Most important take away

Most startup orthodoxy (Founder Mode, OKRs, “always be raising,” “hire a VP of Sales early,” “move fast”) is advice that worked in a specific context and is misapplied broadly. Great founders should default to copying what’s been done for 80% of decisions, but for the 20% of truly important decisions they need to form their own hypothesis based on their company’s specific reality, then act with conviction in the right direction rather than just speed.

Summary

Actionable insights and patterns from Zach Perret (CEO of Plaid) on building, hiring, fundraising, and investing:

Career and leadership advice

• Learn to love the things you must do well. Zach hated recruiting until he reframed it as a competitive game (win/loss: can I find the best person and convince them to join?). If a core CEO responsibility feels painful, find the angle that makes it intrinsically rewarding to you.
• Cold outreach works far more than people think. Short, sincere emails to people you admire — five sentences max — produce meetings, hires, mentors, and customers. Most people respond to genuine interest plus a specific ask.
• “Hire for spikes.” Build teams where individuals are exceptional at one thing even with glaring weaknesses elsewhere; complementary spikes raise the team ceiling far higher than a team of well-rounded but unspiky generalists.
• Re-examine your “experience trap.” Plaid long avoided 15-20 year veterans assuming they wouldn’t hustle. As the company matured into specialized domains (e.g., ML fraud engineering), deep experience became critical — but you still filter for hustle and values. Aim for a deliberate ratio (Plaid uses ~70/30 outsiders vs. industry veterans in financial services).
• Beware the “I just need to hire someone” trap. When a search drags on, you’ll be tempted to settle. Sometimes that’s right, but most often that hire becomes a problem you’ll have to fix later.
• Recognize and celebrate progress, not just excellence. Going from a 2/10 to a 4/10 is doubling the quality. Zach’s flaw is refusing to celebrate “still bad” — but teams need manufactured/interim wins (Big Hairy Audacious Goal + “what’s the first thing we can celebrate that proves we’re on track?”) to sustain momentum.
• Ask “how confident are you and how much do you care?” — Zach speaks with default confidence even on low-stakes opinions, and others over-weight his words. Pair every strong-stated view with calibration so teams know when to push back.

Building product and company patterns

• Use atomic teams for new product bets: smallest sufficient unit (PM + a few engineers, maybe designer/data scientist), protected and even hidden from the rest of the company so the 950-person “how can I help?” deluge doesn’t DDoS them. Give them resources and time to find 1-2 design partners, then do milestone-based funding internally — but don’t over-processize it into an internal VC. Let real signal (a customer “jumping at the bit to buy”) drive escalation, not bureaucracy.
• Pursue “grind problems” — work that’s hard not because it’s intellectually clever but because few are willing to do it (Plaid’s early bank screen-scrapers were one example). The defensibility doesn’t need to exist on day one; it needs the potential to compound via network effects, scale, brand, or accumulated knowledge. Worry about immediate defensibility too much and you’ll miss great grindy bets; worry about late-stage defensibility too little and you’ll waste your time.
• “Snake eating sheep, not an elephant.” When expanding from one product to many, sequence them — one, then the next, then the next. Plaid launched 3 new product areas, went international, and went upmarket simultaneously; go-to-market broke first because every product team lobbies sales, sales can’t carry the explanation alone, and you’re forced into a far more complex specialized GTM model.
• Be right, then move fast. Velocity > speed. Pre-product-market-fit, gather signal and choose direction carefully. Post-PMF, move fast becomes the right default. “Speed in a given direction” — and the direction must also be right.
• “Sweat the details” instead of “micromanage.” Delegate effectively: stay close to what matters, step back from what doesn’t. The misread of Founder Mode is either “devalue execs” or “micromanage everything”; the actual lesson is don’t over-delegate.

Fundraising and capital strategy

• Treat capital as a resource, not a ritual. Plaid raises every 2.5-3 years, not on a manufactured A/B/C cadence at $1M/$5M/$20M ARR. Raise when you need the resource, not because you’ve hit an arbitrary milestone.
• The “always be raising” mantra wastes founder time. Raise infrequently, only when capital is the actual tool needed.
• Don’t hire a VP of Sales early if you can run founder-led sales. Zach surrounded himself with sharp young people who made him more efficient at sales rather than handing it off — he learned more about customers and product than any other path could have provided.
• OKRs were built for manufacturing. Software companies should not blindly copy Intel’s playbook. Goals matter; the mechanism should fit your business.
• Don’t apply revenue metrics before product-market fit. Pushing top-of-funnel metrics pre-PMF is a common, damaging piece of VC advice.
• Enable secondary liquidity for employees, even when it’s messy. After the Visa deal fell through, Plaid worked to let a majority of employees participate in secondaries — necessary because employees had mentally spent the acquisition money.
• Founders should be allowed (and arguably encouraged) to take some secondary themselves. Making “a little bit of money” changes how a founder can think long-term.
• Valuations are essentially marks, not values. Private company valuation is just whatever the next marginal share trades at. Don’t optimize purely for max valuation — Zach feels a fiduciary responsibility to make his investors money rather than maximize headline numbers.

On Founder Mode and contrarian views

• Founder Mode (PG’s essay) will likely be one of the most misused blog posts in startup history. The lessons apply at Brian Chesky’s specific scale and context; most founders are not Brian. Copying Steve Jobs’ Top 100 offsite, or any other specific tactic, without understanding why it worked is dangerous.
• For 80% of company decisions, just copy what’s been done before. Spend your originality budget on the 20% that actually need it.
• Most VC advice is generic and shouldn’t be applied universally. A few VCs add real value (Zach cites NEA’s Ravi Viswanathan for stepping up during a near-collapsed round).
• VC has shifted from boutique industry to commoditized asset class. Average returns must come down because capital has grown faster than great companies. The traditional VC product may be played out.
• M&A is largely shut for big transactions due to regulatory scrutiny. IPOs remain a milestone/fundraising event — go public when you need the capital and the rigor, not for status.

On angel investing and fund-building

• Angel investing as a busy operator is distracting: it’s personally on you, no infrastructure, no leverage.
• If you want investment exposure as a CEO, build leverage: Zach partnered to create Mischief Fund — three full-time partners, he contributes a few hours weekly. The byproducts (staying close to early-stage thinking, network, learnings he ports back to Plaid, blog posts that scale his answers across portfolio founders) are why it’s a positive-sum activity rather than a distraction.

Personal practices

• Make hypothesis-driven decisions. Always start with a tentative answer to any question, then refine. This forces clarity.
• Protect solo thinking time. Zach’s favorite work mode is a notebook on a plane with static playing through headphones — full isolation produces clarity unavailable in normal operating mode.
• Becoming a parent expands your capacity to see depth in people you thought you already knew. Apply that openness to colleagues.
• Define success as excellence in your specific craft, not as more/bigger/faster.

Tech and macro patterns noted

• One-tap experiences (Shop Pay was the wake-up call; Plaid’s 5-year vision is one-tap loan applications via Face ID).
• Meta Ray-Bans called out as a “this is happening” moment for ambient computing.
• AI agents and automated workflows reshaping infrastructure (heavy sponsor presence: UiPath, Attio).
• Financial fraud is growing at an increasing rate in the US — flattening or reversing that curve is a major Plaid bet.

Chapter Summaries

1. The Visa refounding moment — Plaid signed paperwork to sell to Visa in 2020; regulators investigated, and growth during COVID let Plaid walk away. The first decision to sell was 51/49; a year later the decision to walk was 99/1. Hardest decision of Zach’s career.

2. Expanding from one product to many — Plaid simultaneously launched three new business areas, went international, and went upmarket. In hindsight: sequence them (“snake eating sheep, not an elephant”). Go-to-market broke first as sales teams got internally lobbied and customers received muddled pitches.

3. Atomic teams and new product development — Smallest sufficient team for each new bet, protected from the rest of the company, milestone-based internal funding, but not over-processized into an internal VC model.

4. Grind problems and defensibility — Hard, unglamorous work that few will do (early bank screen-scrapers) creates differentiation over time. Don’t over-index on day-one defensibility; do think about long-term moat potential.

5. Recruiting and talent — Zach learned to love recruiting by reframing it competitively. Cold outreach with five-sentence emails works. “Hire for spikes.” The experience trap evolved — deep specialists matter as you mature, but filter for hustle.

6. Hiring mistakes — Settling under search fatigue is the most common harmful pattern.

7. Founder Mode critique — Likely to be misused as justification for either devaluing execs or universal micromanagement. The real lesson: don’t over-delegate the things that matter; sweat the details where it counts.

8. Contrarian Silicon Valley beliefs — VPs of Sales early are often wrong, OKRs are misapplied software-to-manufacturing, speed must be paired with being right (velocity over speed).

9. Fundraising philosophy — Raise every 2.5-3 years only when capital is the tool needed. The seed round near-death story: lead backed out, three “Justins” invested $60K total, kept Plaid alive.

10. Secondaries and the Visa aftermath — Employees mentally spent the acquisition money; when the deal fell through, Plaid worked hard to let employees access liquidity in the next round.

11. Valuation, IPOs, and M&A — Private valuations are marks not values. Large M&A is regulator-blocked. IPO is a milestone, not a goal — go public when you need it.

12. Unmade decisions and personal flaws — Hypothesis-driven decision-making creates the appearance of confidence. Calibrate by asking “how confident, and how much do you care?” Zach’s biggest flaw: not celebrating progress that isn’t yet excellence.

13. Manufacturing wins and BHAGs — Pair big hairy audacious goals with “what’s the first thing we can celebrate that proves we’re on track?”

14. Self-sufficiency and fatherhood — Independence enables Zach’s clear-thinking time but makes him hard to read. Becoming a parent expanded his capacity to see depth in others.

15. Success and competition — Defined as excellence in one’s craft. Competition is motivating but the healthiest version is self-competition.

16. Mischief Fund and angel investing — Solo angel investing is distracting; partnering into a fund creates leverage and keeps him close to early-stage thinking.

17. Future of VC and M&A — VC has commoditized; average returns must fall. Few VCs add real value (NEA’s Ravi cited for stepping up during a difficult round).

18. Quick fire — Recommends The Wealth of Nations. Most contrarian advice: most things VCs tell you, don’t do. Tech most underrated: one-tap experiences (Shop Pay) and Meta Ray-Bans. Five-year vision: one-tap loans via Face ID; flatten or reverse the fraud growth curve.