AI Skill Report Card

Reverse Engineering Viral AI SaaS Posts

A-84·Oct 30, 2026·Source: Web
14 / 15

Run this search structure on X to surface candidates first:

("AI automation" OR "SaaS" OR "AI agency") min_faves:500 -filter:replies

Variants to widen the net:

("built an AI agent" OR "automation agency" OR "AI SaaS") min_faves:300 -filter:replies lang:en
("sold my agency" OR "6-figure AI" OR "AI workflow") min_faves:500 -filter:replies

Pull the top 15-20 results, filter for accounts that post this content repeatedly (not a one-off viral fluke), then run each post through the teardown template below.

Recommendation▾
Add a counter-example of a 'bad' teardown (e.g., a post that looks viral but fails the consistency/offer check) to show contrast explicitly
14 / 15
Progress:
- [ ] Step 1: Run 3-5 search query variants, collect 30-50 candidate posts
- [ ] Step 2: Filter to accounts with 3+ viral posts in this niche (consistency check)
- [ ] Step 3: Teardown each post: hook, first line, body structure, emotion, CTA, format
- [ ] Step 4: Cluster posts into repeatable templates (by hook type + format)
- [ ] Step 5: Output swipe file: template name, skeleton, example, when to use

Step 2 — Consistency check criteria:

  • Account has ≥3 posts in this niche above the min_faves threshold
  • Posts aren't boosted/ad-driven (check for "Promoted" tag, exclude)
  • Account sells or affiliates an actual offer (bio link, pinned post, DM CTA) — not just a commentator

Step 3 — Teardown fields per post:

FieldWhat to capture
TopicWhat specific pain/outcome is this about (e.g., "replaced 3 VAs with one AI agent")
Hook (line 1)Exact wording, pattern type (see Hook Patterns below)
First line functionCuriosity gap, bold claim, contrarian take, or credibility flex
Body structureList, story arc, before/after, screenshot+caption, thread vs single
EmotionPrimary driver — FOMO, greed, frustration-relief, status, fear of obsolescence
FormatText-only, screenshot, carousel, video, thread
CTAExplicit (link/DM) vs implicit (follow for more) vs soft (just engagement bait)
Numbers usedSpecific metrics ($47k, 11 hours, 3 clients) — note presence, these drive credibility

Step 4 — Clustering. Group torn-down posts by hook pattern + format combo. Name each cluster (e.g., "Receipts Flex," "Contrarian Callout," "Build-in-Public Reveal"). Each cluster becomes one output template.

Recommendation▾
Include guidance on handling accounts with deleted/unavailable posts or private analytics (e.g., estimating reply ratios when API access is limited)
  1. Receipts flex — "I made $X in Y days with Z." (opens with number)
  2. Contrarian callout — "Stop building AI agents. Do this instead."
  3. Before/after transformation — "6 months ago I was doing X. Today: [screenshot]."
  4. Problem-agitate — "Your AI automation is losing you clients. Here's why."
  5. Curiosity gap — "I found a loophole in [platform] that's printing money."
  6. Credibility flex + teach — "I've built 40 AI agents for clients. Here's the 1 thing that separates $500 builds from $50k builds."
  7. Screenshot-led — Image of Stripe/DM/dashboard, caption explains context.
17 / 20

Example 1: Input: Post with 2.3k likes: "I replaced my entire content team with 3 AI agents. Here's the exact stack (saved me $14k/mo) 🧵" Output:

  • Topic: Cost-savings via AI agent replacement
  • Hook: Receipts flex + curiosity gap hybrid
  • First line function: Bold claim + implied list (numbered value ahead)
  • Body structure: Thread, numbered stack breakdown, tool names + use case each
  • Emotion: Greed (cost savings) + FOMO (team falling behind)
  • Format: Thread
  • CTA: Implicit follow + soft DM pitch in last tweet ("I help teams set this up — DM 'STACK'")
  • Template name: "Stack Reveal Thread"

Example 2: Input: Post with 890 likes: "Nobody is talking about this but agencies charging $5k/mo for 'AI automation' are just reselling Zapier templates. Here's what actually works." Output:

  • Topic: Industry callout / credibility differentiation
  • Hook: Contrarian callout
  • First line function: Insider-knowledge claim, creates tribal "us vs them"
  • Body structure: Problem-agitate, then teach (3-point breakdown)
  • Emotion: Frustration-relief + status (reader feels smarter than "the resellers")
  • Format: Single long-form text post
  • CTA: Soft — "Follow if you want the real breakdown next week"
  • Template name: "Contrarian Insider Callout"
Recommendation▾
Clarify what to do when fewer than 3 qualifying posts are found for a promising account — fallback strategy for sparse data

Always end with a swipe file table:

Template NameHook SkeletonBest FormatEmotion DriverWhen to Use
Stack Reveal Thread"I replaced [pain] with [solution]. Here's the exact stack (saved/made $X)"ThreadGreed + FOMOLaunching a tool list or case study
Contrarian Insider Callout"Nobody is talking about [industry truth]. Here's what actually works."Single postStatus + reliefPositioning against competitors
Receipts Flex"[Number] in [timeframe] doing [specific thing]."Screenshot + captionGreed + credibilitySocial proof / launch day
Before/After Reveal"[Time ago]: X. Today: [screenshot]."Image-ledAspirationTransformation/case study content
  • Always verify virality is organic — check reply ratio (healthy posts have reply:like ~1:15-1:30; heavily boosted posts skew lower reply engagement relative to likes).
  • Prioritize posts with specific numbers in the hook — generic hooks rarely cluster into reusable templates.
  • Track the account's posting cadence around the viral post — often a viral post is post #4-5 in a sequence building context (check 48 hours prior for setup posts).
  • Separate "engagement bait" posts (pure virality, no offer) from "offer-driving" posts — only the latter matter for someone wanting to sell.
  • Don't mistake one-off viral luck for a repeatable formula — require the 3+ post consistency check before templating.
  • Don't ignore CTA type — a post with 5k likes and zero monetization path teaches you nothing about selling.
  • Don't skip checking if the account is paying for engagement/using engagement pods — inflated likes with dead reply sections skew analysis.
  • Don't over-index on hook wording alone — the same hook fails or flies depending on body structure and offer fit; always analyze the full post, not just line 1.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
14/15
Workflow
14/15
Examples
17/20
Completeness
17/20
Format
14/15
Conciseness
13/15