005 AI for Operators

Rahul has supercharged his work with AI. In this episode, he tell us how he’s strung together systems to get the best out of them, in his work and personal life, including:

• Data analysis - but accurate: Rahul shows how he feeds raw Google-Forms exports into Gemini, then uses App Scripts to derive summary stats auto-generated charts, and more in minutes (not days)

• “Vibe-coding” for operators: You don’t need to be an engineer: grab AI-written code, spot obvious spreadsheet-logic errors, ask the model to fix them, and rerun. Iterative prompt-debugging is the new coding.

• Deterministic vs. agentic AI. Rahul frames a practical litmus test: if identical inputs should always yield identical outputs (e.g., payroll reports, recurring KPIs), push for rule-based automations; reserve LLM “agentic” reasoning for idea-generation and edge-case discovery.

• AI as your ruthless editor. Before publishing insights, he asks ChatGPT to poke holes, surface missing angles, and then rewrite takeaways in Axios-style prose.

• Building a custom board game: Using a single “super-designer” prompt, he built game mechanics, a rulebook, custom cards, and more - helpful if you get board on a rainy day.

• Bring back Clippy! Rahul’s dream: Gemini or ChatGPT living inside Google Workspace, proactively executing your recurring processes without being asked.

What It Does

Scribe is a workflow-documentation platform that records any on-screen process and turns it into a polished, step-by-step guide almost instantly. Start the browser extension or desktop recorder, perform the task once, and Scribe captures every click and key-press, layering explanatory text and spotlighted screenshots. You can tweak wording, blur sensitive data, apply brand colors, and share the result via link, embed, PDF or HTML export.

Why Ops Leaders Should Care

If you want your team to scale, good processes are essential - plus, they actually need to be followed. Scribe makes creating SOPs and playbooks quick and easy, which means your team will spend less time on training and you’ll get better process adherence. Good documentation will also help your company incorporate AI faster: it’s a lot easier to build a workflow or an agent on an established process.

Key Features (Pros & Cons)

Pros

• Robust free plan: Sign up (without a credit card) and try it for free - helps to derisk it.

• One-click capture: Auto-builds guides with annotated screenshots and text for any web or desktop workflow, eliminating manual screenshots and writing.

• Rich editing & branding: Reorder steps, rewrite captions, blur or redact data and apply company branding for external sharing.

• Flexible sharing: Distribute via link, embed code, or PDF/HTML export so you can insert instructions exactly where teams work.

Cons

• Paywalled power features: Desktop recording, redactions, and advanced exports live behind the Pro paywall; the free tier is browser-only and limited to basic PDF export.

• No video: A seemingly glaring gap - no ability to record and display a video as part of the SOP.

An Operator’s Perspective

I tested this out on my own workflows and the process was very easy. You choose what tab you want to capture, then manually complete the workflow as usual. As you move through the steps, Scribe captures all your movements (including screenshots) and provides a running commentary. The workflows I tried required some editing after the fact - both to change the titles and commentary and to remove duplicative steps, but overall the process took only a couple of minutes.

One cautionary note: it’s only possible to redact information if you have a Pro subscription, so don’t share anything you create using the Free version if you work with sensitive data (even something like your calendar).

One other interesting feature: if you have the Chrome Extension open while you’re on other pages, Scribe will suggest workflows related to the tool that you’re using. For example, I’m writing this on Beehiiv and Scribe is suggesting a couple of public workflows that have been recorded related to Beehiiv. This would be more powerful for internal teams - there’s no need to remember a workflow if it’s going to be surfaced for you whenever you’re on that page!

You can see an example output here.

Other Options

Bottom Line

Today: if you want to scale fast, you’re going to need to document processes. Scribe can help you do that quickly and easily, as long as you don’t need to record video.

Tomorrow: as AI takes over more workflows, having good documentation will allow you to automate those processes more quickly.

• Replit announced that it has surpassed $100M ARR: They went from 0 to $10M in ARR in 8 years, and then 10x’d in a year - a reminder of the crazy new benchmarks for impressive growth in the AI era.

• AI can’t run a store…yet: An update on the capabilities of Claude. Articles like this always make me suspicious - the mistakes are so hilariously bad (and fixable!) that it seems like part of a PR campaign to ward off regulation, more than an honest update on AI’s current capabilities.

• Cybersecurity isn’t keeping up with AI: According to Accenture, 77% of organizations lack the data and security practices to guard their AI infrastructure. If you’re looking for an area to invest in that will have your CIO or CTO nodding their head, this could be it.

• Browserbase launched Director.ai: This lets anyone automate the web (think: setting up recurring web scraping jobs, etc.) - useful if you’re not a developer.

• Develop a visible AI strategy, like, now: Companies with a defined strategy are twice as likely to see growth as a result of it, according to a new Thomson Reuters survey.

• A new AI-driven ERP is going up against NetSuite: Campfire, fresh off a $35M raise led by Accel, is one of a crop of upstart ERP systems (including Rillet, Series, Tailor, and others) looking to unseat legacy players like NetSuite. If your CFO or CEO are fed up with NetSuite, it might be worth exploring these systems (you’ll probably pay less and get a better UI) - just be ready for a messy migration!

• Zuck goes on an AI hiring spree: If nothing else, the Meta founder knows how to take bold bets. In addition to the non-acquisition-acquisition of Scale AI founder Alexander Wang, former GitHub CEO Nat Friedman as well as several leading AI researchers from OpenAI and other companies are joining Meta (with massive comp packages) to form their new Meta Superintelligence Labs. He’s not messing around.

• New productivity bundle incoming? Hot on the heels of the acquisition of AI tools company Coda, Grammarly announced the acquisition of fan favorite email client Superhuman, paving the way for a likely future announcement of an AI-first productivity bundle.

Thanks for reading,

Tom Guthrie

If you liked this edition of AI for Operators, share it with another operator you know!