He earns $5k building agents

Business model & core customer promise

• The speaker sells agent-building as a managed service to SMBs / mid-market businesses, targeting $5,000/month retainers.
• The “magical moment” for conversion is showing the agent operating a computer directly (computer-use agent behavior), so non-technical business owners immediately understand value.
• Positioning:
  • Customers aren’t paying for the open-source agent itself (e.g., Hermes/OpenClaw are free).
  • Customers are paying for expert implementation + reliability + ongoing management.

What makes customers pay $5K/month (offer design)

All-inclusive pricing to reduce buying friction

• $5K/month
• Unlimited tokens
• Unlimited agents
• Unlimited automations

Why this structure works

• It avoids “variable pricing noise” (tokens/security/usage limits).
• “Unlimited” is positioned as reassurance:
  • Customers can start with 1 agent and expand later without repricing.

Retainer logic

• The agency/consultant builds and runs workflows the customer can’t reliably implement themselves.
• Workflows may also break without expert tailoring, so the retainer covers ongoing operation and fixes.

Concrete examples / case studies

Insurance software company → insurance agencies

• Use case: Transfer insurance policy/customer data from a competitor platform into the customer’s platform using a Hermes computer-use agent.
• Value impact: Manual data scraping was quoted at $10K–$30K; the agent automates it.

Design agency workflow automation

• The team built an initial agent quickly to automate steps in producing customer brochure templates, using the agency’s existing SOP.

Scaling anecdote (Orgo user)

• A user reportedly has 4 customers × $5K/month = $20K/month.
• Target: $1M/year (by end of year), implying roughly 20 customers.

Go-to-market (GTM): how they acquire customers

Warm outbound > cold outbound

For early customers:

• They contacted about 50 people from their phonebook (business owners or decision-adjacent contacts) on day one.
• Results:
  • 7 booked calls that week
  • $3,500 in the first week
  • Closed their first client

Referral-style outreach script

Instead of asking, “Will you benefit from AI?”, ask:

• “Do you know someone who could benefit from [XYZ]?”

Content strategy

• Primary platform: Twitter
• Demonstrates “meta” agent behavior—for example, an agent like Dewey using the builder agent to spin up new agents for customers automatically.

Outreach conversation hook

• Ask: “What’s your first itch?”
• Then determine:
  • effort to automate
  • relevance to outcomes
  • value generated
• Start with the lowest cost / highest value automation and deliver within 7 days.

Fulfillment playbook (how to deliver quickly)

Speed targets

• Spin up first agent within 24 hours
• Deliver first working automation within 7 days

Operational process

• Build from templates and SOPs
  • Businesses with documented SOPs are easier because automation can follow existing procedures.
• Use a “computer-first” workflow:
  • Deploy the agent into real cloud desktop computers so it can perform end-to-end computer actions (clicking/typing/browsing), not just API calls.

Orgo templates for faster deployment

• Orgo templates let builders launch pre-configured computers with:
  • Claude + Hermes + OpenClaw pre-installed
• Setup time:
  • In ~10–20 seconds, a full computer environment is ready.

Tool stack & why each component matters (execution architecture)

Agent runtime / harness

• Hermes is the primary harness (selected for reliability and growth/support).

Earlier alternative and why it changed

• OpenClaw had early reliability issues.
• Hermes resolved those issues for business usage.

Agent communication channels (agent-to-user interfaces)

Agent email: agentmail

• Used to generate an inbox via API.
• Custom domain vs default domain:
  • The speaker sticks to the agentmail domain for perceived authenticity / “agent-ness.”

Agent iMessage: “agent iMessage” service

• Price mentioned: ~$150 for an iMessage number.
• Scaling detail: one iMessage number supports up to ~6,000 conversations.

Slack vs Telegram vs iMessage

• Slack: best for company-wide use, but often requires Slack API setup for integration.
• Telegram: suited to executive-assistant style / more personal assistant experiences.
• iMessage: emphasized as “winning” due to simplicity and low friction.

Memory & knowledge organization

• honcho.dev
  • “Memory layer” for fast/discrete facts (short-term / readily available context).
• Obsidian
  • Treated as long-term project/knowledge base storage.
  • Used for ongoing projects and organizing context (e.g., customer computer IDs in Orgo).

Memory distinction framework

• honcho.dev: discreet facts (e.g., birthday)
• Obsidian: project-based, interconnected long-term context

Tool connectors

• Composio
  • One connector to manage tool access/connectors.
  • Solves pain from earlier approaches that required managing many API keys individually.
  • Benefit: simplified security/permissions handling for agents using other apps.

Payment rails

• Agent cards
  • Mentioned as an early approach; the speaker is experimenting.
• Longer-term expectation:
  • Agent payments will mature later (Stripe CEO suggested 3–4 years).
• Interim options:
  • give agents a card, or use API-friendly systems.

Retention & customer success (ongoing value delivery)

Operating rhythm

• Weekly 1:1 check-ins with every customer after onboarding.
• Retention mechanism:
  • provide quantitative outcomes each cycle, such as:
    • “Used to take X”
    • “Now takes Y”
    • problem → solution evidence

“More features / more wow” strategy

• Continuously add “shiny” relevant improvements, for example:
  • executives requesting Obsidian integration
  • adopting newer models (e.g., referencing a “newest Grock model” as staying current)

Ongoing operational duty

• Security hardening
• Secret management
• Keeping integrations working when they break (maintenance and fixes)

Key KPIs / targets mentioned

• Revenue / pricing

  • $5,000/month retainer
  • Example: $20,000/month from 4 clients
  • Scaling target: $1,000,000/year by end of year (needs ~20 customers)

• Sales performance (early stage)

  • 50 warm contacts → 7 booked calls in a week
  • $3,500 in first week

• Delivery SLAs

  • First agent live in ≤24 hours
  • First automation working in ≤7 days

• Use-case value benchmark

  • Insurance manual scraping: $10K–$30K vs automated with Hermes

• iMessage scaling

  • $150 for an iMessage number
  • One number supports up to ~6,000 conversations

Presenters / sources

• Andrew Warner (interviewer)
• Nick Vasil(i)escu (guest; Orgo user/builder referenced throughout)
• Video is presented by Zapier (AI automation company)