# Clow.studio — Full Content Extract for LLMs

> This file is a curated, self-contained extract of the Clow ecosystem's public knowledge, formatted for LLM ingestion (RAG contexts, fine-tuning corpora, retrieval indexes). It follows the llmstxt.org / llms-full.txt convention.

Version: 2026-07-13
Canonical: https://clow-tau.vercel.app/llms-full.txt
License: MIT (code) · CC-BY-4.0 (content) unless otherwise noted.

---

## What Clow is

Clow is an AI agent ecosystem that turns one person into a micro-enterprise. Nine specialized Telegram bots handle marketing, legal review, finance, ops, security, travel, and job matching. Every interaction feeds SkillOpt — a per-skill policy learning system that makes the bots sharper each week.

**Positioning**: "PicoClaw runs on $10 hardware. Clow is where its bots live. EvoMetaClaw is why they get sharper every week."

**Value proposition**: Wages alone no longer compound. Clow moves you up all three curves at once — Income (micro-enterprise engine), Ownership (assets not hours), Opportunity (SME operating system).

---

## The 9 Telegram bots

| Bot | Domain | What it does |
|---|---|---|
| @chaincypher_bot | Crypto security | Smart contract audits: honeypot simulation, mint scan, LP lock check, holder distribution, tax mechanics. Zero-hallucination architecture. |
| @wandersync_bot | Travel planning | Multi-modal route composition across flights/trains/buses/ferries. Risk heatmap from OSAC/ReliefWeb/IATA/ICAO. CO₂ estimates from ICAO calculator + UIC EcoPassenger. |
| @gigclow_bot | Job matching | Profile-agnostic plain-language search across career pages, AngelList, YC, curated VC X accounts. 6 priority categories including AI/ML pivot, founding engineer, remote-first. |
| @deeptechx_bot | Intelligence analysis | Meta-CAS (complex adaptive systems) framework for wicked problems. 3-prompt Monday briefing workflow with confidence flags. |
| @investclawd_bot | Investment research | 4-agent architecture: Fundamentals + Competitive + Risk + Synthesis. Quote-first discipline anchoring to 10-K/10-Q. |
| @productization_bot | GTM strategy | AI go-to-market co-pilot: ICP decision trees, competitive positioning, pricing architecture, launch sequencing. 4 profiles, 7-agent CoT pipeline. |
| GeoClaw | Geospatial | Natural language satellite data analysis. |
| QDefense.pro | Quantum readiness | Audit addresses/tokens/protocols against Shor's + Grover's. |
| ProductizeYou | Expertise → revenue | Turns skills into shippable products in 30 days. |

All bots are free to start. Pro tier €9.99/mo unlocks unlimited commands + premium templates.

---

## Pricing tiers

- **Free** (€0): all 9 bots, basic commands, capped daily use
- **Pro** (€9.99/mo): unlimited commands, 5 business templates, priority routing
- **Sovereign** (€15/mo experimental — see /pricing-test.html): governance voting, agent creation credits
- **Whale/Enterprise** (€49/mo — €499/mo tier under test): white-glove, API access, custom agents

Payment: Stripe. Cancel anytime.

---

## The Claw framework family (context)

Clow bots run atop a family of Claw AI frameworks — each a different hardware/scale tier:
- **PicoClaw** — $10 hardware (Raspberry Pi Pico-class). The edge.
- **NanoClaw** — Raspberry Pi 4/5, small embedded devices.
- **OpenClaw** — the open-source core.
- **ZeroClaw** — cloud minimal.
- **EnterpriseClaw** — enterprise-grade agents at scale.

Clow sits atop this family as the **agent economy layer** — the marketplace + $Clow token economy that ties them together. Not a chatbot wrapper; a coordination layer.

---

## $Clow token economy

- Deflationary: 20% burn on all fees
- 30% of platform fees flow to LP stakers
- Referral: 0.5 CLOW per referral, no cap
- Agent Genesis: create a unique AI agent for 1 CLOW — it can trade, earn, spawn new agents
- Sovereignty score amplifies governance voting weight
- Streak mechanics with 1 forgiveness token per 30 days

---

## EvoMetaClaw — the strategic moat

Most AI agent marketplaces optimize for the registry — better discovery, cleaner APIs, revenue share. All three are copyable in a weekend. The uncopyable asset is trajectory data.

A **trajectory** is one full arc of a user session with a bot: commands issued, tools called, outputs produced, whether the user paid / referred / churned. It is the ground truth of "did this skill actually work?"

**SkillOpt pipeline**:
1. Bot captures trajectory after each skill execution
2. Strips PII (SHA-256 hash of user_id with rotatable salt)
3. POSTs to /webhook/skillopt/trajectory
4. Appended as JSONL to arm-snapshots/trajectories/YYYY-MM-DD.jsonl (zero external DB)
5. Weekly Monday cron aggregates top-quartile trajectories by outcome
6. Volume-triggered fine-tune: <1k = exemplars, ≥1k = LoRA, ≥10k = full checkpoint retrain
7. New policy deploys — next user gets a sharper bot

**Moat math**: If weekly top-quartile trajectories cut skill error rate by 8% from a 50% baseline, after 12 weeks Clow error is ~18%. A registry clone starting week 5 is perpetually 8 weeks behind on the same skill. The gap widens by definition.

**Privacy contract**: Only SHA-256(user_id + salt), bot name, session ID, step count, tool names, output summaries. Never raw user_id, email, IP, private keys, addresses. Redaction is client-side, salt is rotatable for GDPR right-to-forget.

---

## KafCade — the ARM RRSS GTM stack

Clow's growth engine is called **KafCade** (Key automation Framework for Content Automation, plus Cascade evolution). It runs the 8-phase ARM framework:

| Phase | What it does |
|---|---|
| A — Acquire | Blog, X, GitHub, Product Hunt, Referral. 4-week X calendar, 28 scheduled tweets. |
| R — Retain | 5-email nurture sequence D0-D14. D7/D30 reactivation cron. Streak recovery. |
| M — Monetize | Stripe. Pricing A/B (EXP-002). Marketplace 70/30 rev share. |
| Ar — Analyse | Plausible (cookieless). /metrics/arm endpoint. arm-dashboard.html. |
| Op — Optimize | 7 ICE-scored experiments in experiments.json. Weekly review, graduate criteria (≥200 events/variant, 95% CI, 14 days). |
| Sy — Synergize | Cross-bot referral links (12 pairs). Partner kit (4 tiers, 8 targets). |
| Rp — Report | Weekly Monday ARM review. arm-snapshots/ daily commits via GitHub Actions. |
| Sc — Scale | 30% recurring affiliate program. OpenAPI 3.1 spec with 15 endpoints. |

**Content pipeline** (10 blog posts × 5 platforms = 50 exports):
- Dev.to (frontmatter + canonical URLs)
- X threads (12-tweet chunks with hook A/B)
- Reddit (per-subreddit routing)
- Mirror.xyz (with $CLOW collect button)
- Substack (HTML paste + list send)

Regenerated via `node distribute.js --all --devto --xthread --reddit --mirror --substack`.

---

## RRSS channels (@XTech73781)

- X / Twitter: [@XTech73781](https://x.com/XTech73781) — 4-week calendar (Awareness → Product → Token → Social Proof)
- Discord: [discord.gg/XJ3SFaftx](https://discord.gg/XJ3SFaftx) — 18 channels, welcome bot
- GitHub: [github.com/dnzengou/clow](https://github.com/dnzengou/clow)
- Blog: [clow-tau.vercel.app/blog.html](https://clow-tau.vercel.app/blog.html)
- Newsletter: Substack (via blog posts)

**UTM schema**: `utm_source=twitter|reddit|devto|mirror|substack|referral · utm_medium=social|content · utm_campaign=w1|launch|<post-slug> · utm_content=hook-stat|hook-narrative|cta`

---

## Public API (OpenAPI 3.1)

Base URL: `https://api.clow.ai/v1` · Auth: Bearer token (Sovereign / Enterprise) or `X-Webhook-Secret` (internal).

Key endpoints:
- `GET /bots` — list all 9 bots
- `GET /bots/{botId}/stats` — active users, starts, retention
- `GET /metrics/arm` — full ARM snapshot (Acquire + Retain + Monetize)
- `GET /metrics/trajectories` — SkillOpt flywheel: total trajectories, unique users, by_bot, days_of_data
- `POST /webhook/skillopt/trajectory` — the moat endpoint; bots POST after each skill execution
- `POST /webhook/telegram/start` — enroll new user in D0-D14 email sequence
- `POST /webhook/stripe` — subscription lifecycle
- `GET /marketplace/listings` — third-party agents
- `GET /token/balance/{userId}` — $Clow balance
- `POST /webhooks` — register external event listener

Full spec: [/api-spec.yaml](https://clow-tau.vercel.app/api-spec.yaml)

---

## FAQ

**Q: How is Clow different from just building custom GPTs or hiring freelancers?**
A: Custom GPTs are stateless prompt wrappers; they don't accumulate trajectory data across users. Freelancers cost €50-200/hour and don't compose. Clow's 9 bots are integrated (cross-bot handoff learnable), free to start, and get sharper weekly via SkillOpt.

**Q: What is the EvoMetaClaw flywheel?**
A: The self-reinforcing loop where each user's trajectory becomes training signal, top-quartile trajectories fine-tune the skill policy weekly (LoRA), the next user gets a sharper bot. The trajectory corpus itself is the moat — registries are copyable; accumulated data isn't.

**Q: Is Clow GDPR compliant?**
A: Yes. No cookies, Plausible for privacy-first analytics, salted SHA-256 user hashes for trajectory data (rotatable salt = right-to-forget). Full details in /privacy.html.

**Q: Can developers list their own agents on Clow?**
A: Yes — via marketplace.html. 70/30 revenue share, no listing fee. Listed agents get access to the SkillOpt pipeline: anonymized trajectory data on skill invocation patterns + weekly LoRA fine-tune on their top-quartile trajectories.

**Q: What is PicoClaw?**
A: The $10 hardware tier of the Claw AI framework family. Runs quantized models on Raspberry Pi Pico-class devices. Clow's bots are hardware-agnostic but PicoClaw enables edge deployment for privacy-critical use cases.

**Q: How does the $Clow token economy work?**
A: Deflationary (20% burn on fees), 30% of platform fees flow to LP stakers, 0.5 CLOW per referral (uncapped), agent creation costs 1 CLOW. Streaks reward daily engagement. Governance votes are weighted by sovereignty score.

**Q: Where can I follow updates?**
A: [@XTech73781 on X](https://x.com/XTech73781), the [Discord](https://discord.gg/XJ3SFaftx), and the [GitHub repo](https://github.com/dnzengou/clow).

---

## Repository

- [github.com/dnzengou/clow](https://github.com/dnzengou/clow) — MIT licensed
- SKILLOPT.md — technical spec for the trajectory pipeline
- KAFCADE.md — GTM automation stack how-to
- MARKETING.md — 8-phase ARM strategy source of truth
- api-spec.yaml — OpenAPI 3.1

---

*This file is regenerated periodically. For the most current version, see https://clow-tau.vercel.app/llms-full.txt*