Heidelberg AICurriculum

Here's how to build & sell AI systems using Claude Code

AI Foundations ·2026-02-20 ·2 min read

Summary written by us from the video's transcript. The video, and everything in it, is AI Foundations's work.

Learn how to quickly build and sell AI-powered agentic systems for niche clients using Claude Code and Lerty, from setup through creating databases, dashboards, and custom agents.

Takeaways

  • Claude Code can generate full relational databases in Lerty with realistic seed data in seconds.
  • Lerty’s organization model lets you host up to 50 client workspaces on the agency plan, making a single $500/month sale break even on tooling costs.
  • Dashboards turn raw tables into user‑friendly views (lists, filters, buttons) that clients can interact with without technical knowledge.
  • Custom agents powered by Lerty’s MCP server can act as conversational assistants, pulling data from the database and delivering prep briefs for coaches or other professionals.
  • A free resource bundle supplies a system prompt, 50 sellable agentic system ideas, and pricing guidance to accelerate niche selection and sales outreach.

Why Build Agentic Systems Now

There is growing pressure across professions—plumbers, lawyers, dentists—to adopt AI solutions even though they lack time or expertise. Builders can meet this demand by delivering tailored AI systems that generate significant recurring revenue for clients.

Essential Tools: Claude Code and Lerty

Claude Code (a paid Anthropic model) handles the heavy lifting of generating code and configuring Lerty; a subscription starts at $20/month. Lerty provides the infrastructure—organizations, databases, agents, dashboards—and its agency plan ($24/month) supports up to 50 client organizations, making a single $500‑month sale cover tool costs.

Getting Started: Resources and Setup

A free resource bundle includes a system prompt for Claude Code, 50 pre‑built sellable agentic systems, and pricing guidance (minimum $500/month). Users must install Claude Desktop, sign in, and obtain an API key from Lerty to let Claude edit the workspace.

Building a Coaching Client Management System

Create a new organization (e.g., “Elite Coaching”), paste the Lerty API key into the system prompt, then ask Claude Code: “Build a coaching client accountability engine…seed with four realistic clients.” Claude instantly creates three relational tables—Clients, Sessions, Action Items—and populates them with mock data.

Designing Dashboards for End‑User Interaction

Within the organization, add a dashboard (e.g., “Client Management Dashboard”). Use sections, columns, and views to display upcoming sessions, filter by status, and add buttons for quick actions. The visual layout mimics a simple web app that clients can navigate without manual table work.

Creating a Custom AI Agent

Add a custom agent, connect it to an Lerty MCP server, and select a model (e.g., GPT‑5.2 via OpenRouter). Paste system instructions that define the agent as an “executive assistant” able to query the database, summarize upcoming sessions, and generate prep briefs. Enable only necessary tool permissions (read/write) to keep costs low.

Selling and Scaling the Solution

Invite paying clients by email; once they accept, they gain instant access to their organization’s workspace. A $500‑month subscription covers Lerty’s agency plan and Claude Code fees, leaving profit. The 50 pre‑built systems provide a menu of niches to target for rapid client acquisition.