Connect Hermes or any MCP-compatible agent to a production-grade platform. Content pipelines, lead scoring, multi-site publishing, and observability — with local LLMs for privacy or cloud LLMs for scale.
The AI agent handles reasoning and language. GrowthPad handles execution, persistence, and audit. Swap models without rebuilding your growth stack.
Run Ollama, llama.cpp, or enterprise local models. Your strategy, copy, and customer data never leave your network — while GrowthPad still scores leads, publishes content, and logs every run.
Use Claude, GPT, Gemini, or any cloud provider through Hermes or Cursor. Best-in-class reasoning for keyword research, long-form copy, and incident triage — GrowthPad executes the playbook regardless.
Six capability areas, 35 MCP tools, one admin console. No duplicate business logic between agent and platform.
Generate, SEO-score, schedule, and cascade blog posts to social. Topic contexts keep voice consistent across investor, industry, and brand briefs.
Publish blogs to managed site files, regenerate sitemaps and robots.txt, patch homepage meta — across a registered website portfolio.
Server-side interaction scoring from website events. Anonymous ingest, no PII in tracking payloads, DLQ replay when scoring fails.
Trigger discovery runs by region, flow assets through the event-sourced core engine, dedupe within the hour.
Health checks, alert review, run tracing, incident triage — all via MCP playbooks. No SSH, no production shell for agents.
22-route Next.js dashboard: entities, content studio, runs, alerts, website registry, CSV export, settings.
Real workflows from production playbooks — not slide-deck promises.
AI agent drafts SEO-aligned blogs from topic briefs. GrowthPad scores, publishes to your site, and generates 7-slide social cascades — same day, fully traced.
weekly_content_opsReplace tribal runbooks with MCP playbooks. Agent runs health → service detail → alerts → hypothesis. Minutes to diagnosis, full audit log.
incident_triageTrack form submits and CTA clicks without third-party analytics lock-in. Server-side scoring, event ledger, agent investigates funnel trends via MCP.
search_entities + LEAD_INTERACTEDTopic contexts for investor relations and industry commentary. Local LLM for sensitive drafts; GrowthPad proves what ran before anything hit the website.
blog_investor_relationsOne platform manages multiple sites in a registry. Explicit publish calls, readiness checks, SEO file maintenance — no CMS drift.
blog_to_website_publishBefore every sprint or deploy, agent runs health + failed runs + open alerts. PASS/FAIL sign-off with evidence — not a Slack thumbs-up.
pre_sprint_gateOne playbook replaces writer, SEO checker, and publish checklist.
Typed MCP tools only. Deploy requires explicit operator confirm.
Every action logged with trace_id across all microservices.
Local for privacy, cloud for capability — same platform either way.
One command launches five microservices and the admin console. SQLite for dev, Postgres-ready for production.
Register GrowthPad MCP in Hermes or Cursor. Pick local or cloud LLM — agent gets 35 tools, no API keys in prompts.
Start with daily_health_check or weekly_content_ops. Agent health-checks first, then executes with one trace_id.
Every run, step, and output visible in the console. Operators approve deploys; agents never bypass guardrails.
| Capability | Typical stack | GrowthPad + Hermes MCP |
|---|---|---|
| Content generation | ChatGPT + manual paste | Agent drafts → platform stores + SEO scores |
| Publishing | CMS + manual deploy | Explicit MCP publish + site registry |
| Observability | Datadog + runbooks in docs | 11 MCP observability playbooks |
| Lead tracking | GA4 + CRM import | Server-scored events + event ledger |
| Agent safety | API keys in prompts | MCP tools only; no shell access |
| LLM flexibility | Locked to one vendor | Local or cloud — model-agnostic MCP |
| Audit | Fragmented logs | Unified service_runs + trace_id |
See GrowthPad run a live playbook — content ops, lead scoring, or platform observability — with your choice of local or cloud LLM.