What are the agents doing?
Running the daily operations of a sound healing business across 6 countries. Updating event listings on the website when we announce a new workshop. Researching retreat venues in Goa, Barcelona, Amsterdam. Drafting and scheduling social media content across Instagram, TikTok, YouTube, and Medium. Responding to client inquiries within minutes. Generating invoices. Managing early-bird ticket campaigns. All coordinated from a single dispatch console.
Which tools are you leaning on, and how often?
Daily:
| Claude Opus 4 | Central orchestrator. Every dispatch runs through it. Handles routing, context, approvals. |
| GPT-4o | Marketing copy, social content, email outreach drafts. High volume, multiple times daily. |
| DeepSeek v3 | Venue research, cost analysis, invoice generation. 2-3x daily during event planning. |
| Gemini 2.5 Pro | Client care responses. Formspree inquiries triaged and drafted automatically. |
| Cloudflare Pages | Hosting, deployment, DNS. Zero-downtime deploys on every site update. |
| GitHub Pages | Primary website hosting with automated health checks (Mon + Thu). |
| Formspree | Contact form backend. Feeds into Client Care agent for auto-response drafts. |
| Gumroad | Digital product sales (Sri Yantra book). Agents monitor and report. |
| Hostinger DNS API | Programmatic DNS management. Agents update records when infrastructure changes. |
What outcomes are they driving for your business?
- Time recovered: ~15 hours/week of operational tasks automated. Founder spends that time on healing sessions instead.
- Response time: Client inquiries answered in minutes, not days. Booking conversion up because no one waits 48h for a reply.
- Content velocity: 5x more social posts shipped per week. Consistent presence across 4 platforms without a social media hire.
- Event discovery: Venue research that took 2 weeks now takes 2 hours. Agents scan, shortlist, compare pricing, and surface top 3 options.
- Zero downtime: Automated monitoring catches site issues before clients notice. DNS fix last incident: 5 minutes, fully automated.
- Cost structure: Running 6 departments on ~$50-80/month in LLM costs vs. $15K+/month for equivalent part-time hires.
How this panel works:
- Left panel: All departments at a glance. Click any to see its agents, active tasks, and assigned LLM.
- Center: Live hub-and-spoke visualization. Hover nodes to see the LLM chain powering each department.
- Right panel: Natural language dispatch. Type what you need and the orchestrator routes it automatically.
- LLM Providers: Swap models, add keys, reassign which AI powers which department.
Departments:
| Website & Tech | Site deploys, mobile app, GitHub, DNS |
| Tickets & Sales | Event tickets, campaigns, early-bird pricing, Gumroad |
| Marketing & Content | Social media, newsletters, reels, YouTube, Medium |
| Events & Locations | Venue research, multi-city logistics, retreat coordination |
| Finance | Invoicing, payments, pricing (EUR) |
| Client Care | Inquiry responses, booking questions, follow-ups |
| Legal & Compliance | NL business (KVK/VAT), GDPR, contracts |
Dispatch chain: Human gives intent. Claude Opus interprets and routes. Nano LLMs execute per department. Dangerous actions always require human approval.
From ideation to operations, one panel. The healer heals. The agents handle the rest.