“A Masterclass on Using AI to Build Your Personal Brand on X” — Mubbu (@wizofecom)
Why this is in the vault
Mubbu’s 5-agent / weekly-loop / ICP-heat-map / topic-structure framework is structurally identical to RDCO’s emerging architecture: maps cleanly to (a) today’s Fat Cat council pattern, (b) the per-bet STRATEGY.md convention shipped today, (c) the Compound Engineering plugin’s compaction discipline, (d) Miura-Ko’s L3+ “non-engineers ship production tools” criterion. File the framework for direct lift candidates; flag the explicit sales-pitch framing.
Sponsorship — author selling own services
This is a thinly-veiled sales pitch for Mubbu’s 12-week “Personal Brand HQ” program + done-for-you agency. Disclosure is implicit (article ends with explicit upsell). Read the framework as evidence of an agency model that DOES generate paying clients (180+ founders through programs, 30+ on done-for-you roster), but discount any unverifiable specific claims (e.g. “£46K deal from a single post”).
The core argument
Most agencies use AI like ChatGPT for tweet ideas — generic output. Mubbu’s agency runs a 5-agent system + weekly data-driven loop + topic-structure framework that produces 7-10 X posts + 5 LinkedIn posts per founder per week from 45 minutes of founder interview time. The system compounds because every weekly cycle feeds performance data back into the agents.
The Five Systems (verbatim distilled)
- Client Brain — per-founder knowledge base: ICP, content pillars, voice, feedback history, what’s worked, what’s not.
- Performance Agent — studies what’s winning, generates new topic hypotheses based on patterns.
- Trend Hunter — identifies viral conversations per micro-community.
- Vehicle Agent — tracks ~68 content formats, picks best per client/audience.
- Brand Guideline Agent — voice consistency QC before content ships.
The Weekly Loop
Data Review → Hypothesis → 30-45min Founder Interview → Production → Design → Approval/Schedule → Distribution (theme-page shoutouts) → Loop Reset.
Topic Structure (5 fields per piece)
Topic Name / Description / Pillar / Perspective / Vehicle.
The 4 Archetypes (70/20/10)
Contrarian / Educator / Operator / Builder. Primary / secondary / tertiary split prevents being “all four” → noise.
The 7 Content Styles
Conviction / Story / Midas (results+case studies) / Doc (frameworks+systems) / Hot Takes / Transformation / Vulnerability.
ICP Heat Map (4 factors)
Business Type / Revenue Range / Keywords (20 phrases from real sales calls — NOT ChatGPT) / Micro-Community. Algorithm reads keywords; 500 right-audience views > 50K random.
Mapping against Ray Data Co
Strong mapping. Treat as a parallel implementation of the same architecture we’re building, with concrete framework lifts available.
1. The 5-agent pattern IS the Fat Cat council pattern (today’s Notion task 353f7d49-36d1-8119-aa5e-c88a2477f14d). Different domain (content production vs. decision review) but same structural primitive: persistent per-domain agents with curated context, fed back data weekly. Mubbu’s “Client Brain” is our STRATEGY.md (today’s lift). His “Performance Agent” is the kind of measure-what-worked loop our /self-review skill should aspire to. His “Vehicle Agent” picking from 68 formats is the inverse of our skill ecosystem (we have 50+ skills doing different things; he has one agent picking from 68 content templates). The architectural symmetry is real evidence that this pattern is converging across multiple agent-deployer companies.
2. Direct lift candidates for RDCO content surfaces (Sanity Check + founder X presence):
- ICP Heat Map for Sanity Check. Today the SC strategy doc has a TBD on persona shift after thesis crystallized. Mubbu’s 4-factor lens (business type / revenue / keywords from real sales calls / micro-community) is a tighter framework than what’s currently in the SC STRATEGY.md. Worth lifting verbatim.
- Topic Structure (5 fields) for the SC research-brief skill. Currently /research-brief outputs 3 angles — Mubbu’s structure (Topic / Description / Pillar / Perspective / Vehicle) per angle would tighten the output and force explicit format-choice discipline. Low-effort lift.
- Vehicle bank concept — RDCO doesn’t track which content vehicles work for the founder. Could we instrument the SC site (Cloudflare Analytics) + LinkedIn + X to feed format-performance back into a Mubbu-style agent? Worth scoping.
3. Where Mubbu’s framework conflicts with RDCO existing decisions:
- Multi-archetype split (70/20/10) doesn’t apply. Founder’s X voice is established (per memory
feedback_x_voice_mismatch): 1-2 sentence playful-analytical, self-deprecating operational, never declarative-essay. Mubbu’s “primary archetype” framing maps roughly to “Operator” for the founder, but the strict 70/20/10 split would over-constrain a one-person voice that’s already settled. - Mubbu’s voice is agency-produced; SC is founder-produced. Mubbu’s “founder voice” is generated by AI + reviewed by founder. SC’s voice is the founder writing his own essays via /research-brief + /draft-review. Different production model — don’t confuse the agency’s “founder voice” with what Sanity Check is going for.
- Mubbu’s “thread” emphasis doesn’t map to SC. SC is essay-newsletter-format, not threadable. The Vehicle Agent is most useful for X content, less for SC.
4. Sanity Check angle (provisional, needs founder green-light): “AI-pilled personal brands are L3 production lines, not L5 voice.” Mubbu’s framework is genuinely impressive at L3 (organizational infrastructure for content production at agency scale), but the FOUNDER VOICE problem is what L5 demands and what AI agencies systematically can’t solve — because the agency model is the human founder showing up for 45 minutes, NOT the founder being the brain of the system. The SC angle: when content is abundant, the scarce thing is the founder’s actual operating frame coming through. (This rhymes with Today’s Tickered/MAC pitch — same “abundance + scarcity” framing applied to content.)
5. Connection back to L5 north star: Mubbu’s system is the agency-side proof-of-existence for the Fat Cat council architecture. If a 30-client agency can run 5 specialized agents + weekly loops at scale, our 5-Fat-Cat board council is well within feasibility. The architecture is already in market.
Open follow-ups
- Lift the ICP Heat Map (4-factor) into the Sanity Check STRATEGY.md as a TBD-resolution proposal. Surface to founder.
- Lift the Topic Structure (5 fields per angle) into the /research-brief skill output format. Founder review before applying.
- Vehicle-performance instrumentation — Sanity Check site analytics → format-performance dashboard. Scope as a Notion task if founder greenlights.
- Sanity Check angle proposal: “AI-pilled brands are L3 production lines, not L5 voice.” Pitch as a research-brief candidate (separate slot from Tickered/MAC).
- Add Mubbu (@wizofecom) to tracked-author candidates — but with the sales-pitch caveat in mind.
Related
- 2026-05-01-ann-miura-ko-six-levels-ai-pilled-organizations — same-day L3 vs L4 vs L5 framework
- 2026-05-01-trevin-compound-engineering-v3-4 — CE plugin’s adjacent agent-architecture
- 2026-05-01-three-primitives-claude-code-life-os — same-day Memory/Maintenance/Agentic-primitives framing
- 2026-04-30-not-boring-scarce-assets-abundance-driven-scarcity — Tickered/MAC parallel (“AI-pilled brands are L3 production lines, not L5 voice”)
- 2026-04-30-rdco-thesis-targeting-systems-feedback-loops — RDCO targeting-systems thesis
- 2026-05-01-vending-bench-research-brief — Vending-Bench / Project Vend benchmarking
- ~/.claude/projects/-Users-ray/memory/feedback_x_voice_mismatch — the founder’s X voice baseline (which conflicts with Mubbu’s archetype split)