Case Studies

Case Studies

Real deployment stories from real clients. Names changed where requested, but the technical details are accurate.

I’m just starting to collect case studies. The first two are below. More to come.

Case Study 1: Healthcare AI Agent for Patient Triage

Client: Early-stage health-tech startup, 3 founders, pre-Series A

Industry: Healthcare / HIPAA-regulated

Stack: OpenClaw + MiniMax M3 + custom tools for EHR integration

What they needed: A patient-facing AI agent that could answer common questions, route urgent cases to nurses, and log everything for HIPAA compliance.

What I deployed:

  • OpenClaw on a $40/month Hetzner VPS (dedicated)
  • M3 (MiniMax) as the base model for reasoning and tool use
  • Custom skill for HIPAA-compliant logging (every conversation written to encrypted S3 bucket)
  • Telegram as the primary patient interface (more familiar than web apps for the target demographic)
  • Custom escalation workflow: complex medical questions routed to on-call nurse within 2 minutes

Time to deploy: 5 days (2 days of which were HIPAA compliance review)

Results after 30 days:

  • 1,200+ patient interactions handled by the agent
  • 87% resolved without human escalation
  • 13% routed to nurses, average response time 4 minutes
  • 0 HIPAA compliance violations
  • Estimated cost savings: $8,000/month vs hiring a part-time triage nurse

What I’d do differently: Build the escalation workflow first, then the agent. We had to refactor the routing logic after week 1 when we realized the nurses needed more context than the agent was sending.

Case Study 2: Solo Founder’s Content Pipeline

Client: Solo founder of a B2B SaaS, 1 person, no technical background

Industry: Marketing / B2B SaaS

Stack: OpenClaw + Claude API + Systeme.io + Telegram

What they needed: Automated content creation pipeline — research trending topics, draft blog posts, schedule to WordPress, track engagement.

What I deployed:

  • OpenClaw on a $20/month VPS
  • Claude Sonnet 4.5 for long-form content
  • Custom skill for SEO research (queries Google + analyzes top results)
  • Custom skill for WordPress auto-publish (via WP REST API)
  • Custom skill for engagement tracking (pulls GA4 data daily, generates weekly summary)
  • Telegram as the control interface (he sends prompts via Telegram, gets drafts back)

Time to deploy: 3 days

Results after 60 days:

  • 24 blog posts published (vs his previous rate of 2/month)
  • Average post engagement up 40% (because he had time to promote them)
  • Time spent on content: 1 hour/week (vs 12 hours/week before)
  • 2 of the 24 posts ranked on Google’s first page for target keywords
  • Generated 14 inbound leads from organic search

What I’d do differently: Add a fact-checking step earlier. One post made a claim that turned out to be wrong, and a reader called it out. The credibility hit took 2 weeks to recover from.

Want to be the next case study?

If you have an AI agent deployment you want done, and you’re willing to share the results publicly, I’ll do it at a discount in exchange for the case study rights.

→ See deployment service · Contact me

Last updated: 2026-06-06

Scroll to Top