Founder Split and the Zigsaw Spark
Zigsaw began as a classic co-founder split: product vs. go-to-market. We built fast and won the YC Dedalus hackathon, which put us on blast. The attention was oxygen—we closed design partners off the back of that momentum. It taught me that distribution is oxygen too; without it, the best tech flatlines.
If you think you have no competitors, that's the red flag. Embedded LLM ad-tech looked “blue ocean,” but the absence of rivals meant the market itself was cold. We learned to search for heat, not quiet.
Finding the Right Problem
We chased “cool” until we chased “pain.” Talking to founders every day made it obvious: the winning wedge is the problem with budget and urgency. For us that was creative ad throughput—teams drowning in backlogs and compliance reviews. When we tied LoRA fine-tuned assets to measurable lift (CTR, CPA), pilots stopped being “interesting” and started being contracted.
Pivot Hell and Dual Zoom
Pivots are supposed to feel smart; mostly they feel like doubt. The rare skill is zooming into the weeds—latency budgets, model constraints, UX affordances—while holding the big picture (market, price, channel) in view. Losing either is how you waste a quarter.
Courage is leaving the lab to talk to investors, customers, and hires. Mental strength is pushing through the imposter feeling on skills you weren’t “trained” to have. Discomfort is pitching a vision with conviction long before it’s “ready.” Grit is shipping v0.1 hardware knowing it might break tomorrow. And the emotional volatility of balancing research rigor with startup speed is…constant.
Hiring, Culture, and Speed
- Hire for bias-to-ship: builders who demo daily and measure weekly.
- Culture of receipts: every bet gets a metric and a timebox.
- Async by default; synchronous only for decisions or debugging.
- Clear owners; no feature without a DRI and a success metric.
How We Operated
- Find desperate customers with hair-on-fire problems; manually recruit them even if it’s unscalable.
- Launch an MVP fast, charge early to test commitment, and iterate in tight feedback loops.
- Pick 1–2 metrics and do 90/10 work: 10% effort for 90% of the value that moves those numbers.
- Avoid bad advice traps: don’t perfect before launch; don’t rely on partnerships to sell for you; plan scalability while doing the manual work.
Fundraising Lessons
Money follows momentum, not theory. The best pitch was always a live demo with real customer logos, unit economics, and “we ship every week” energy. Data room hygiene mattered: short deck, crisp metrics, zero fluff. Investor questions rhymed—defensibility, distribution, margin, and “why now.” Having answers from live pilots beat any TAM slide.
Product and Go-to-Market Loops
- Ship → watch usage → cut the fat → deepen the winning edge (speed, quality, price).
- Pair every feature with a distribution motion: launch post, customer webinar, or targeted outbound.
- Keep infra debt low: kill features that aren’t pulling revenue or learning.
- Charge early to force clarity on value; free users rarely give signal.
Resilience Without Blindness
Resilience is keeping your morale intact when the setback lands. Obstinacy is refusing to change your mind when the data says you should. The trick is rational persistence: focus on expected value, adjust beliefs quickly, and keep the morale unbroken while the plan evolves.
Most days were unglamorous: debugging latency regressions, rewriting cold email copy, handling churn. The wins were earned by compounding small, boring fixes. The lesson: momentum is built, not found, and it survives only if the team can absorb bad news without losing speed.