Stop Blaming AI. Your Systems Are Broken.
The speaker argues that AI implementation failures in businesses stem from broken underlying processes, not AI itself. Drawing on decades of corporate and entrepreneurial experience, he explains that without clear workflows, single sources of truth, and proper delegation structures, neither AI nor any other productivity tool will succeed. He then demonstrates how his own AI agent team is structured using simple folder systems built on the same organizational principles that apply to human teams.
Summary
The video opens with the speaker framing this as his most important video, citing a statistic that 95% of companies attempting AI implementation have failed. His central thesis is that this failure reflects broken business processes, not AI's limitations — AI failure is merely a symptom of pre-existing organizational dysfunction.
To support this, he shares a detailed case study from his corporate career around 2020, where he took over a team of five people managing 150 parallel global projects. He discovered that a single departing employee had become a critical bottleneck, holding all institutional knowledge accumulated over 15 years. By mapping business processes visually and redistributing work by function rather than by individual, his team increased project performance by 40% with fewer people — without any AI or automation involved.
He introduces the concept of a 'single source of truth' as a foundational productivity principle, contrasting it with the common corporate dysfunction of scattered information, redundant meetings, and miscommunication through meeting minutes written by non-experts. He argues that most businesses fail not because of missing tools, but because people don't know what they're doing day-to-day, KPIs are misleading ('watermelon metrics' — green on the outside, red on the inside), and there is no continuous improvement feedback loop.
The speaker then positions AI on a spectrum between delegation and automation. Unlike pure automation (which breaks when inputs vary), AI handles variation but produces inconsistent outputs — roughly 80% quality. He argues this is still superior to human assistants, who cost more, work fewer focused hours, and are subject to personal unpredictability. He calculates that paying $200/month for Claude versus $2,000/month for a human assistant, while receiving comparable or better output consistency, makes AI economically compelling.
In the third section, he reveals his practical AI agent setup using a simple local folder structure he calls PKA (Personal Knowledge Assistance). The structure includes a Business Knowledge Management (BKM) folder with SOPs, expert knowledge, and brand assets; a team folder containing named AI agents (Larry the orchestrator, Nolan for HR, Pax for research, Iris for brand design, Charter for infographics, and a QA agent); and an owner's inbox for reviewing agent outputs. He emphasizes that this requires no coding, no GitHub downloads, and no complex prompt engineering — just well-organized text files that give AI agents the context and guardrails they need.
He also describes a persistent memory and self-improvement mechanism: agents log each session in a journal file, enabling them to learn from past interactions and maintain consistency over time. A QA agent reviews work before it reaches the owner's inbox, creating an internal feedback loop. The speaker concludes by urging both individuals and business owners to fix their foundational productivity systems first — understanding goals, projects, tasks, and workflows — before layering on AI, just as automation always came last in his methodology.
Key Insights
- The speaker argues that the 95% failure rate of AI implementation in businesses has nothing to do with AI itself, but reflects unclear and dysfunctional underlying business processes — AI failure is simply a diagnostic symptom of pre-existing organizational breakdown.
- The speaker claims that AI sits between delegation and automation on a spectrum — unlike automation it tolerates variation in inputs, but unlike human workers it delivers a consistent 80% output quality, which he argues exceeds the real-world consistency of human assistants who cost far more and work focused for only 3-4 hours per day.
- The speaker describes 'watermelon KPIs' — metrics that appear green on the outside but are red on the inside — as a systemic corporate problem where teams manipulate numbers to satisfy management, causing leadership to make wrong decisions and then be blamed for outcomes that were caused by bad data from below.
- The speaker contends that the most dangerous jobs are not high-skill roles but low-effort, time-consuming tasks like data entry and copy-pasting — work that was previously delegated to cheaper human workers — because AI now performs this work at a fraction of the cost with no meaningful loss of quality.
- The speaker reveals that his entire AI agent team — including an orchestrator, HR agent, researcher, brand designer, infographic creator, and QA agent — is built using nothing more than a local folder structure of plain text files, with no code, no GitHub downloads, and no complex prompt engineering, demonstrating that organizational clarity matters more than technical sophistication.
Topics
Transcript
[0:00] I consider this the most important video I ever made and I launched this channel in 2017 with the first videos about productivity and my experience working in corporate and climbing the ladder there and I shared over the time all the insights and struggles that we have thriving in a business world. In 2022 I met my co-founder Paco Cano. He built four businesses in parallel in the past 20 years and he came to the same conclusions when it came to productivity [0:30] in a business. So we talk here about decades of productivity advice that helped us to be highly efficient inside business and we have seen this over and over again. Today it's AI. In…
Full transcript available for MurmurCast members
Sign Up to AccessMore from ICOR with Tom | AI Productivity
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.
An AI Just Handed Me a Fake $67B Statistic
The speaker shares how an AI confidently generated a completely fabricated $67.4 billion statistic while researching AI hallucinations, then explains his systematic approach to combating AI misinformation: using dual independent search engines to cross-check answers and identify where hallucinations occur.
AI Is Quietly Taking Your Memory
AI companies are racing to own users' context and memory through integrated tools like Claude in Slack and ChatGPT's auto-memory features, creating a lock-in trap. The speaker argues users should own their context (expensive, irreplaceable) in local plain text files while renting the AI model (cheap, replaceable), and demonstrates his personal knowledge assistant system built on markdown files that work with any AI model.
My AI Team Now Has an Interface. All 12 Agents. Free.
Tom, creator of the iCola methodology, releases My PKA version 3 with a free interface (My PKA Cockpit) that provides 12 pre-built AI agents for productivity, note-taking, health tracking, document management, and task planning—all running locally in a folder structure accessible via Claude and a web interface.