SWLW #702: How I created OpenClaw, Your CEO is suffering from AI Psychosis, and more.
Oren Ellenbogen · email · 2026-04-20 Reader 🖥️
Effective tech leadership requires anchoring in measurable value and compounding effort rather than chasing the appearance of productivity, whether through AI theater, performative suffering, or shallow busyness.
practical · mixed — A curated newsletter collecting links with editorial commentary. Each entry is a practical recommendation for tech leaders, framed as 'read this because it will change how you lead.' The curator's commentary is argumentative-practical: here is a pattern, here is why it matters to you.
Structure
| Part | Role |
|---|---|
| This Week's Favorite: OpenClaw / Peter Steinberger | Opening hook — illustrates the thesis through a positive exemplar: someone who found passion by building something real with agents, not by managing dashboards |
| Culture section: AI Psychosis, Richard Hamming/Shannon, Suffering Isn't Leadership | Core evidence — three articles that attack different forms of the same anti-pattern: mistaking the feeling of productivity for actual value creation |
| Peopleware section: Management Wins, Being Strategic | Constructive counterpart — how to actually build value as a leader: track real wins, think strategically rather than reactively |
| Inspiring tweets / closing | Coda — reinforces the theme with aphorisms about outlier thinking and user-centricity |
Key terms
- AI Psychosis Emotional Impact of AI Work
A specific pathology in executive decision-making where leaders mistake token-burning AI dashboards and agent hierarchies for genuine productivity, getting the dopamine of delegation without measuring output - Compounding
The multiplicative (not additive) nature of sustained effort — someone working 10% harder produces not 10% more but 2x more over a career, because knowledge and productivity compound like interest - Pathological pessimism
A default negative lens on work that caps innovation and energy, distinct from healthy skepticism — a leadership dysfunction that may require coaching or therapy to address - Strategic thinking
Not planning on a longer timeline, but an entirely different lens: shaping what's possible rather than optimizing what's in front of you
Propositions
- explicitMost current enterprise AI adoption produces almost nothing of measurable value, despite looking and feeling like productive managementconf: high
- explicitKnowledge and productivity compound multiplicatively, not additively — a 10% effort advantage produces a 2x career output advantageconf: high
- explicitWorking on important problems is a prerequisite for doing important work — problem selection matters more than raw abilityconf: high
- explicitCustomer and user perception is the ultimate arbiter of company and product value, not internal metrics or AI-powered dashboardsconf: high
- explicitTreating leadership suffering as a badge of honor is itself a leadership dysfunction, not a sign of dedicationconf: high
- implicitAgents that improvise are fundamentally different from chatbots that give up — the distinction marks a new paradigm in AI-assisted buildingconf: medium
- implicitUrgency and intensity, combined with an ownership mindset, are the primary differentiator between top performers and average performers of equal intelligenceconf: high
- explicitManagers need 'brag documents' just as much as individual contributors do — small wins compound into strategic insightsconf: high
Arguments
Leaders should anchor AI adoption in measurable customer value rather than in the feeling of commanding AI agent fleets strength: high
Explicit premises:
- Current AI platforms are designed to make operators feel like they're commanding a fleet
- These platforms produce dashboards and governance layers that mimic management
- They provide the dopamine of delegation without measuring output
Implicit premises:
- The feeling of productivity is reliably distinguishable from actual productivity only through external measurement
- Executives are systematically vulnerable to confusing process sophistication with outcome quality
- The AI vendor ecosystem has an incentive to maximize the feeling of control rather than actual results
Inferences:
- Organizations that measure AI ROI by token spend or agent count rather than customer outcomes will destroy value
- The current wave of enterprise AI adoption may produce a correction similar to previous technology hype cycles
Sustained intensity on important problems is the primary driver of exceptional career outcomes strength: medium
Explicit premises:
- If you do not work on an important problem, you will not do important work (Hamming)
- Knowledge and productivity compound like interest
- A 10% effort difference compounds to a 2x output difference over a career
Implicit premises:
- Most people do not choose their problems deliberately — they inherit them from organizational context
- The compounding effect applies to intellectual capital, not just hours worked
- There exists a reliable way to identify which problems are 'important' in advance
Inferences:
- Organizations that allow talented people to self-select their problems will dramatically outperform those that assign work top-down
Narratives
- Peter Steinberger's journey creating OpenClaw — 'one of the most productive humans on the planet' finding passion in building software again through AI agents
→ Implicitly argues that genuine productivity comes from passionate building with new tools (agents), not from managing or delegating to them — a positive counterexample to the AI Psychosis pattern - Richard Hamming sharing an office with Claude Shannon at Bell Labs, observing why some equally smart people produce Nobel-worthy work while others produce forgettable work
→ Supports the proposition that problem selection and sustained intensity matter more than raw intelligence — provides historical authority for the compounding argument
Questions addressed
- ✅ Why does so much enterprise AI adoption feel productive but produce little measurable value?
Because the platforms are designed to produce the feeling of management (dashboards, hierarchies, governance) rather than actual outcomes — the dopamine of delegation substitutes for measurement - ✅ What separates people who do important work from equally capable people who don't?
Problem selection (working on important problems) and compounding intensity (sustained effort that multiplies rather than adds) - ⭕ How should a tech leader transition from reactive management to strategic leadership?
Pointed at via the Gefroh series but not fully developed in the newsletter itself — it gestures toward 'shaping what's possible' as a different lens
Gaps
- incompleteThe newsletter juxtaposes the OpenClaw story (passionate builder using AI agents productively) with the AI Psychosis critique (executives using AI agents unproductively) but never articulates what distinguishes productive from unproductive AI agent use. The implicit answer seems to be 'building vs. managing' but this is never made explicit or tested.
Read next: Read the actual OpenClaw video/article and the AI Psychosis article side by side to extract the distinguishing criteria - incompleteThe compounding argument (10% effort → 2x output) is stated as fact but has no empirical backing cited. The Hamming lecture is anecdotal authority, not evidence. The actual compounding dynamics of knowledge work are far more complex than financial interest.
Read next: Cal Newport's 'Deep Work' or Ericsson's deliberate practice research for more rigorous treatments of effort-to-outcome dynamics - incompleteThe curator's editorial voice synthesizes across articles but never interrogates the tension between 'seek measurable value' (AI Psychosis commentary) and 'shape what's possible rather than optimize what's in front of you' (strategic thinking commentary). These are potentially contradictory: strategic thinking often requires investing in things that are not yet measurable.
Read next: Roger Martin's work on strategy as an integrative activity, or Clayton Christensen on the innovator's dilemma of measurability bias
Related documents
- AI Enthusiasts Are in a Race Against Time, AI Skeptics Are in a Race Against Entropy
- How People Are Really Using AI in 2026
- Please stop the AI Confidence Theater
- How tech workers actually feel about AI in 2026 | Annual AI sentiment survey (Noam Segal)
- Itchy Brain (Still Burning: Michael Grinich & Kent Beck on AI adoption, enterprise developer tools)