💥Personal AI Operating Stack… | Data Driven VC Summary

1. Key Themes

Theme 1: The Extreme Power Law in VC — Median Performance Is Near Breakeven

The distribution of VC fund returns is far more skewed than most LPs appreciate. Carta's data on 400+ US funds from 2016–2018 vintages shows the bottom half of managers barely return capital.

"The 25th percentile fund sits at 1.06x net TVPI and the 50th percentile at 1.38x, both described in the data as close to a total miss on invested capital... The 99th percentile is a different business entirely: Top 1% funds post 13.12x net TVPI, an order of magnitude above the median."

Theme 2: AI Agent Capability Has Crossed a Critical Threshold — and Cost Economics Are Compelling

Computer-use agent performance has doubled in one year, and the cost structure now materially undercuts human labor — particularly offshore BPO.

"OSWorld-Verified Jumped From 42% to 85% in a Year: The best model scored 42% a year ago; today's leader, Claude Fable 5, scores 85%, above the ~72% human baseline... Inference runs $6-8/hour typical ($3-15 range) versus ~$10/hour for offshore BPO and $30-45/hour for US back-office labor, pencilling out to 70-80% gross margin against US labor even at the high end."

Theme 3: AI "Harness" Companies Are Commanding 2021-Era Multiples — But With 3x Faster Growth

The market is repricing top AI SaaS companies at peak multiples, but this time the underlying growth rates justify it differently.

"The 100x ARR Multiple Is Back: The article notes a 2021 piece on the '100x ARR multiple' as a historical premium; five years later, multiples are back at similar levels, but reported growth rates underneath are running roughly 3x faster... Legora, Sierra, and Ramp saw theirs accelerate starting January 2026, driven partly by sustained growth and a more favorable fundraising market."

Theme 4: Hype Precedes Winning Categories by 5–6 Years — But the Biggest Companies Rarely Come from the Hyped Topic

Sequoia's Hacker News analysis shows that hype is a leading indicator of a space, not of which company wins in it.

"LLMs first cracked the top 15 in 2016 and didn't hit #1 until 2022; AI coding entered at #12 in 2021 and topped the list by 2026; crypto entered in 2011, well before its 2017 and 2021 peaks... The Top Company of a Given Year Is Rarely the Hyped Topic of That Year: Airbnb (2008) launched while Google dominated the conversation, Uber (2009) arrived amid low-level programming chatter, and Anthropic (2021) was founded while crypto consumed the internet."

Theme 5: The Moat in AI Agent Businesses Is Shifting from Model Capability to Workflow Context

As foundational computer-use capability commoditizes, what creates durable defensibility is proprietary operational knowledge — not model performance.

"As raw computer-use capability becomes commoditized, buyers pay for the specific, hard-to-replicate knowledge of how one company's workflow actually runs... For a VC evaluating a computer-use agent startup, the thing to check is how much company-specific operational knowledge the product has already captured, since that's what determines defensibility once the underlying model becomes a commodity."

2. Contrarian Perspectives

Contrarian 1: The "Jockey" (Team) May Be Overrated as the Primary Investment Criterion

The conventional wisdom in early-stage VC is to back the team above all else. Stanford's data across 50 companies tracked from business plan to IPO challenges this.

"Fewer than three-quarters of CEOs at IPO had held the role at initial investment, while each company's core business line stayed stable."

This suggests the "horse" (the business idea and market) is more durable than the original "jockey." Teams rotate out; businesses persist. The implication: VCs who anchor too heavily on founder identity may be systematically mispricing business quality.

Contrarian 2: Current High AI Multiples Are Not a Bubble — They're Mathematically Justified by Faster Growth

The consensus reaction to 100x ARR multiples is often "bubble." But the article argues the current cohort is actually growing ~3x faster than equivalently-priced 2021 companies did.

"These multiples require growth roughly 3x faster than a normal SaaS company, which is what these five names are actually posting right now. If a portfolio company can't show that same growth premium, a 25-125x ARR multiple doesn't apply to it, no matter how similar the product category looks."

Contrarian 3: Sustained Hype — Not Novelty — Is the Durable Signal

Most investors screen for emerging or underappreciated trends. The Sequoia data suggests the opposite signal may matter more: persistent attention over many years.

"The 'Musk-verse' has held a top-15 spot for 14 consecutive years, the only topic with that kind of staying power in the dataset."

The implication is that durability of community interest — not recency or surprise — may be the more reliable indicator of investable magnitude.

3. Companies Identified

Harvey

Description: AI legal platform
Why mentioned: Case study in "AI harness" valuation; crossed $100M ARR as part of a cohort used to analyze current ARR multiples
Quote: "Harvey, Legora, and Sierra each announced crossing $100M in annual recurring revenue within nine months of one another"

Legora

Description: AI legal assistant platform
Why mentioned: Same cohort; notable for multiple acceleration in January 2026
Quote: "Legora, Sierra, and Ramp saw theirs [multiples] accelerate starting January 2026, driven partly by sustained growth and a more favorable fundraising market"

Sierra

Description: AI customer experience platform
Why mentioned: Part of the $100M ARR cohort used to benchmark AI harness multiples
Quote: "Harvey, Legora, and Sierra each announced crossing $100M in annual recurring revenue within nine months of one another"

Ramp

Description: AI-powered financial operations platform
Why mentioned: High-end data point in AI harness ARR analysis at ~$1.4B ARR
Quote: "Ramp (~~$1.4B ARR) and Decagon (~~$35M ARR) bracketing the group at the high and low end"

Decagon

Description: AI customer support agent company
Why mentioned: Low-end data point in the same AI harness multiple analysis at ~$35M ARR
Quote: "Ramp (~~$1.4B ARR) and Decagon (~~$35M ARR) bracketing the group at the high and low end"

Anthropic

Description: AI safety and large language model company
Why mentioned: Used as a case study in the hype-vs-built analysis; founded in 2021 while crypto dominated discourse
Quote: "Anthropic (2021) was founded while crypto consumed the internet"

Airbnb

Description: Short-term rental marketplace
Why mentioned: Case study showing that the biggest company of a given year rarely comes from the hyped topic of that year
Quote: "Airbnb (2008) launched while Google dominated the conversation"

Uber

Description: Ride-sharing and delivery platform
Why mentioned: Same hype-vs-built case study
Quote: "Uber (2009) arrived amid low-level programming chatter"

Exa

Description: AI-native search engine / web data infrastructure
Why mentioned: Newsletter sponsor; described as turning "the web into live, structured data that your agents and pipelines can act on"
Quote: "AI-forward firms use Exa to source companies by thesis, enrich target businesses programmatically, and monitor signals across the market in real time"

4. People Identified

Peter Walker

Description: Data analyst at Carta
Why mentioned: Author of the fund-level power law analysis using 400+ US venture funds from 2016–2018 vintages
Quote: "Peter Walker at Carta shared new data on 400+ US venture funds from the 2016, 2017, and 2018 vintages, illustrating how extreme the power law in VC really is"

Konstantine Buhler

Description: Partner at Sequoia Capital
Why mentioned: Conducted the Hacker News hype-vs-built ranking analysis spanning 2007 to present
Quote: "Sequoia's Konstantine Buhler ranked the most hyped topics on Hacker News for every year since 2007 and mapped each year against the most valuable company actually founded that year"

Ilya Strebulaev

Description: Professor at Stanford GSB
Why mentioned: Conducted the "jockey or horse" VC decision-making study using surveys of 1,000+ VCs and tracking of 50 companies from business plan to IPO
Quote: "Stanford GSB's Ilya Strebulaev revisits the 'jockey or horse' debate using survey data from over 1,000 VCs plus a study tracking 50 VC-backed companies from business plan to IPO"

Tomasz Tunguz

Description: Founder and General Partner at Theory Ventures
Why mentioned: Author of the ARR multiple analysis for AI harness companies
Quote: "Tomasz Tunguz asks how the market values an 'AI harness', using Harvey, Legora, Sierra, Ramp, and Decagon as the data points"

Yohei Nakajima

Description: VC and AI practitioner (known for BabyAGI)
Why mentioned: Shared his personal AI operating stack as a real-world case study of agentic VC workflow
Quote: "Yohei Nakajima posted his personal AI operating setup, a hardware-plus-agent stack that lets him work on almost anything from anywhere, hands-free"

Fabrizio Serafini, Seema Amble, and Eric Zhou

Description: Partners/analysts at a16z
Why mentioned: Co-authored the survey of computer-use agent deployments tracking one year of progress
Quote: "Fabrizio Serafini, Seema Amble, and Eric Zhou at a16z surveyed a year of computer-use deployments and found the frontier has shifted from clicking the right button to reliably doing the job"

Andre Retterath

Description: Author of Data Driven VC newsletter
Why mentioned: Newsletter author and curator of all insights in this issue
Quote: "Hi, I'm Andre and welcome to my newsletter Data Driven VC which is all about becoming a better investor with data and AI"

5. Operating Insights

Operating Insight 1: Structure Your AI Workflow as a Hierarchical Agent System Organized by GitHub Repos

Yohei Nakajima's personal operating stack offers a concrete blueprint for replacing a SaaS tool stack with an agentic hierarchy. The key structural insight is that roles (chief of staff, manager, builder) are model-agnostic and projects are portable via repos.

"Each project (work, research, or personal) lives in its own GitHub repo holding its instructions and knowledge, and any project can itself manage other projects by spinning up a new repo underneath it... Chief of staff, manager, and builder are role definitions that Claude, ChatGPT, Codex, or Claude Code can each execute, which is what lets a task started on one AI resume on another without losing context."

Operating Insight 2: When Evaluating AI Agent Startups, Prioritize Depth of Captured Workflow Knowledge Over Model Performance

For operators building in the computer-use agent space, the strategic imperative is to invest in capturing proprietary customer workflow data early — before model capability becomes a commodity selling point.

"For a VC evaluating a computer-use agent startup, the thing to check is how much company-specific operational knowledge the product has already captured, since that's what determines defensibility once the underlying model becomes a commodity."

6. Overlooked Insights

Overlooked Insight 1: VC Fund Contract Terms Are Not Currently Structured for Founder Turnover Risk

The "jockey or horse" analysis surfaces a structural gap in how VC funds design investment terms. The newsletter briefly flags that rising founder-centric bets create a mismatch with observed IPO-stage CEO turnover — but doesn't fully unpack the term-sheet implications.

"Founders raising earlier pushes team weight up, likely explaining recent team-first wins. It also raises the odds of a mismatch between the founding team and what the company later needs, showing up as turnover or contract terms built for that risk. Think now about what scenario you can foresee, and whether your fund is prepared for it."

Overlooked Insight 2: The Hacker News Hype Signal Has a Dark Side — Recent Hyped Topics Have No Confirmed Winner Yet

The analysis is often read as a sourcing tool for spotting early signals. But the newsletter quietly buries a cautionary note that the most recent hyped categories have not yet produced a clear breakout company — meaning the framework is retrospectively clean but prospectively uncertain.

"Recent years' 'hyped' topics haven't produced a clear winner yet, per Sequoia's own data, so early attention alone isn't enough to pick the outcome."