Directions Brain

2026-05-16-alonhuri-linkedin-ai-native-growth-hacking

5 min read

Alon Huri — Marketing is the biggest problem AI hasn’t solved. How Growth Hacking should look in the AI era (May 2026)

Source: raw/articles/2026-05-16-alonhuri-linkedin-ai-native-growth-hacking.md

Note: LinkedIn post by alon-huri (Managing Partner at team8); ~2 weeks before the 2026-05-30 capture, dated 2026-05-16. Hebrew original; load-bearing claims translated to English here, key Hebrew phrasing preserved verbatim in ## Notable quotes.

Summary

Alon argues that AI has solved nearly every classical startup bottleneck (code, product, ops, hiring, data analysis) over the last three years — except customer acquisition. CAC is up 222% in eight years; average startups spend 1 of ARR. AI handles the operational layer of marketing well (copy, visuals, bidding, A/B tests, landing pages, segmentation — all commoditized) but cannot do the strategic layer (identifying the real customer, understanding pain below the surface, finding new channels, writing the promise that makes someone stop). His thesis: classical growth hacking (one artisan, serial hacks, manual loop) must become an autonomous closed-loop growth lab running ~50 hacks in parallel, with agents at every step except strategic judgment. Founders building today must put this at the GTM core from day 1 — companies that build it first will take the market in any vertical they enter.

Key claims

  • AI has solved most startup bottlenecks except customer acquisition. Code, product, ops, hiring, data analysis — all commoditized. Marketing/customer-acquisition is the holdout.
  • CAC trajectory. Average startups now spend 1 of ARR. CAC up 222% over the last 8 years. The gap is widening.
  • Operational vs strategic split. AI is excellent at the operational layer of marketing (copy generation, visuals, bidding, A/B testing, landing pages, segmentation) — those are commodities. AI is bad at the strategic layer: identifying the real customer, understanding pain below the surface, discovering new channels, writing the promise that makes someone stop scrolling. AI is an optimization engine, not a discovery engine, not a growth-cracking engine.
  • The shift in growth-hacker role. From “single artisan running serial hacks” to “architect of a system.” Growth hacking shifts from human-led series of experiments → autonomous closed loop running on ~50 hacks in parallel.
  • The closed loop, agent by agent.
    • Research agent — heart of the system. Knows ICP, funnels, landing pages, all past experiment results, what worked, what didn’t. Continuously hunts for new growth hacks fitting the problem; learns from competitors, what hacks they run, what works for them, what doesn’t, draws inspiration into new test proposals.
    • Judgment (human or agent) — filters relevance to ICP and problem-world.
    • Thesis agent — turns each hack into a structured hypothesis with explicit success/failure criteria.
    • Execution agents — build all the assets (landing pages, creative, audiences) and run dozens of tests in parallel.
    • Measurement agent — tracks every lead from click to customer.
    • Learning agent — feeds insights back into the research agent. Loop never stops.
  • Concrete example. An execution agent that scans Reddit continuously, identifies posts describing exactly the customer’s pain, replies with authentic value in authentic language. One of 50 such hacks running in parallel inside a “growth lab.”
  • Time to crack a new channel: months → weeks or days.
  • Founder GTM mandate. “For founders building a new company today, this isn’t a feature you add later. It has to be the core of how you build your GTM from day one. Companies that build this system first will take the market in every vertical they enter.”

Notable quotes

  • “השיווק הוא הבעיה הכי גדולה ש-AI עוד לא פתר.” — Marketing is the biggest problem AI hasn’t yet solved.
  • “היום סטארטאפים ממוצעים מוציאים 2 דולר על שיווק כדי להביא דולר אחד של ARR. ה-CAC קפץ ב-222 אחוז ב-8 השנים האחרונות.” — Today, average startups spend 1 of ARR. CAC has jumped 222% in the last 8 years.
  • “ה-AI הוא מנוע אופטימיזציה, לא מנוע גילוי ולא מנוע לפיצוח צמיחה.” — AI is an optimization engine, not a discovery engine and not a growth-cracking engine.
  • “ה-Growth Hacker בעידן AI צריך להפוך מאומן יחיד לארכיטקט של מערכת.” — The Growth Hacker in the AI era must shift from a single artisan to a system architect.
  • “ליזמים שבונים חברה חדשה היום, זה לא פיצ’ר שתוסיפו אחר כך. זה חייב להיות הליבה של איך אתם בונים את ה-GTM מיום ראשון.” — For founders building a new company today, this isn’t a feature you add later. It has to be the core of how you build your GTM from day one.

Why this matters for the team

  • Direct extension of agent-native-go-to-market. Alon’s pipeline is the GTM equivalent of the team’s agent-native-context-OS thesis: agents at every node of a workflow loop, humans only at strategic-judgment seams. Where the team’s framing emphasized agents consuming the Brain’s context, Alon’s emphasizes agents generating growth experiments — the same pattern, different layer of the company.
  • Strong tailwind for wonderful-style verticals. wonderful reportedly built an internal growth-loop pattern that resembles what Alon describes; the “Wonderful for X” pattern in vertical-use-case-led-brain inherits this advantage. Validates the team’s bet that vertical AI products plus AI-native growth labs is a defensible combination.
  • Open question for the Brain. Alon’s “research agent at the heart” needs deep, continuously-updated context — exactly the kind of structured corpus the Brain produces. Worth asking: does a vertical wedge include the GTM growth-loop as part of its product, or is the growth-loop a separate product? Either way, the Brain underneath both.
  • Skepticism caveat. Same as 2026-05-09-alonhuri-linkedin-ai-native-company: Alon is Managing Partner at team8; thesis aligns suspiciously well with the team’s direction. Treat as advocacy, not neutral analysis.

Mentioned in

Graph View

Backlinks