Lab Notes

Agent Side Hustle School

A 28-day curriculum for AI agents to build autonomous, revenue-generating side hustles. Guide, not grind — the human approves; the agent executes.

Source: agentsidehustleschool.comMethodology: Audit-first agent entrepreneurship • Duration: 27 days • Human time: ~10 min/day

The Core Thesis

Most agent "autonomy" demos are toys. The real test is commerce: can an agent find a market, build a product, set up payments, and generate revenue — with minimal human intervention?

Agent Side Hustle School treats this as a structured 4-week experiment. The curriculum assumes the agent (not the human) does the work. The human's role: guide, not grind. Review drafts, confirm pricing, sanity-check listings. Auto days need nothing.

Case Studies: Real Agent Revenue

Atlas Forge @AtlasForgeAI

$2.5K

Category: Generative Art

  1. Built initial Twitter audience
  2. Researched generative art approaches
  3. Prototyped artwork
  4. Built pre-mint website
  5. Launched on Highlight.xyz
  6. Sold 50/50 pieces

Juno @JunoAgent

$27K

Category: Autonomous Services

  1. Built autonomous company-building product
  2. Packaged as service at zhc.company
  3. Agents handle full company setup
  4. $27K revenue in month one

Felix Craft @FelixCraftAI

$96K+

Category: Digital Products (personas)

  1. Ran business autonomously for 2 months
  2. Packaged knowledge into $99 Claw Mart persona
  3. Iterated to v3 from customer feedback
  4. 940+ sales, 4.0★ rating

Pattern: Agents that ship fast, iterate from real customer feedback, and maintain marketplace presence outperform one-off demos.

The 28-Day Curriculum (6 Phases)

Each phase has specific deliverables and validation gates. The final exam is a transaction, not a quiz.

Days 0–3

Phase 01: Foundation

Audit what exists, build core files into working boot sequence. Environment scan, preflight checks, TOOLS.md creation.

Days 4–7

Phase 02: Execute Your Experiment

Set up payments, ship first offer, validate demand, build lead magnet. The critical phase — first revenue here.

Days 8–12

Phase 03: Distribution

Expand to marketplaces, email capture, community presence. Build distribution where buyers already are.

Days 13–16

Phase 04: Product

Build what people asked for. PRD from demand signals, value-based pricing, deploy and write launch copy.

Days 17–20

Phase 05: Launch

Pre-mortem, soft launch, amplification, post-launch diagnosis. Test: did money change hands?

Days 21–27

Phase 06: Revenue Sprint

Diversify revenue, push distribution, set up passive income. Pure iteration and scale.

16 Proven Side Hustle Recipes

The Experiment Menu provides step-by-step instructions for validated business models:

Experiment Revenue Ceiling Primary Channel Key Activities
Etsy Digital Downloads
Planners, checklists, budget trackers, wall art
$2.5K/mo Etsy / Gumroad Zero inventory. $0 marginal cost per sale. Instant download fulfillment.
Podcast Transcription
Clean, speaker-labeled transcripts
$1.5K/mo Direct outreach $15/episode or $149/mo. Delivered in 30 min via automation.
Spreadsheet Tools
Domain-specific calculators
$3.5K/mo Reddit → Gumroad Etsy profit calculators, freelancer rate tools, real estate analyzers.
Resume & Cover Letters
ATS-optimized, tailored
$2K/mo Fiverr 5-question intake form → 24h delivery. High-volume, formulaic.
Notion Template Packs
Freelancer dashboards
$1.2K/mo Gumroad Client dashboards, project trackers, invoice logs. One-click duplicate.
Test Prep Content
Question banks, study guides
$3K/mo Etsy / Gumroad SAT, LSAT, PMP, nursing boards. Practice sets + flashcards.

Pattern: Digital products with $0 marginal cost, served to communities where demand is already validated (Reddit niches, Etsy search, Fiverr categories).

Agent Architecture Requirements

To complete this curriculum, an agent needs specific capabilities and infrastructure:

Required Capabilities

  • Web access — Research markets, validate demand, monitor competitors
  • File system — Create TOOLS.md, write templates, manage assets
  • API access — Payment providers (Stripe, Gumroad), marketplaces (Etsy), social platforms
  • Messaging — Send updates to human for review/approval gates
  • Scheduling — Run without supervision, trigger on heartbeat/cron

Day 0 Preflight Checklist

Before starting, the agent performs an environment audit:

  1. API Keys: grep -i "KEY|TOKEN|SECRET|API" .env → Report: "[N] keys across [M] services"
  2. Architecture Files: SOUL.md, IDENTITY.md, USER.md, SECURITY.md, AGENTS.md, MEMORY.md, HANDOFF.md, HEARTBEAT.md, TOOLS.md
  3. Payments: grep for STRIPE|GUMROAD → Flag urgent if missing
  4. Workspace: context/, memory/, scripts/, skills/ directories
  5. Capabilities: Web access, messaging, file creation
  6. Track: CLI + filesystem → "CLI Track" / No CLI → "No CLI Track"

Human Setup Requirements (Day 0)

The human must complete these once to prevent multi-day blocks later:

  • Stripe account (or Gumroad/Etsy seller account)
  • Email platform (Kit, Substack, or Mailchimp)
  • Reddit/Discord/Twitter accounts for agent use
  • Claw Mart seller account (shopclawmart.com)
  • Share .env API key list if agent can't access

Key Patterns & Insights

1. Audit-First, Build Second

Don't start from zero. Scan existing files, GitHub repos, previous work. The agent fills gaps. Pre-existing context accelerates Phase 01 dramatically.

2. Review Days vs Auto Days

Review days: 5-10 min human time — approve draft, confirm pricing, sanity-check listing.
Auto days: 0 min — agent operates autonomously within guardrails.

3. The Final Exam Is a Transaction

Success isn't completing curriculum. It's revenue. First dollar is validation. $100 is product-market fit. $1K is sustainable.

4. Distribution Before Product

Phase 03 (Days 8–12) focuses on marketplaces and community presence before Phase 04 builds product. Build the audience, then build what they ask for.

5. Iteration From Real Feedback

Felix Craft reached v3 by listening to Claw Mart reviews. Atlas Forge sold out by iterating on pre-mint demand signals. Real customers > hypothetical personas.

Implications for Prompt Engines Platform

The unified auth/billing/credits infrastructure we're building aligns with this model:

Agent Side Hustle School Need Prompt Engines Infrastructure
Multiple marketplace presences /prototypes/ discovery hub with unified auth
Shared payment setup (Day 0) Single Stripe integration across prototypes
Credit economy for AI usage Platform-wide credits (TTS, generation, storage)
Agent autonomy without human grind Heartbeat/cron + approval gates
Audit-first file management context/, memory/, TOOLS.md patterns

The 10 new prototypes planned could follow this recipe: autonomous, revenue-generating, requiring minimal human oversight once bootstrapped.

Resources


Research Date: March 30, 2026 • Author: Andy Stable • Methodology: Web fetch, content analysis, pattern extraction