7 min read

A $5,000 Agent Took Vercel From 10 SDRs to 1

Vercel's COO on the $5,000-a-year agent that took its SDR team from 10 to 1. Plus why 94% of Bending Spoons' code is AI-written and what Anthropic's S-1 shows.

A $5,000 Agent Took Vercel From 10 SDRs to 1

Compounders like Bending Spoons and Vercel are shifting the operational baseline from AI-assisted to AI-native, replacing entire human workflows with $5,000-a-year agents that consistently out-execute 90th-percentile specialists.


The Intake

📊 10 episodes across 8 podcasts

⏱ 10 hours 6 minutes of intelligence analyzed

🎙 Featuring: Harry Stebbings, Erik Bernhardsson, Anish Acharya, Luca Ferrari, David Tisch, Jeanne DeWitt Grosser, Andrew Huberman, Oliver Cameron, Bill Perkins, and Hector Serrano.

🆕 348 emerging themes | 20 surprising insights | 1 debate detected


Presented by

Velocity Road

Turn AI spend into EBITDA—with proof every quarter.

Velocity Road finds where AI actually pays, installs the governance your board can trust, and builds the systems that move the number. Run all year. Proven every quarter.

Learn More

The Big Shift

The "Unreasonable" Efficiency of the AI-First Operating Stack

We have moved past the pilot phase of AI. The signal this week from top-tier operators is that AI is no longer a tool for augmentation—it is becoming the baseline architecture for the "compounder" business model. At Bending Spoons, CEO Luca Ferrari reveals that 94% of the company's code is now AI-written, and 90% of data processing is handled without human intervention. This isn't about marginal gains; it's about a structural shift in how tech companies generate free cash flow.

Why it matters: The "SaaS moat" is being redrawn. Ferrari argues that while AI makes replicating software cheaper than ever, the real defense lies in distribution, habits, and trust. Meanwhile, Vercel COO Jeanne DeWitt Grosser demonstrated the brutal math of this shift: their lead-qualification agent costs less than $5,000 annually and took its SDR team from 10 people to 1 while maintaining 90th-percentile performance. The new competitive bar for growth-stage companies is no longer just "using AI"—it’s achieving this level of radical capital efficiency.

"The lead qualification agent... costs us under $5,000 a year between infrastructure and tokens... The architecture reflects exactly what [the rep] was doing manually, but now it performs like the 90th percentile rep 100% of the time."
— Jeanne DeWitt Grosser, COO of Vercel on The Official SaaStr Podcast

The move: Audit "human-intensive" workflows like SDRs, support, and routine data cleaning against the $5k/year agent benchmark. Companies failing to redeploy that headcount into creative or high-leverage roles within 12 months leave their margins exposed to more efficient competitors.


The Rundown

① The Death of "High Ownership" Dogma in Seed.

David Tisch (BoxGroup) argues that the obsession with 15-20% ownership at seed is outdated. Successful later rounds now see only 3-8% dilution, allowing broad portfolios to return funds. (David Tisch on 20VC)

→ The signal: Competitive seed rounds are becoming "access-first," and managers who pass on generational founders just to hit an ownership percentage are suffering from adverse selection.

② Performance-Based UGC as a Scalable Growth Lever.

🆕Hector Serrano (Wispr Flow) details a shift from vanity-reach organic creator posts with 500M views and low conversion to performance-based ads. These use "smart retainers" that pay creators up to $13k/month. (Hector Serrano on Lenny's Podcast)

→ The signal: Creative quality is audience-relative; for technical users, "ugly" screen recordings often outperform polished influencer content.

③ Anthropic’s S-1 Reveals the "Inference Gap."

The filing shows $16.3B in H1 2026 revenue against an $80B net loss, driven by future compute commitments and debt. (TechCrunch Team on Equity)

→ What to watch: The disclosed risk that AI liabilities could exceed the entire capacity of the global insurance industry, creating an "uninsurable" risk category for frontier models.

④ The End of 90% Software Gross Margins.

Erik Bernhardsson (Modal) notes that the AI stack captures 20-30% at each layer, effectively capping margins as compute costs flow through every vendor. (Erik Bernhardsson on 20VC)

→ Why it matters: Investors must reset expectations for SaaS profitability; the "marginal cost of $0" no longer applies to AI-heavy applications.

⑤ World Models as the New Robotics Foundation.

Oliver Cameron (Odyssey) is moving away from task-specific robot training toward "world models" that learn physics from video, requiring only 20-30 hours of fine-tuning for new tasks. (Oliver Cameron on This Week in Startups)

→ The signal: 🆕Odyssey has raised $370M to build simulated training environments, betting that general physical intelligence will solve the "brittleness" of current autonomous systems.


Signal Board

🔥 HEATING UP

• Modal Sandboxes: Now a $300M run-rate business as coding models improve and demand for ephemeral environments spikes. (Erik Bernhardsson on 20VC)

• Bending Spoons: The "compounder" model involves acquiring established apps like Eventbrite and automating them to 2.5x EBITDA leverage. (Luca Ferrari on a16z Show)

• Industrial Autonomy: 🆕Bedrock Robotics is raising $600M to move from pilots to commercial excavator operations. (Sean O’Kane on Equity)

👀 ON WATCH

• Orbital Compute: Google is testing "Project Suncatcher" to explore if Starship launch cadences make space-based data centers viable. (Jason Calacanis on TWiST)

• Circular AI Financing: Concerns are rising that Nvidia’s demand is artificially propped up by investing in its own customers. (Scott Galloway on Pivot)

• Post-Training RL Limits: Effectiveness is currently capped by "feedback loop" duration. Tasks taking longer than a week remain hard to automate. (Erik Bernhardsson on 20VC)

❄️ COOLING OFF

• SDR Headcount: Vercel’s 10-to-1 reduction is the new blueprint for GTM engineering. (Jeanne DeWitt Grosser on SaaStr)

• Organic UGC: Shift from "vanity metrics" and viral views toward performance-tracked creator commissions. (Hector Serrano on Lenny's Podcast)

• The "Lean Startup" Label: David Tisch argues the doctrine is a "fake" for venture-scale companies that require aggressive capitalization to win markets. (David Tisch on 20VC)


The Debate

Does AI Optimization Erode the Human Experience?

AI mediation of parenting and social interactions is creating tension between "efficiency" and "intimacy."

🐂 The bull case: Agents like 🆕Dazzle or Miso AI save time on administrative parenting and research, reserving human focus for direct connection. As Jeanne DeWitt Grosser (Vercel) notes, automation allows people to be "redeployed into more valuable roles."

🐻 The bear case: Amanda Silberling (TechCrunch) argues executives are "out of touch," turning social behavior into an optimization problem. Using AI to describe one's own child means "basic human interaction gets more and more optimized out of actually occurring."

Our read: The market currently prioritizes the "efficiency" side of the ledger, but consumer backlash against "AI-mediated intimacy" is a latent brand risk for startups that touch the parent-child or dating categories.


The Bottom Line

The winners in "SaaS 2.0" will be companies with the most automated, high-margin operating stacks rather than simply the best models.


Episode Guide (Web Version)

The Twenty Minute VC — "Erik Bernhardsson, Modal CEO: Who Wins the Inference Market"

Runtime: 62 min | Host: Harry Stebbings | Guest: Erik Bernhardsson (CEO, Modal)

Best for: Infrastructure investors and CTOs navigating GPU capacity planning.

Bernhardsson breaks down the economics of the "new cloud." He argues that inference will fragment into specialized products and explains how Modal pools demand across 25 neoclouds to offer on-demand scale without multi-year commitments.

"I think inference is going to be an enormously large market. It's like the new cloud basically." — Erik Bernhardsson

▶ Listen · Apple Podcasts

The a16z Show — "How Bending Spoons Built a $20 Billion Compounder"

Runtime: 44 min | Host: Anish Acharya | Guest: Luca Ferrari (CEO, Bending Spoons)

Best for: Growth equity and PE partners looking for the next-gen "Constellation Software" model.

Luca Ferrari details the Bending Spoons strategy: acquiring established tech, moving them to a shared infra stack, and utilizing AI automation (94% of code) to drive free cash flow for reinvestment.

"We buy technology companies where we see an opportunity to make them a lot more successful... try to transform them so they're a lot more cash generative." — Luca Ferrari

▶ Listen · Apple Podcasts

The Official SaaStr Podcast — "Going from 10 to 1 SDRs with Vercel's COO"

Runtime: 22 min | Host: SaaStr | Guest: Jeanne DeWitt Grosser (COO, Vercel)

Best for: COOs and Sales Leaders looking to gut-renovate their GTM costs.

A tactical breakdown of how Vercel built a "GTM engineering" team to replace 90% of their SDR function with a $5k/year agent. Essential listening for anyone managing a large sales org.

▶ Listen · Apple Podcasts

The Twenty Minute VC — "David Tisch on Why High Ownership is BS"

Runtime: 77 min | Host: Harry Stebbings | Guest: David Tisch (BoxGroup)

Best for: Emerging managers and LPs evaluating seed fund construction.

Tisch challenges the venture status quo, arguing that the "20% ownership" rule is dead. He shares how Cursor returned his $275M fund multiple times despite a lower-ownership entry.

▶ Listen · Apple Podcasts

Lenny's Podcast — "How a 19-year-old intern built the growth engine behind Wispr Flow"

Runtime: 88 min | Host: Lenny Rachitsky | Guest: Hector Serrano (Wispr Flow)

Best for: Early-stage founders struggling with customer acquisition costs.

Serrano outlines the mechanics of "Paid UGC," explaining how to recruit creators and coach them to build ads optimized for conversion over viral reach.

▶ Listen · Apple Podcasts

This Week in Startups — "Odyssey CEO Oliver Cameron on World Models"

Runtime: 84 min | Host: Jason Calacanis | Guest: Oliver Cameron (Odyssey)

Best for: AI researchers and robotics founders.

Explores the shift from task-specific robotics to "world models" that learn general physics, followed by an analysis of simulation theory.

▶ Listen · Apple Podcasts

Pivot — "Trump's War on AI and Google's Meltdown"

Runtime: 61 min | Host: Kara Swisher | Guest: Scott Galloway

Best for: General partners tracking the intersection of tech policy and public sentiment.

A breakdown of Microsoft/Alphabet earnings, the "circular financing" risk in the GPU market, and Elon Musk's geopolitical influence.

▶ Listen · Apple Podcasts

Equity — "Anthropic’s S-1 is here. Does AI pay off?"

Runtime: 36 min | Host: Kirsten Korosec | Guests: TechCrunch Team

Best for: Late-stage investors and public market analysts.

An autopsy of the Anthropic S-1, detailing the disconnect between enterprise growth and escalating compute liabilities.

▶ Listen · Apple Podcasts

The a16z Show — "Andrew Huberman’s Three Rules for Focus"

Runtime: 49 min | Host: Daisy Wolf | Guest: Andrew Huberman

Best for: Stretched CEOs looking for high-ROI health protocols.

Huberman distills health into three non-negotiables: morning light, low-volume strength training, and sleep. He specifically addresses founder burnout and focus windows.

▶ Listen · Apple Podcasts

My First Million — "$100M hedge fund manager: Do not make the mistake of dying rich"

Runtime: 83 min | Hosts: Shaan Puri, Sam Parr | Guest: Bill Perkins

Best for: High-net-worth founders thinking about "the exit" and life after liquidity.

Perkins argues founders should target a specific date for peak net worth instead of a dollar figure, noting that ego drives risk aversion more than financial downside.

▶ Listen · Apple Podcasts

PARTNER

Not sure where AI fits in your operations? Start with the data.

Velocity Road's AI Readiness Assessment maps your organization against 7 operational dimensions and shows exactly where AI creates ROI — in under 10 minutes.

Take the Assessment -> →

Avi Savar

Get Business Intelligence in your inbox

Deal trends, operational insights, market conditions, exit strategies. Free.