Get AI summaries of any video or article — Sign up free
How AI Creates SOPs in Minutes (ft. Hayden Miyamoto) thumbnail

How AI Creates SOPs in Minutes (ft. Hayden Miyamoto)

Tiago Forte·
5 min read

Based on Tiago Forte's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

SOPs scale operations only when knowledge is converted into a structured framework with multiple levels, not a single static document.

Briefing

AI can turn a company’s “tribal knowledge” into structured SOPs fast—by using a master prompt that encodes business identity, process definitions, and output rules, then generating a full, multi-level workflow with ownership (RACI) and KPIs. The practical payoff is scalability: when key people are out sick or on vacation, documented procedures let others step in, train new hires, and keep operations consistent.

The workflow starts with defining SOPs as more than a single document. SOPs are framed in three levels: a high-level end-to-end view (even spanning months), mid-level SOPs that cover one department or segment, and tactical guides for one person that spell out the most granular steps. For Acquir—an acquisition-focused company—the example “core process” runs from sale through delivery success for its Accelerator Plus product, beginning after an invoice is sent and ending when an LOI is signed and phase two is paid.

Before AI, the process was mapped manually on a whiteboard and then recreated in Excalidraw. That effort took about six months and still required ongoing improvement. The AI-driven approach begins with a multimodal “screenshot” moment: after capturing the Excalidraw diagram, the creator asks AI what it is and is surprised by how well it understands the diagram’s meaning. From there, the system shifts to prompt-driven generation.

A “master prompt” acts like a business operating system for the model. It injects style preferences (concise wording, ask clarifying questions one at a time, answer based on internal knowledge, wait for confirmation) and business identity (value proposition, brand voice, mission, core values). With the project files providing context about Accelerator Plus, AI generates a visual diagram of the core process with clear start and end points, step-by-step handoffs, activities, and success criteria. It also surfaces decision logic—such as how to prioritize when multiple deals could match multiple AEs—then asks for confirmation.

To make the SOP actionable, AI outputs RACI-style responsibility structure: Responsible (who does the work), Accountable (the single person who owns the KPI outcome), Consulted (two-way input needed), and Informed (one-way notification). The same process generation can branch into parallel tracks—for example, on-market versus off-market deal paths—then converge where the workflow meets again. The result includes key roles, key performance metrics, and an HTML artifact that packages the workflow into a usable deliverable.

From the core process map, AI can generate separate “guide” artifacts for specific functions, such as an onboarding call script or investment thesis development. The creator emphasizes a review loop: the generated SOP may be overly detailed, so a responsible party (e.g., Nathan) checks what’s missing or unnecessary. When AI produces too much, the fix is iterative—remove sections that don’t fit the company’s actual needs.

Finally, the transcript highlights how to operationalize these outputs beyond chat: using AI to generate a lightweight React app that displays SOPs via clickable pages, then exporting the code to be hosted elsewhere for non-technical team members. The broader claim is time-sensitive: setting up an AI “co-pilot” for SOP creation is work upfront, but businesses that do it early can extract significant value before the approach becomes standard.

Cornell Notes

The core idea is that SOPs become scalable only when they’re systematized—turned from “tribal knowledge” into repeatable, multi-level documentation. A master prompt can encode business identity, process definitions, and strict output rules so an LLM generates an end-to-end core process (with decision points, KPIs, and handoffs) and then produces tactical guides for specific functions like onboarding. The workflow is strengthened with RACI ownership so every task has clear Responsible, Accountable, Consulted, and Informed roles. Iteration matters: generated SOPs are reviewed by the responsible owner and trimmed or corrected to match reality. Packaging outputs into a shareable app (e.g., a React-based interface) makes SOPs usable for teams, not just readable in chat.

Why treat SOPs as a system rather than a single document?

SOPs are framed in three levels: (1) an end-to-end explanation of the whole workflow, potentially spanning months; (2) mid-level SOPs focused on one department or segment; and (3) tactical guides for one person with the most granular steps. This structure prevents knowledge from staying trapped in individuals’ heads and supports scaling, onboarding, and coverage when key people are absent.

How does the master prompt make AI-generated SOPs match a specific company?

The master prompt injects business identity and output preferences into every request. It includes style rules like concise wording, asking clarifying questions one at a time, answering based on the model’s knowledge, and waiting for confirmation. It also includes company mission, core values, and brand voice. Project files provide product context (e.g., Accelerator Plus), letting AI generate a process diagram that starts and ends at the correct business milestones.

What does RACI add to an SOP beyond step-by-step instructions?

RACI clarifies ownership and accountability. Responsible is the party doing the work (often one person). Accountable is the single person who owns the KPI outcome—having multiple accountable owners risks nobody being accountable. Consulted identifies two-way input needed from others, while Informed covers one-way notification. The transcript emphasizes that each task should map to these roles.

How does AI handle branching workflows like on-market vs off-market deal paths?

After generating the core process, AI can create parallel tracks for different paths (e.g., on-market marketplace listings versus off-market outreach). It walks through steps (A1, A2, A3, A4) with decision points and KPIs for each track, then shows where the tracks converge. This makes the SOP reflect real operational divergence and re-alignment.

What’s the practical review loop to keep AI outputs from becoming “wrong but detailed”?

Generated guides can be overly exhaustive compared with actual company practice. The workflow is to review artifacts with the responsible owner (e.g., Nathan), identify gaps or unnecessary sections, and then iterate—such as removing sections that don’t fit the company’s needs. The goal is to combine AI’s completeness with human judgment about what’s truly required.

Why package SOPs into an app (like a React interface) instead of leaving them in chat?

Chat outputs are harder for teams to use consistently. A lightweight app can present SOPs as clickable pages, making the system user-friendly for non-technical staff. The transcript notes that Claude isn’t ideal for full hosting, so code can be exported and hosted in other tools (e.g., Vercel/Netlify-style hosting), turning drafts into a shareable internal resource.

Review Questions

  1. How do the three SOP levels (end-to-end, department segment, and tactical guide) change what a team member should expect to find in the documentation?
  2. What is the difference between Responsible and Accountable in RACI, and why does the transcript insist Accountable should be only one person?
  3. What specific master-prompt behaviors (e.g., question timing, confirmation, confidence scoring) are intended to reduce hallucinations and overwhelm?

Key Points

  1. 1

    SOPs scale operations only when knowledge is converted into a structured framework with multiple levels, not a single static document.

  2. 2

    A master prompt can encode business identity (mission, values, brand voice) and process definitions so AI-generated SOPs fit the company’s reality.

  3. 3

    Generating a core process map with clear start/end points, handoffs, success criteria, and decision logic makes downstream SOPs more consistent.

  4. 4

    RACI ownership turns procedures into accountability systems by assigning Responsible, Accountable, Consulted, and Informed roles for each task.

  5. 5

    AI outputs still require human review to remove unnecessary detail and close gaps between “what’s currently done” and “what should be done.”

  6. 6

    Packaging SOPs into a shareable interface (e.g., a React-based app) improves adoption, especially for teams that aren’t comfortable reading raw chat outputs.

  7. 7

    There’s a time advantage in setting up an AI co-pilot for SOP creation early, because the setup effort comes before the widespread shift to this workflow.

Highlights

The transcript treats SOPs as a three-tier system—end-to-end, department-level, and tactical single-person guides—so knowledge can scale beyond individuals.
A master prompt that enforces concise, one-question-at-a-time clarification and confirmation can make AI outputs reliable enough to use as a starting point.
RACI is used to assign ownership in generated SOPs, with Accountable restricted to one person to avoid “nobody owns the KPI.”
AI can generate parallel workflow tracks (on-market vs off-market) with KPIs and decision points, then show where paths converge.
Turning SOP artifacts into a clickable React app is presented as a practical way to make documentation usable for non-technical team members.

Topics

Mentioned

  • SOP
  • LOI
  • RACI
  • AE
  • KPIs