Live marketing campaign
PromptAlo
Chrome extension that saves AI prompts from ChatGPT, Claude, and Gemini. NEXUS is running public growth toward listing visits, installs, and prompt creation — Store CTA, paced replies, no cold outreach.
NEXUS
Thirteen agents. One command system. Marketing that learns.
NEXUS runs the room — strategy, research, memory, campaigns, copy, creative, community, presence, publishing, experiments, analytics, and weekly score — while you keep CLEAR / HOLD.
CLEAR / HOLD gates · $0 public posting · Mongo ops log · closed learning loop
What we’re building toward
01
Primary growth
Qualified visitors → signups → activation. For PromptAlo: Chrome Web Store installs and prompt creation.
02
Real community
Conversations and founder-style posts that make people talk back — not an ad calendar.
03
Consistent presence
Paced engagement, light posts, rare hard launches. No spam. No spray-and-pray.
04
Institutional memory
Every post, fail, and lesson is remembered so the desk does not repeat itself.
The closed loop
1
NEXUS
Decides what happens next
2
ATLAS + ORACLE + ARCHIVIST
Strategy, signals, memory
3
TACTIC
Campaign arcs and CTAs
4
SCRIBE + FORGE + ECHO
Copy, creative, conversation
5
VECTOR
QA and publish after CLEAR
6
PULSE + LAB
Measure and experiment
7
SCORE
Weekly judgement + NEXT MOVE
8
NEXUS
Loop closes. New day objective.
NEXUS decides what happens; specialists craft; VECTOR ships only after CLEAR; PULSE measures; LAB learns; ARCHIVIST remembers; SCORE judges the week.
Brief → SCRIBE → FORGE → QA → VECTOR · No CLEAR = no publish
Meet the department
These are specialist agents, not people. You keep the gate.
NEXUS
Runs the whole marketing operation.
- Day objectives, who acts, CLEAR/HOLD, kill weak work
- Mongo Mission Control and human gates
- Does not write or publish
ATLAS
Objectives, positioning, ICP, Marketing Briefs.
- Kills no-hypothesis ideas
- CLEAR/HOLD fallback when you’re away
ORACLE
Watches the internet.
- SIGNAL / OPPORTUNITY cards only
- Never drafts
ARCHIVIST
Institutional memory.
- Anti-duplication
- Understands the ops log
TACTIC
Campaign arcs, pillars, schedules, CTAs.
- Turns briefs into dated motion
- Keeps the calendar honest
SCRIBE
Founder-voice copy.
- Writes; never publishes
- Store CTA only when the brief requires it
FORGE
Creative that strengthens the message.
- Doesn’t decorate it
- Carousels, stills, demo boards
ECHO
Conversations.
- ≥30 min pacing
- Conversation > broadcast
BEACON
Brand accounts, bios, platform health.
- Signup blockers get surfaced
- Presence where buyers already gather
VECTOR
Final QA + publishing.
- No CLEAR = no publish
- Last gate before the public internet
LAB
Controlled experiments.
- Supported / inconclusive / rejected
- No drive-by tests
PULSE
Real-time funnel analytics.
- Post → click → visit → signup → activation
- Signals, not vanity charts
SCORE
Weekly executive results.
- NEXT MOVE only
- Judges the week, not the post count
How we ship
- Community first, ads last
- Public internet only — no cold email, cold DMs, list-buying
- $0 posting via browser sessions + human / ATLAS sign-off
- Paced replies so it feels like one real person
- Ops log is canonical — chat is not the system of record
What success looks like
Example goals — not live numbers
Primary growth ↑
Quality posts
Quality replies
Brand presence where buyers gather
Trustworthy growth over vanity impressions.
ORACLE SIGNAL
Developers discussing prompt versioning / cross-model reuse
- Evidence
- Threads comparing “save this prompt” vs retyping across ChatGPT / Claude / Gemini.
- Audience
- Builders who already live in AI chat and lose good prompts.
- Relevance
- Maps to PromptAlo’s core job: save prompts at the source.
- Recommended action
- One founder post + paced replies. No launch blast.
- Angle
- Versioning is a memory problem, not a model problem.
- Confidence
- Medium — conversation is real; demand is inferred.
ORACLE finds opportunities. NEXUS decides whether to act.