Whitney Hazard

Tony Xu on DoorDash's 3Q25 earnings call: Great companies start as experiment factories: "A lot of times, when you're building a company, what you're really doing is you're starting with lots of experiments. Some of those experiments make it into products. Some of those products can then be graduated into commercialization, where you're trying to test whether or not you have a good business. And then some of those candidates ultimately yield big businesses that generate the cash flows to allow you to invest in the next set of products and experiments." #DoorDash #localcommerce #fooddelivery

2

1 Comment

The Wall Street Journal asked me how we track AI costs at Zapier. Honest answer? We're still figuring it out. We recently started looking at token use across the company. When someone's burning through 5-10x more tokens than their peers, that's a place to get curious. Sometimes it's an anti-pattern to coach our way out of. But sometimes? It's a sign of innovation. A new best practice. A pattern we want to spread. The instinct for a lot of companies will be to enforce token limits first, and ask questions later. I think that's backwards. Build enough trust that people will experiment freely, and enough transparency that you can see what's working. You wouldn't take paint away from a talented artist. You'd give them more. https://lnkd.in/gYG4CAna Kudos to Katherine Bindley for spotlighting this topic!

259

33 Comments

Your Team's Attention Is the Actual Bottleneck, Not Your Data Every private lending org runs into the same ceiling eventually: reps can only make so many calls in a day, marketing has a fixed budget, and leadership only has so much bandwidth to review the pipeline. Given that ceiling, "how do we find everyone who could possibly borrow" is the wrong question to be optimizing for. You already know approximately who could borrow. That list has existed for years. The question that actually determines revenue is: where does the next unit of attention go? That reframe changes what "good data" even means. A system that hands your team 5,000 more qualified names hasn't solved anything if your reps can only work 200 of them this week. What actually helps is a system that tells you which 200 to work first, and, just as importantly, which names to leave alone until the timing changes. Attention is the scarce resource in this business. Everything else is just infrastructure around it. Full breakdown – link below 👇

2

1 Comment

Chatted with a PE partner today. Lots of startups target tech/other startups. Many other places to sell into, but a few things to beware of. PE is a different model - Costs are carefully monitored - Finance is carefully forecasted - Decisions are (usually) made thoughtfully Selling into early stage tech/venture backed is great because people have tremendous responsibility and tend to try things quickly. PE is not necessarily the same. Not all PE is the same - Turnarounds - Roll ups - Leveraged buy outs - So. Many. More. Models. Each model has its own operational model, and even within an industry they can be different. Vista Equity Partners is famous for bringing its operating model into its portfolio companies. Other PE firms aren't as involved. My friend said that his turnaround firm typically doesn't get involved–they just hire a new team. The turnaround part is more about the purchase–not a hands-on approach. Do you sell product or services into PE? What have you learned?

8

4 Comments

Hot take: I don't think MCPs are here to stay. I think AI will be smart enough to use the tools and services we have today (API, SDK, UI) to do anything a human/dev can.

63

28 Comments

Driving AI adoption in a finance org is a lonely job. The board wants results, the teams want proof, and you're in the middle with a mandate and no playbook. For two years, the standard advice was "ask every team where AI could help." If you ran that playbook, you did what everyone recommended. It produced ideas. It rarely produced change. Here's the playbook I give finance leaders instead. I call it O-R-D-E-R. 1. Own it from the top. Your teams can't spot where AI helps until they've worked with it. Exposure comes before ideas, not after. So don't open with a survey asking where AI could help. Open with leadership choosing one tool and telling teams: use it on your daily work for a month. Once people feel what AI does well and where it breaks, the ideas come, grounded in reality instead of headlines. 2. Responsibility to one person. A transformation lead whose only job is to drive this. "Everyone's responsible" means no one is. This person chases the blockers, owns the timeline, and reports on one thing: is the number moving. Without them, AI adoption becomes everyone's side project, and side projects lose to month-end every single time. 3. Define the number that hurts Not "where can we use AI." Take the 13-day close, split it into AP days, AR days, cash days, and point a team at each one. A number gives you everything a vague mandate can't: a baseline, a finish line, and a way to know if any of this is working. Teams rally around "cut AP from 5 days to 2." Nobody rallies around "explore AI." 4. Everyone on one dataset. On a normal day, procurement, AP, controlling and accounting each run their own world. That's the job. But a transformation project is different: shared goal, shared data. One source of truth, not four versions of it. The close isn't slow inside any one team, it's slow in the handoffs between them, and you can't fix a handoff each side measures differently. 5. Resist the early quit. AI gets things wrong at first. So did your last hire. The losing teams stop at the first "see, it's not ready." The winning ones treat early errors the way they'd treat a new joiner's mistakes: correct them, feed them back, and watch the accuracy climb. The gap between those two reactions is the entire difference between teams that transform and teams that pilot forever. The teams getting results don't have better technology. They have a clearer order, and they hold it. Finance teams, I'm curious: how are you driving AI adoption in your org right now?

44

8 Comments

Fun dinner last night with an exciting group of founders and builders in San Francisco. The conversation went many places...we got into how pricing models are being upended in the AI era, whether the application layer is disappearing or being fundamentally reinvented, what actually predicts AI adoption inside organizations (spoiler: it's not age or technical background), and how go-to-market is shifting as the cost of building software collapses. A few threads : * The app layer isn't dying, it's shapeshifting. Foundation models are pushing into the application layer, but that doesn't eliminate apps. It reinvents them as the context-aware scaffolding around intelligence. The next generation of durable software businesses will be built with . * Curiosity is the real leading indicator. The strongest predictor of AI effectiveness inside an organization isn't domain expertise or seniority : it's the intrinsic drive to go try things. Most performance management systems don't even measure for this. That's a blind spot. * Pricing is an important unlock, and most companies are stuck. We're shifting from seat-based to outcome-based, but most teams haven't defined what "outcome" means for their product yet. True outcome based pricing that actually aligns with how customers experience value is hard to build but a massive competitive advantage. What I love about these conversations is that the questions don't have clean, binary answers, and everyone in the room knows it. That honesty is what makes the thinking sharper. Thanks to Matthew Jacobson, Kelly Zheng, and the ICONIQ team for bringing together such a thoughtful group.

284

17 Comments

AI is changing the structure of most companies we see. Lots more coming from us on how startups can build faster, leaner teams -- and where we see all of this going. Great piece from Pete Flint to kick it off: https://lnkd.in/dcUebcJj

57

1 Comment

The VC herd is shifting course from once-safe, now-strained SaaS to Aerospace & Defense, an alluring prospect that plays by different rules. SaaS isn’t the predictable play it once was. AI is rewriting the playbook: erasing moats and shifting buyer needs, eroding the predictability that made SaaS so attractive. The secret SaaS sauce wasn’t magic, just choreography: product, fundraising, and traction timelines that neatly lined up. Easy to take for granted. The SaaS timelines are predictably in sync. → Product development moves quickly, and you have control. No physics, no hardware. → Fundraising is a well-oiled machine. Do X, and you got Y. Deep pools of investors who know the playbook. → Commercial traction is straightforward. For years, if you understood what your customer needed and built it, they bought. A&D is different. The timelines don’t cooperate. → Product is harder. In SaaS, QA is a bug report. In A&D, QA means pummeling hardware in vacuum chambers, at hypersonic speeds, on vibration tables, at extreme temperatures. → Fundraising is less predictable. Fewer seasoned investors, shifting milestones, heavy upfront capex and R&D, dependence on grants. → Commercial traction is murky: buyer cycles are political, slow, and opaque. SBIRs and STRATFIs help, but often create false positives. This doesn’t make A&D less attractive. But, it does make it different. The excitement is real: DIU’s FY 2024 appropriation was nearly $1B, and since 2016, it has awarded $5.5B in contract ceilings to commercial companies, many of them startups. The Big Beautiful Bill added $156B to DoD’s $831B FY 2025 budget, bringing total defense resources close to $1T. In A&D, timelines rarely align. Investors need a playbook that assumes gaps, delays, and unpredictable orders. Success will go to those who underwrite to that reality.

51

3 Comments