into existing processes. He promotes his myICOR methodology and scaffold folder as resources for others wanting to build similar AI team systems.
Key Insights
- The speaker previously had to manually collect 126 invoices from various platforms just before a holiday, and similar tasks occur every quarter with 36+ invoices, demonstrating how accumulated small subscription payments create significant administrative burden
- Claude was set to a specific goal within the prompt, which prevented it from constantly asking for permission and allowed it to work autonomously; this is critical for hands-off AI execution
- Work streams are instruction files stored in the local folder that allow AI to understand recurring procedures without re-explanation; this enables future requests to require only a single word like 'Find invoices' instead of full instructions each time
- AI automation is more valuable than traditional rule-based automation because it understands context and adapts when email subjects or processes change, whereas traditional automations break and require constant maintenance
- Claude completed the entire 36-invoice task in 5 minutes with minimal human interaction (only logging into 2-3 portals when automated access was blocked), then autonomously closed the tasks in ClickUp and proposed follow-up communication with the accountant
Topics
Transcript
[0:00] I just completed uploading 36 invoices for my accountant to process later. if you're a freelancer, self-employed business owner, you know that You need to do the same, right? Searching for the invoice for this specific tool that you used, logging in, downloading it, or forwarding the email that you had with the PDF and all these things, they compound and can consume a lot of time. in fact, those 36 invoices I uploaded today are not my record because just before I went to holiday, I had to upload 126 invoices [0:32] that I had to find from various systems. I'm talking here about different subscriptions that I had in different software tools, where I need the invoices…
Full transcript available for MurmurCast members
Sign Up to AccessMore from ICOR with Tom | AI Productivity
Claude finished the work. Then I lost it in the chat.
A creator demonstrates a comprehensive local-folder-based system for managing AI-generated deliverables asynchronously, featuring an interactive browser interface that tracks decisions, changes, and conversations without relying on external databases or web applications. The system enables collaborative editing, visual annotations, and automatic context preservation across multiple parallel AI projects.
Claude writes better weekly reports than I did. Here is proof.
The speaker demonstrates how AI-powered weekly reports can automatically synthesize journal entries, notes, and captured content into personalized insights without manual review. Two different implementations are shown—one dashboard-style and one essay-style—proving the same underlying system can be customized to individual preferences and consumption habits.
I stopped doing my weekly review. The AI does it now.
The speaker has replaced traditional weekly reviews and manual note-linking with an AI-powered system built on a local folder structure called a 'scaffold.' This approach eliminates the time-consuming review ritual while enabling the AI to automatically connect notes and surface meaningful patterns that humans would miss.
I Made Opus and Fable Grade Each Other. Opus Admitted It Lost.
The creator demonstrates real-world AI applications in their productivity business using Claude Fable and Opus, showing how Fable excels at comprehensive analysis of complex systems while comparing its performance to Opus on specific tasks. The key finding is that Fable's holistic understanding justifies its higher cost for business-critical audits and optimizations.
Our AI invented the numbers. We caught it.
The hosts discuss the importance of double-checking AI outputs and validating information, emphasizing that this is not a new problem but rather a longstanding practice required with any information source. They explore how AI tools can be customized to individual workflows and contexts, arguing that understanding foundational methodology matters more than specific tools.