Blindspots

A living document. Updated as patterns emerge. Read this when you're making a big decision and want someone to check your wiring.

Last updated April 12, 2026 — synthesized from 14 Reflections across Chesky, Fadell, Hockey, Gokul, Tobi, Des, Ben Horowitz, Katie Burke, Parker Conrad, Jean Lee, Karpathy, Niccol, Brooks, and Carroll.

1. Builder Over Seller

You default to building. Refining a brief, perfecting a role title, tightening a design — these feel like progress because they produce artifacts. But artifacts don't close revenue.

Gokul Rajaram scores FeatureOS at ~1 out of 8 on the moat scale. Distribution moat: 0. Network effects: 0. That's the danger zone.
Brian Chesky says founder mode means staying close to the work that determines survival. You're in founder mode on aesthetics. The survival work is distribution.
Your CLAUDE.md has detailed sections on design, brand, craft, team, and strategy. The word "sales" doesn't appear once.

"Is the thing I'm working on right now the thing that gets us to $1M ARR fastest?"

Your default under pressure is to refine rather than push forward.

Today's exhibit: a 500+ note Obsidian vault with custom CSS, a homepage dashboard, 14 person folders with Reflections, and a /summarize skill that dispatches parallel AI agents. Beautiful. Revenue impact: zero.
Ryder Carroll teaches simplicity with one notebook. You summarized his 10-minute video using parallel AI agents into a 500-note vault. The irony should not be lost.

"Would the outcome change if I stopped here?" If no — stop. Ship it.

3. Consumption as Substitute for Action

You consume at elite volume. 18 video summaries, 39 books, 500+ reference notes. The system for learning is world-class. The conversion rate from learning to doing is not.

Andrej Karpathy: you're spending more tokens organizing your own thinking than building AI into the products people pay you for.
William Hockey: value is in 20,000 pages about ONE boring thing. Are you reading 20,000 pages about product feedback infrastructure, or skimming 200 pages each across 100 interesting topics?
Ben Horowitz: you consume CEO content instead of building CEO peer relationships. YouTube is not a room.

When did learning become more comfortable than doing?

4. Loyalty Delaying Hard Calls

You carry people carefully. That's a strength — until it isn't.

Katie Burke: you asked for unfiltered CEO feedback in March. What have you done with the uncomfortable parts?
Ben Horowitz: management debt in a 5-person team compounds faster than in a 500-person team. There's nowhere to hide underperformance.
You make 52%-right decisions on product all day. You avoid 52%-right decisions on people.

"Am I designing this role because the company needs it, or because I need a place to put this person?"

5. Written Clarity as a Shield

You write your way through problems instead of talking through them.

Alison Wood Brooks: on your last customer call, how many follow-up questions did you ask vs. how many times did you pivot their answer into a pitch?
A 10-minute conversation with a churned customer would teach you more than a 3,000-word summary of someone else's conversation.

"Could I resolve this faster with a 15-minute call?"

6. Taste as a Bottleneck

Taste that lives in one person's head doesn't scale.

Tobi Lütke spent 20 years going deep on ONE domain. You're across FeatureOS + SupportWire + blog + skills + vault. Is the craft directed at one thing deeply enough?
Tony Fadell: has the press release been written? Not for the vault — for the product a customer is going to buy.

"What did Linear look like at MY stage?" — not what they look like now.

7. SupportWire Is Undefined

SupportWire has been "pre-launch" long enough that the phrase has lost meaning.

Des Traynor shipped Fin — 1M resolutions/week, 27+ subsystems, $50M+ invested. That's the market you're entering.
Andrej Karpathy: is the AI actually built, or is "AI" a positioning word on a landing page?
There is no press release. There is no torture test. There is no product marketing.

If you can't describe SupportWire's advantage over Fin in one sentence, the product doesn't have one yet.

8. No Distribution Operator

Gokul Rajaram: every great founder needs an "Eric." You don't have one.
Your gaps are all GTM. Your strengths are all product. The compound that matters isn't FeatureOS + SupportWire. It's Karthik-the-product-person + someone-who-sells.

Can you describe your sales motion with the same specificity you describe your design principles?

9. Principle Overload

Eight operating principles. Six team rules. Five brand guidelines. A 14,000-word CLAUDE.md.

"Move fast but don't ship garbage" + "I'd rather ship one thing that's beautiful" = tension with no resolution.
Brian Niccol: what is the core experience? You haven't answered this for FeatureOS.

If you could only keep three of these principles, which three? Those are real. The rest are preferences.

The Pattern

Every Reflection points at the same loop:

Stress → improve the website → summarize another video → build another tool → revenue stays flat → stress.

The vault is beautiful. The graph is rich. The revenue is the same as it was three months ago.

Five Moves

Define the FeatureOS core experience in one sentence. Strip everything that conflicts. Niccol did this at Starbucks in week one.
Ship SupportWire in 90 days. Write the press release (Fadell). Set a hard deadline.
50%+ of the week on distribution. Customer calls, pipeline, partnerships. Not product. Not the vault.
Call 10 churned customers this month. No pitch. Just questions (Brooks).
Find a founder peer group. 3-5 bootstrapped founders. Monthly. Real humans.

The blindspots you're aware of are manageable. The dangerous ones are the ones you haven't named yet.