Who he is
Affiliation: Founder & CEO, Uppit AI LLC (his education company; Skool profile headline "Founder & CEO @ Uppit AI," read 2026-09-21) and founder of AI Automation Society (a Skool community, Oct 2024–). Role check run 2026-09-21: he co-founded TrueHorizon AI (an enterprise AI-agent agency) in January 2025 with two co-founders and exited in December 2025 "to refocus on education and community" — his own account; TrueHorizon's current About page lists founders Milan Tahliani and Tyler Baughcome and does not list him. Before that: Business Intelligence Analyst, Goldman Sachs (Salt Lake City; dashboards and Alteryx automations); University of Iowa Tippie, class of 2024, Business Analytics + Marketing. Full name Nate Herkelman; 24 (born April 2002); Chicago.
Scale (his numbers unless noted): YouTube 1M+ subscribers (independent snapshots: 968K on 2026-09-01 with +88K in the prior 30 days; 1M, 512 videos, 53.8M views in Sep 2026); AI Automation Society 459.9k free members (live count 2026-09-21); AI Automation Society Plus 3.5k members at $129/month; 70K+ LinkedIn followers; a book, Becoming AI Native (~July 2026).
One-line position: "AI automation made easy" — a person who cannot code can turn AI workflows into income by climbing a ladder (freelance modules → AI consulting → agency → teaching), and as of 2026 the tool for every rung is a coding agent (Claude Code or Codex) run as a personal "AI operating system."
Provenance: captured Aug 13, 2026 from a text from Mark: "he helps people with our kind of experience create AI consulting companies." The v1 page was that capture stub — no claims, no review, no affiliation. This is the first real look.
Discipline & technical bet
Creator-economy educator — not a researcher, and no longer an operator. His discipline is packaging. Two technical bets, in sequence. 2024–25: n8n (an open-source, drag-and-drop workflow tool) as the substrate for "AI agents without code" — the n8n Masterclass, "AI Personal Assistant 2.0," the template kits the Microsoft Learn page lists (2025-02). 2026: coding agents as an operating system. Since May 2026 his flagship is a "Claude Code OS" — a folder of context files, connections and scheduled skills that "runs" his business — formalized as the Four Cs (Context, Connections, Capabilities, Cadence) and the Three Ms (Mindset, Method, Machine), and shipped as a free MIT-licensed starter kit, nateherkai/AIS-OS (1.4k stars, 420 forks, read 2026-09-21). The kit's design rules are a real technical stance: workflows over agents (fixed steps first, model judgment only where needed), simplicity, safety guardrails. He says the pattern is tool-agnostic — he moved the whole system to "a different coding environment" in about two minutes — and his late-August build (SnagTime, a free Calendly clone) used Codex, not Claude. So the wager is on the harness (the scaffolding around a model), not on any vendor. Trademarking both frameworks ("The Three Ms of AI™," "The Four Cs of an AI OS™," © 2026) says what he thinks the moat is: the vocabulary.
Key claims (Says)
- v1 framing (Mark, 2026-08-13): "helps people with enterprise/consulting experience create AI consulting companies." Graded in two halves — mechanism and audience separately. Held — mechanism: the paid tier does teach building an AI automation agency — courses "One Person AI Agency" and "Get Clients from YouTube," a promise of "$5,000 – $20,000 per month" client contracts, content unlocks at 30 days ("first client") and 90 days ("scale") (Plus about page, read 2026-09-21; aifunnelinsider, updated 2026-09-08). Wrong — audience: his stated mission is "non-technical people who want leverage," the Plus pitch is "starting from zero," and his bio is University of Iowa 2024 → one analyst job → YouTube; there is no consulting background and no senior-operator track (about page; Microsoft Learn bio, 2025-02). The "our kind of experience" in Mark's text is what Nate lacks, not what he serves.
- Empirical: you don't need to code — no-code / low-code is enough to build automations clients will pay for. Drifted: by mid-2026 he "spends most of his time in Claude Code" (a terminal coding agent), the flagship courses are "Build & Sell Claude Code Operating Systems" (2+ hours) and "Build & Sell with Claude Code" (10+ hours), and the recent-uploads list in the 2026-09-01 CreatorDB snapshot is Claude Code / Codex end to end — a website-design skill, a Deepseek-harness-vs-Claude-Code test, shared memory for ChatGPT and Claude, the Calendly clone. The pitch is now "no coding experience required" while living inside a coding tool — a real shift the phrase papers over. His short-lived "Claude Code Kickstart" challenge promised to take members "from Terminal Error to AI-Native n8n Builder," which is precisely the bridge no-code was supposed not to need.
- Empirical: businesses pay $5K–$20K a month in retainers (a fixed monthly fee for ongoing work) for AI automation; charge "$5,000 just for the AI business audit, then make another $20,000 to $100,000 implementing" (Plus page; 4-stage blueprint, 2025-12-21). Unverified — not graded: the only client-side evidence is TrueHorizon's Topgolf case study (5,000+ calls a month handled, missed calls 60% → 0%, ~75% resolved without a human; no fees stated) and a third-party write-up of a $6K-a-month YouTube-strategist retainer; a due-diligence roster that sampled Uppit AI's own site found "delivery evidence is thin" — three testimonials, no metrics, no case studies (verified 2026-07-26). Check under predictions.
- Empirical: "I scaled my agency past $100,000 a month, exited it" (Skool post, 2026-08-11; about page). Held — the exit: TrueHorizon exists, lists Topgolf and BarkBox as customers, and no longer lists him (About page, 2026-09-21); the roster records the Jan–Dec 2025 tenure "by his own account." Revenue unverified: no filing, no third-party figure; the $100K a month is self-reported and — by his own thread — was a team's number: "in order for me to achieve that scale, I built a team around me" (X, 2026-08-11).
- Empirical: his own business does "$1 million per year" (Plus page, agency framing) → "$3M/yr" (X, 2026-05-01) → "$5M+ per year" (free-community page, live 2026-09-21). Drifted: the thing being measured moved from the agency to the education business without the page copy saying so. The top number is arithmetically plausible — 3.5k Plus members × $129 × 12 ≈ $5.4M gross before churn and annual discounts — which is exactly the reviewers' point: the income offered as proof of the agency model is subscription income from teaching the agency model (scamrisk, 2026-01-22). A "$231,000 in 30 days as an AI consultant" figure circulates in reviews; the original video or ad was not located this cycle and is not credited here.
- Empirical: "87 out of every 100 AI projects die in the demo phase and never make a dime" (Skool post, 2026-08-11); IBM's 2026 CEO study puts regular worker AI use at 25% against 86% of CEOs who believe their teams are ready — "a 61-point gap" he calls the business opportunity (X article, 2026-08-11). Not graded — the primary studies were not fetched this cycle; the pilot-failure figure matches the 88% in Rich's own 8.1 abstract, so it is at least the consensus number.
- Normative: sell outcomes, not AI features; route every project into one of three business buckets; walk clients from education → audit → project → retainer; build the system for yourself before selling it; get five real conversations before forcing a niche (X article, 2026-08-11). Method, not prediction — noted, not graded; this is the slice of his material with the most transferable content for Rich.
- Credentials he cites: "4-time Skool Games winner" — Held: Q1 and Q4 2025 wins confirmed by his own community posts, the Q4 post saying "all 4 quarters of the year." "Featured by Forbes / Microsoft" — Held: a Forbes small-business piece names him at 650K subscribers (2026-04-24); the Microsoft Learn "Learn With Nate Herkelman" creator page exists (updated 2025-02-06). "Verified n8n Expert Partner" — unverifiable: n8n's expert-partners page names no partners (a closed pilot) and the roster found no directory entry for Uppit AI (2026-07-26); TrueHorizon lists n8n among its logos, so the credential may belong to the agency he left. "#1 Amazon bestseller in four categories" — unverified: category badges are hourly snapshots and the Amazon listing did not render this cycle; plausible, not confirmed.
Notable predictions — with falsifiable checks
- (2025-12) An "AI partner business" (agency) can "easily hit over $1 million in annual revenue" and scale to "$10 million per year or more" (4-stage blueprint). Live, unconfirmed. Check: by 2027-06-30, at least one AAS+ member business documented publicly with a named client and $1M+ annual revenue, not counting Nate's own; none found this cycle → Wrong at that review.
- (2025-12) Skool post: "pricing for AIS+ will increase in 2026." Held — $99 → $129 a month (Plus page, live 2026-09-21; aifunnelinsider, updated 2026-09-08).
- (2026-05) A personal AI operating system is "something anyone willing to put in two weeks of learning can build for themselves" (thusie, 2026-05-23). Live. Check: by 2027-01-31, AIS-OS at 3k+ stars with commits after 2026-10 and the 7-day-challenge certification still running → Held; repo stalled or challenge retired → Wrong.
- (2026-08) The one-person AI business — "just you, a computer, and Claude Code" — is the entry model, and the relationships "later evolve to $20,000" retainers (X, 2026-08-11). Live. Check: by 2027-03-31, does his funnel still teach this ladder unchanged, and has he published any member retainer count? A second tool pivot before then grades it Drifted.
- (2024–25, implicit) n8n is the substrate; learn n8n. Drifted — the 2026 curriculum and uploads moved to coding agents (Class Central lists "Build Anything with Claude Code and n8n — Beginner's Guide"; CreatorDB recent titles, 2026-09-01). Check: if n8n-titled uploads fall below one in five by 2027-01-31, log as Wrong for anyone who bought the n8n-first thesis in 2025.
Revealed behavior (Does)
- Time: quit Goldman two months after his first upload (Sep → Nov 2024; Apify, 2025-07-31) — the bet was on audience, not consulting. Cadence since: ~5.6 uploads a week, 500+ videos in 24 months, 4.8k Skool contributions, "active 8 hours ago" on the day of this review.
- Money: co-founded the agency (Jan 2025), left it after 11 months (Dec 2025), kept the education business. He teaches the model he exited; the revealed preference is the "10/10 income potential" rung of his own ladder — teaching — where he claims "over $150,000 per month in pure profit" (blueprint, 2025-12-21).
- Giveaways as acquisition: a 2+ hour free course plus the AIS-OS repo (2026-05-01), a 10+ hour "Build & Sell with Claude Code" course, a 7-day challenge with certification, 100+ n8n templates, a free Calendly clone with its repo (~2026-08-28). Free tier 459.9k → paid 3.5k, roughly a 0.8% conversion.
- Pricing: raised Plus from $99 to $129 with one week's notice, announced in the same post as the Q4 Skool Games win (2025-12).
- Kill fast: "Claude Code Kickstart" (525 members) launched and archived within 2026; its content resurfaced as a "$49-value" bonus on the $4.99 book offer.
- Hires: 16 admins run the free community; the agency had "a team around me." No research or engineering bench in public; the roster found no case studies on Uppit AI's own site.
- IP: trademark marks on the Three Ms and Four Cs (2026) — the first sign of moat anxiety in a business that gives everything else away.
Feels
Wants to be the person who made AI "understandable and profitable" for people who don't code — the mission line on his site — and to prove the Goldman exit was right. Fears being the shovel-seller whose gold rush moves: the 2026 sprint from n8n to Claude Code, and the trademarks, read as someone watching the ground shift under a 460K-person audience.
Hears
Skool's creator leaderboard and its community-growth playbook; the n8n and Claude Code release cycles; and, at a scale no consultant has, the daily questions of 460K beginners. Almost no academic, policy or enterprise-strategy input — TrueHorizon was the enterprise channel, and he left it.
Sees
What non-technical people actually get stuck on when they try to automate work, in volume and in real time; what sells on YouTube in AI (3.4% engagement against a 1.5% category median); and, from one year inside an agency, what a mid-market client will sign for. He does not see a transformation from inside a large organization, and his numbers show what packaging does, not what delivery does.
Incentive map
Re-derived this cycle: maximal commercial conflict, and structurally circular. He is paid by subscriptions (up to ~$450K a month gross at current Plus numbers, less for annual members), the book, YouTube and affiliates; the product is belief that the agency model works, sold mostly to people who will never build one. He cannot say: "most members won't land a client," "the one-person-agency market is saturating" (460K members chasing the same HVAC firms and real-estate offices — ippei's critique, 2026-05-20), "n8n is losing to coding agents" (he said it by action, not by words), or "the $100K a month was a team's number and I left." He also depends on two platforms (Skool, YouTube) and two vendors' roadmaps (n8n, Anthropic/OpenAI) he does not control, and holds a Microsoft "Learn With" creator relationship. Rule-25 discount: maximal on every revenue, outcome and "what clients pay" claim; low on the how-to content, which anyone can verify by running it.
Theories aligned with
- It takes intelligence to use intelligence — he is the live counter-test: he says access plus "two weeks of learning" is enough; the theory says wielding is the bottleneck. His own funnel is the data point — 459.9k free, 3.5k paid, a handful of named success stories.
- Augmentation thesis — the rhetoric ("AI gets the grunt. You keep the talent.") is augmentation; the sales line he ran at TrueHorizon, "scale without increasing headcount," is displacement at the hiring margin — the entry-level effect, sold as a feature. Mechanism and valence point in different directions; both logged.
- Organizational singularity — the one-person business on an AI OS is a bet that the smallest organizational form is what survives the transition.
What he's reacting against
- Big consulting: "McKinsey, BCG, and Accenture charge hundreds of thousands of dollars just for an AI consultation or audit" (blueprint, 2025-12) — his audit-then-implement ladder is priced as the underbid.
- Demo-phase AI: "87 out of every 100 AI projects die in the demo phase" — the pitch is shipped outcomes proved "with a number."
- Coder gatekeeping — "without any coding experience," in every bio since 2025.
- Passive AI use — "go from watching AI videos to building your own AI Operating System" (free-community page).
- The corporate ladder he left — the Goldman-analyst-to-creator story is the origin myth in every profile.
Where he overlaps / splits (with Rich)
Carried from v1 — why Mark flagged him: directly relevant to Agency Lab (consulting for betterment using AI), the AI Coaching Practice ($300/hr positioning), Fractional Leadership positioning, and the "leave the company" exit strategy.
- Overlaps — the ladder. Rich's Agency Lab flagship, the Fractional AI Operations Officer (Diagnostic Sprint → 90-day Pilot-to-Production → retainer), is structurally Nate's "education → audit → project → retainer," and both anchor on the same failure rate (Rich 88%, Nate 87%). Both sell a codified method over a résumé.
- Splits — altitude and price. Rich's offer is embedded executive leadership for $50M–$500M firms; Nate's is one-person n8n / Claude Code builds for HVAC companies, real-estate agencies and coaches, sold to a $129-a-month audience. Nate is downstream of Rich's kind of experience, not a mentor to it: usable for tooling, funnel mechanics and offer packaging; not for positioning.
- Splits — the exit. Nate's "leave the company" story is a 22-year-old with a two-month YouTube runway; Rich's is a semi-retirement income stack with a MAGI cliff. Same verb, different physics.
- Overlaps with Mollick on practical AI for non-technical workers; splits on evidence — Mollick runs studies, Nate runs a funnel.
- Overlaps with Ng on democratization through education; splits on depth and on selling outcomes.
- Downstream of Karpathy — the n8n → Claude Code pivot is the "Software 3.0 / vibe-coding" shift arriving in the creator tier a year later.
- Unwitting evidence for Frey, against Brynjolfsson's early optimism — "scale without increasing headcount" is the hiring-margin displacement the Canaries payroll data now measures, sold as a product.
- Splits with Zitron — a 460K-member community paying to learn to sell AI to each other is, in Zitron's frame, a bubble artifact; Nate's counter is the Topgolf call numbers.
- Same form as Shapiro — YouTube-native AI voice monetized by community; different content (Shapiro theorizes, Herk ships templates).
Track record
- Audience: first video Sep 2024 → 230K (Jun 2025, Apify) → 500K+ (Feb 2026, Atak) → 650K (Apr 2026, Forbes) → 968K (2026-09-01, CreatorDB) → 1M+ (his site, Sep 2026). Verified trajectory; Atak's "fastest-growing AI creator on YouTube" is plausible on these numbers.
- Skool Games: four quarterly wins in 2025, Q1 and Q4 confirmed by his own posts.
- TrueHorizon AI (Jan–Dec 2025): a real company with real enterprise logos (Topgolf, BarkBox, Avalara, JP Morgan on its site) and case studies carrying operational metrics but no fees; his exit after 11 months.
- Uppit AI LLC: "delivery evidence is thin" — three testimonials, no metrics, no case studies (roster, 2026-07-26).
- Book: Becoming AI Native, ~July 2026 ("just arrived," community post 2026-07-20); $4.99 promo / $19.99 list; bestseller claim unverified.
- Claude Code Kickstart: launched, 525 members, archived (2026).
- Pattern: exceptional at audience and packaging; unproven at durable delivery; one product killed and one company exited inside 24 months; every revenue figure self-reported.
Empirical vs normative
- Empirical: most AI projects die at demo; non-coders can ship workflows with n8n and now with coding agents; mid-market clients sign retainers for automation; the substrate is moving from low-code to coding agents.
- Normative: everyone should build their own AI OS first; sell outcomes, not features; workflows over agents; five real conversations before a niche; "AI gets the grunt. You keep the talent."
- Conflict note (CONVENTIONS rule 25): the highest commercial conflict on the bench per dollar of claim — the revenue figures are the marketing; weight the how-to, discount the outcomes.
Weak spots / open questions
- No independent verification of any revenue figure ($100K a month, $1M, $3M, $5M+, $150K a month profit, "$231K in 30 days"); the last was never located at source.
- Survivorship built in: he teaches the business he exited, and his income is from teaching it — the reviewers' critique stands unanswered (scamrisk, 2026-01-22; ippei, 2026-05-20).
- Audience fit for Rich: none of the material targets senior operators; Mark's framing was wrong on this axis.
- "n8n Expert Partner" unverifiable from n8n's public page; "#1 bestseller" unverified; a sentence that "Uppit AI merged with OpenLake to form TrueHorizon AI" (Jan 2025) appeared only in a search-engine snippet and is not on TrueHorizon's About or case-study pages as of 2026-09-21.
- Platform and vendor risk: Skool + YouTube + Anthropic/OpenAI roadmaps; a second pivot inside 18 months is the base rate here, not the exception.
- Saturation: the "One Person AI Agency" × 460K members problem — every student is a competitor for the same small-business clients.
- Identity note: a "Nate Perry" appears among TrueHorizon's engineers — a different person; do not conflate.
- Carried from v1 — Rich's to-do list (captured Aug 13, 2026), status at this review: (1) watch 2–3 YouTube videos to assess quality — open, Rich's; (2) join the AI Automation Society free call — open, Rich's; (3) evaluate whether his model aligns with the Agency Lab vision — partly answered above: same ladder, different altitude; (4) note frameworks or pricing models worth stealing — extracted: education → audit → project → retainer; $5K audit then paid implementation; $5K–$20K a month retainers; Four Cs / Three Ms; workflows over agents; five conversations before a niche.
Rich's take
- (your synthesis here)
Delta log
2026-09-21 — v1→v2 migration + first validation (bulk run)
- Grades: 5 Held / 3 Drifted / 1 Wrong; 4 items unverified and not graded (retainer pricing as his own result; n8n Expert Partner; Amazon #1; the "$231K in 30 days" figure); 3 predictions live with dated checks; 0 unfalsifiable.
- Wrong: the v1 audience framing — "people with enterprise/consulting experience." His mission is "non-technical people who want leverage" and the paid tier starts "from zero" (about page; Plus page, read 2026-09-21).
- Drifted: "no code needed" → lives in Claude Code / Codex (X, 2026-05-01; CreatorDB, 2026-09-01); own-business revenue $1M → $3M → $5M+ with the referent moving from agency to education (Plus page; X 2026-05-01; free page, live 2026-09-21); n8n-first thesis → coding agents (Class Central; CreatorDB).
- Held: teaches agency-building (Plus page; aifunnelinsider, updated 2026-09-08); the agency exit (TrueHorizon About, 2026-09-21; roster, 2026-07-26); Skool Games 4/4 in 2025 (Q1, Q4 posts); Forbes (2026-04-24) and Microsoft Learn features; the 2026 price-increase call ($99 → $129).
- Most surprising delta: Mark's "our kind of experience" mentor is 24, two years out of college, exited the agency he teaches after 11 months, and now runs a ~$5M-a-year education business whose 2026 curriculum moved wholesale from n8n to Claude Code.
- Employer/role check (mandatory): the capture stub (2026-08-13) carried no affiliation at all. Found: TrueHorizon AI co-founder Jan–Dec 2025, exited; current title Founder & CEO, Uppit AI (Skool profile, 2026-09-21). Incentive map derived for the first time: maximal.
- Post-review sources (after the file's 2026-08-17 stamp): Skool "I Cloned Calendly" post (~2026-08-28); CreatorDB snapshot 2026-09-01; aifunnelinsider update 2026-09-08; developereducators, Sep 2026; Skool free, Plus and profile pages read live 2026-09-21.
- Not verifiable this cycle: the "$231,000 in 30 days" origin; the Amazon category badges; the n8n partner credential; the Uppit AI / OpenLake merger sentence; the book's publication date (July 2026 by community post only). Eighteen searches plus primary fetches; the last several changed no grade.
- Tier: quarterly — a creator who changed substrate (Alteryx → n8n → coding agents), entity (Uppit → TrueHorizon → Uppit), price and headline revenue inside 18 months will move again before a semiannual check.
- Confidence: low-medium — every load-bearing number is self-reported; the trajectory and the pivots are independently observable.
Sources
- His own pages (read 2026-09-21): nateherk.com/about · nateherk.com · AI Automation Society (free, 459.9k) · AI Automation Society Plus ($129/mo, 3.5k) · Skool profile ("Founder & CEO @ Uppit AI") · GitHub nateherkai/AIS-OS (MIT) · becomingainative.com · book offer page ($4.99 promo, bonuses)
- His posts, dated by ID: "Build a One-Person AI Business in 2026" (X article, 2026-08-11) · "$100k/mo … I built a team around me" (X, 2026-08-11) · Skool post for the same video (2026-08-11) · "12 Lessons From 5,000 Hours Building With AI" (2026-08-04) · "I asked Claude Code to make me as much money as possible" (2026-06-25) · "My Claude Code OS Runs my $3M/yr Business" (2026-05-01) · free 2+ hour course + repo (2026-05-01) · "I Cloned Calendly" / SnagTime, built with Codex (~2026-08-28) · Q4 2025 Skool Games win + 2026 price notice (2025-12) · Q1 2025 Skool Games win · Claude Code Kickstart (archived)
- Profiles and features: Apify — "From BI to AI" (2025-07-31) · Microsoft Learn — Learn With Nate Herkelman (updated 2025-02-06) · Forbes — AI podcasts and guides for business owners (2026-04-24) · Atak — 17 AI YouTubers (2026-02-20) · thusie — the Claude Code AI OS, Four Cs (2026-05-23) · Workreap — 4-stage blueprint (2025-12-21) · BuiltWithAgents — $6K/mo YouTube-strategist retainer (undated) · Class Central — Build Anything with Claude Code and n8n · LinkedIn (not fetched; search snippets show both "Co-Founder & CGO @ TrueHorizon AI" and "Founder & CEO @ Uppit AI" headlines)
- Company / verification: TrueHorizon AI — About (founders Tahliani, Baughcome; Nate not listed) · TrueHorizon — Topgolf case study · The Expert Roster — Uppit AI LLC (verified 2026-07-26) · n8n expert partners (no public list) · Amazon listing (did not render) · community post — book "just arrived" (2026-07-20)
- Metrics: CreatorDB (snapshot 2026-09-01: 968K, +88K/30d, 5.6 uploads/wk) · Developer Educators (Sep 2026: 1M, 512 videos, 53.8M views)
- Reviews (affiliate-adjacent; used for critiques, not facts): aifunnelinsider (2026-03-13, updated 2026-09-08) · ippei (updated 2026-05-20) · Skoolmakers creator page (2026-04-12) · ScamRisk (2026-01-22)
- Rich-side: Fractional AI Operations Officer — Agency Lab flagship (8.4) · v1 capture stub, Aug 13, 2026, via Mark (text message)
migrated v1→v2 2026-09-21 · backup: _archives/nate-herk-ai-consulting.html.bak-20260921 · converted-from: nate-herk-ai-consulting.md (captured 2026-08-13 via Mark) · Building domain · Agency Lab