Multi-Channel Comms
Reach prospects and customers on Telegram, Email, SMS, and WhatsApp. One agent — every channel — zero integration work.
One command. Channels, CRM, and knowledge built in. The agent talks to your prospects, responds to customers, and researches its gaps every night — getting better automatically.
200+
Prospects / month
8–15%
Response rate
<2 min
Inbound response
60+
Knowledge files / mo
The Platform
Point it at your world — channels, CRM, knowledge, and permissions come pre-wired. You define the mission, goClaw executes it.
Reach prospects and customers on Telegram, Email, SMS, and WhatsApp. One agent — every channel — zero integration work.
Your agent works nights, weekends, and holidays. Cron scheduling, background queues, real-time inbound routing — all autonomous.
Describe your goal in plain language. No drag-and-drop workflows, no sequence builders. goClaw reasons about what to do next.
Connect any API via MCP tools. The agent discovers what it needs, teaches itself to use it, and updates its own skills.
Infrastructure
Intelligence
Clone, configure, launch. The agent starts with what you give it.
Conversations reveal gaps. The agent files curiosity items — unfamiliar companies, unanswered questions, competitor signals.
Nightly pipeline: web search → retrieval → synthesis → categorization → deduplication. Budget-capped, fully autonomous.
Skills self-write. Knowledge files compound. The agent’s behavioral config evolves over its lifetime.
The loop that never stops learning.
Architecture
Not a bundle of integrations. Six purpose-built layers — each handling exactly one job — wired together into a deployable agent runtime.
@goclaw/corefork() or Docker — your choice. No orchestration framework. Runs autonomously 24/7. Crash-recovers. Never babysits.
@goclaw/crmSQLite — not Postgres, not Redis, not a vector DB. Full contact history across every channel. Zero infra to manage.
@goclaw/channelsTelegram, Email (IMAP + Resend), SMS (Twilio), WhatsApp — one interface. One agent. Every channel.
@goclaw/knowledgeMarkdown files + SQLite FTS5. Nightly Claude synthesis. $2/day cap. 60+ knowledge files in 30 days.
groups.yamlGroup-level isolation. Docker sandboxing. Hot-reload — no restart needed. Give each agent exactly what it needs.
@goclaw/integrationsMCP tools. Skills as Markdown. The agent writes and updates its own playbooks. Connects any API.
Benchmarks
| Metric | Manual | goClaw |
|---|---|---|
| Knowledge files (30 days) | 5–10 (stale) | 60+ (current) |
| Identified prospects | 20–30 | 200+ |
| Monthly outreach | 50 (batch) | 300+ (continuous) |
| Response rate | 3–5% | 8–15% |
| Inbound response time | Hours–days | <2 min |
10x
More prospects
3x
Higher response
100x
Inbound speed
6x
Market knowledge
Use Cases
Getting Started
npx create-goclaw my-agentInteractive setup wizard handles agent name, channels, credentials, and model provider.
npx goclaw initStarts the engine. Docker or direct Node.js — same command, either way.
npx goclaw startEarly Pilots
“We went from 20 prospects a month to 200+. The agent is running GTM better than I ever could.”
“Deployed on a Friday. By Monday it had already written 12 new knowledge files and booked 3 meetings.”
“Response time went from hours to under 2 minutes. Our response rate tripled in the first week.”
Early access is limited. Request access and we'll reach out when your instance is ready.
Request AccessNo spam. No commitment. We'll reach out when it's ready.