👋 Hi, I’m Andre and welcome to my newsletter Data Driven VC which is all about becoming a better investor with data and AI.
💰Find new LPs in our database of 30k+ active family offices, fund of funds, pensions & more via Final Close

Brought to you by Harmonic - The Startup Discovery Engine
The best investors use Scout to understand startups, the people behind them, and the markets they compete in.

Welcome to another Data Driven VC “Insights” episode where we cover the most interesting research and reports about startups, GPs, LPs, AI & automation.
Venture Is Dead (Again)
Jerry Neumann at Colossus argues in Venture is Dead (Again) that venture works only when partners take small, staged bets on truly uncertain startups, and that ballooning fund sizes are breaking that model for the second time.
Capital Is Concentrating: Funds over $1B have taken 72% of venture capital raised so far in 2026, up from 34% in 2025 and 16% ten years ago, while the firms raising them aren't growing their partnerships to match.
Too Much Money per Partner: Benchmark's $2B raise this June gives each partner about $400M, up from about $120M in 2024. Neumann argues that much money per partner forces VCs to either make bad deals or take on less uncertainty by investing later or buying established companies.
The 1980s Are Repeating: Back then, capital per VC professional tripled, median IRR fell from above 30% to about 8%, and firms drifted into buyouts. Today, Lightspeed has raised $750M+ to buy companies and rebuild them with AI, and Thrive is buying ADT.

✈️ KEY TAKEAWAYS
Neumann expects venture to recover as it did after the 1980s, when Kleiner and Accel stayed early and became dominant in the 1990s, so LPs should watch smaller funds dedicated to uncertainty. For GPs, the test is whether fund size still matches the number of companies each partner can actively support.

Family Offices Shift Toward SPVs
Hamza Shad at Carta finds that family offices are moving capital from funds into SPVs, which may help explain why emerging managers are taking longer to close their funds.
A Third of Commitments Now Go to SPVs: In H1 2026, 33.7% of family office commitments of $500K or more went to SPVs, up from 8.6% in 2019. Most of the jump came after 2023, which Shad ties to stalled VC distributions plus AI mega-rounds and tender offers.
Deal-Level Control Carries Risk: SPVs let family offices decide deal by deal, but Shad notes they concentrate risk and may pull investors into rounds that are already hot and richly priced.
Funds Still Win on Dollars: Because blind-pool checks remain larger, funds captured 73.5% of family office dollars deployed in H1 2026, above their 66.3% share of commitments, so funds still absorb most of the capital even as SPVs win more of the deals.

✈️ KEY TAKEAWAYS
For emerging managers, the shift may help explain slower closes, as family offices weigh deal-level flexibility against a 10+ year fund commitment. The case for the fund rests on what Shad says SPVs give up: the diversification and power-law exposure the VC model depends on.

Upgrade your subscription to access our premium content & join the Data Driven VC community

AI Natives Pull Away From AI-Enabled SaaS
Standard Metrics' AI vs SaaS Growth Race report compares AI natives, where AI is the product, with SaaS companies that added AI to an existing product, and finds the growth gap widening.
A Widening Gap at Every Size: In Q3 2026, top-quartile AI natives with $1-20M in annualized revenue grew 275% year over year against 102% for AI-enabled SaaS. Above $20M, it was 197% against 53%, with gaps up from 41 and 13 points in Q1 2023.
Smaller AI-Enabled SaaS Is Sliding: At $1-20M, AI-enabled SaaS growth slid from 165% in Q1 2023 to a low of 92% in Q2 2026 before ticking up to 102%. Above $20M, it bottomed out in 2024, so the gap there comes from AI natives accelerating.
The Lead Holds Beyond the Outliers: At the median, AI natives at $1-20M grow 80% a year against 30% for AI-enabled SaaS. Part of that reflects a smaller base, but the report says the lead holds within narrower size bands too.

✈️ KEY TAKEAWAYS
For investors, a SaaS company adding AI features has not closed the gap in this data, so diligence should test whether AI is the product or an add-on. The lead holds when age is held constant: among companies founded since 2020, AI natives lead by 666 points in healthcare and 471 in software.

From Agent Tasks to Agent Goals
LangChain CEO Harrison Chase argues in From tasks to goals that as models get more capable, agents are moving from completing assigned tasks to pursuing ongoing goals, which changes how they are built and managed.
Outcomes Replace Instructions: Chase sees two shifts: from specifying an action to specifying an outcome, and from a bounded assignment to an ongoing responsibility. Drafting one post is a task, while maintaining a blog is a goal.
Background Agents Need New Plumbing: Goal-driven agents work in the background, so they need triggers that wake them, the judgment to decide when acting is worthwhile, a way to ask humans for clarification, and limits on which tools they can use.
Feedback Has to Compound: To track progress toward a goal, Chase pairs a measurable metric with human corrections, since chasing a metric alone can backfire. Memory lets each correction shape future work, so users don't have to repeat themselves.

✈️ KEY TAKEAWAYS
Chase frames the shift as delegating decisions about what work to do and when, which moves the user's role to setting direction and reviewing progress. For investors, the building blocks he lists (triggers, permissions, feedback loops and memory) point to where new agent infrastructure demand could form.

Find new LPs in our database of 30k+ active family offices, fund of funds, pensions & more via Final Close

How Often Founders Leave
Prompted by Jensen Huang's resignation from NVIDIA after more than 33 years as CEO, Carta's Hamza Shad looked at how often founders leave the startups they started, using data on privately held US companies.
Nearly Half of 3+ Founder Teams Split: In the 2019 to 2021 cohorts, 44% to 47.5% of companies with three or more founders lost at least one within five years, against about 30% of two-founder teams.
Early Departures Are Rising: First-year founder losses at companies with 3+ founders doubled from 11.5% for the 2019 cohort to 23.3% for 2025. Two-founder teams show the same pattern, rising from 5.2% to 11.4%.
Solo Founders Rarely Leave: In the same cohorts, only 8.3% to 10.2% of solo-founded companies lost their founder within five years, and first-year losses have stayed between 1.7% and 2.5% in every cohort.

✈️ KEY TAKEAWAYS
For investors backing teams of three or more, the data supports agreeing vesting, departure terms and clear roles before the round closes. Solo founders depart less often, but each departure removes the company's only founder, so succession planning deserves a question in diligence.

5 GTM Plays Working Right Now
Kyle Poyar at Growth Unhinged shares 5 GTM plays working right now, drawn from operators at incident.io, Clay, Vercel and others.
Context Makes AI Agents Work: incident's CMO replaced his ad agency with a Claude skill that cut ABM campaign setup from 2-3 days to 3 minutes, which he credits to its access to approved messaging, CRM data and intent signals.
Attribution Must See Past the Last Click: Clay's agent found that a $200k deal credited to a Google ad was preceded by a free trial, a network referral and two webinars, which last-touch attribution would have missed.
Automation Scales High-Touch Channels: Vercel went from 1-2 to 3-6 field events a month without adding headcount, and executive dinners produce 5+ sales-qualified opportunities each.
✈️ KEY TAKEAWAYS
Poyar's common thread is taking a channel that already works and scaling it through automation while keeping what made it effective. For founders and portfolio teams, that means auditing existing channels for AI leverage before adding new ones.
That’s it for today!
Stay driven,
Andre




