How AMP Uses AI Agents to Run a Lean Startup
AMP CEO and co-founder Rachel Woods has explained how the eight-person AI company uses structured AI-agent workflows to manage business operations without expanding its workforce at the same rate as its workload. By combining documented processes with automation, AMP supports sales, marketing, customer service and administrative activities while maintaining a lean operating team.
Key Takeaways
- AMP operates with four full-time employees and four part-time contractors
- Rachel Woods estimates the company would need at least four to five additional workers without its AI-supported operations
- The company organizes AI agents into repeatable workflows called playbooks
- Each playbook defines tasks, inputs, expected outputs, decision points and quality standards
- AMP reported spending $8,510 on Claude Code in September for background automation, separate from individual AI subscriptions
AMP Uses AI Agents for Small Business Operations
AMP has structured its daily operations around AI agents that handle defined responsibilities across several business functions. Woods described an operating model in which automation supports the existing team rather than requiring the company to hire additional employees for every recurring task.
The company currently has four full-time employees and four part-time contractors. According to Woods, maintaining the same level of operations without AI support would require at least four to five more workers. Her estimate illustrates how the company uses automation to manage responsibilities that might otherwise increase staffing requirements.
AMP’s approach reflects a broader shift toward AI-supported startup operations, where founders use technology to manage internal documentation, marketing and repetitive administrative work. Rather than relying on automation for isolated tasks, AMP builds repeatable processes that can be tested and improved over time.
The company organizes these processes into playbooks that outline what an AI agent must accomplish and how its work should be evaluated. This structure allows AMP to apply automation consistently while maintaining human involvement in the broader business.
For Woods, the objective is not to remove employees from operations but to create a system in which a relatively small team can manage a wider range of responsibilities. Documented procedures establish expectations for AI agents, while employees remain responsible for supervising and improving the workflows.
How Rachel Woods Structures AMP’s AI Agent Playbooks
AMP’s automation strategy centers on repeatable workflows known as playbooks. Woods explained that each playbook provides detailed instructions for completing a specific assignment, giving AI agents clear expectations rather than unrestricted responsibility for an entire business function.
Every playbook defines the task, the information required to complete it, the expected output, relevant decision points and the quality standards the result must meet. These components establish a consistent framework for carrying out work and reviewing its accuracy.
The input requirements identify the information an agent needs before beginning a task, while expected outputs specify what it must produce. Decision points guide the workflow when different conditions arise, and quality standards provide criteria for determining whether the work is acceptable.
Woods also described a testing process in which AMP evaluates AI workflows against human performance. The company revises its instructions when necessary before allowing agents to operate with less supervision. This process helps identify weaknesses in a workflow before they become part of routine operations.
One application involves an AI agent that reviews sales calls against company standards and answers employee questions using internal guidance. The system combines performance assessment with access to established company information, allowing employees to reference internal procedures while reducing some of the manual work involved in reviewing calls.
This example demonstrates how AMP uses playbooks for both operational tasks and internal support. The same documented approach can be applied to different responsibilities while preserving clear instructions, measurable expectations and opportunities for human review.
AI Automation Supports Sales, Marketing and Customer Service
AMP uses AI-supported workflows across sales, marketing, customer support and administration. Although these functions involve different activities, the company applies the same playbook structure to organize tasks and establish expectations for automated work.
In sales, AMP has described an agent that evaluates calls against company standards and responds to employee questions using internal guidance. Other businesses are developing AI agents for sales to support activities such as lead generation, outreach and customer engagement, demonstrating how specialized automation can address different stages of commercial operations.
AMP also includes marketing and customer support among the functions supported by its AI workflows. Administrative responsibilities form another part of the company’s operating model, allowing employees to manage multiple areas of the business with assistance from documented automated processes.
The arrangement does not mean that AI agents independently manage these departments. Instead, the company assigns specific tasks to automated systems and establishes instructions for completing them. Human employees and contractors remain part of the operating structure, particularly when workflows require testing, adjustment or oversight.
Woods’ estimate of four to five additional workers provides context for the potential staffing implications. Rather than increasing headcount solely to accommodate routine work, AMP uses automation to expand the capacity of its existing team.
The company’s approach illustrates how smaller organizations can integrate AI into established responsibilities without treating the technology as a complete replacement for employees. Its operating model depends on combining automation with documented processes and human judgment.
AMP Reports $8,510 in Monthly Claude Code Spending
AMP’s reliance on AI agents also involves direct technology costs. Woods reported that the company spent $8,510 on Claude Code in September for background automation, excluding individual AI subscriptions.
Photo Credit: Unsplash.com
The distinction is important because the reported figure represents a specific category of automation spending rather than the company’s entire AI budget. Individual subscriptions were accounted for separately, meaning the total cost of its AI tools was higher than the disclosed Claude Code expenditure.
The spending also provides context for Woods’ staffing estimate. Although she believes AMP would require at least four to five additional workers without its AI-supported processes, the company must still invest in the technology used to maintain those workflows.
AMP’s operating model therefore involves balancing staffing requirements with software expenses. Its employees and contractors work alongside systems that perform defined tasks, while the company continues to allocate resources toward automation and quality control.
The reported costs offer a practical perspective on lean startup operations. AI can reduce the need for additional manual capacity in certain workflows, but maintaining reliable automated processes requires financial investment, documented instructions and ongoing evaluation.
Frequently Asked Questions
How does AMP use AI agents in its business operations?
AMP uses AI agents to support sales, marketing, customer service and administrative activities. The company organizes tasks into documented playbooks that specify instructions, expected outputs and quality requirements.
Who founded AMP?
Rachel Woods is AMP’s CEO and co-founder. She has described the company’s AI-supported operating model and how documented workflows help its small team manage business responsibilities.
How many employees and contractors work at AMP?
AMP operates with four full-time employees and four part-time contractors. Woods estimates that the company would need at least four to five additional workers without its AI-supported operations.
What are AI playbooks in AMP’s operating model?
AI playbooks are repeatable workflows containing instructions for specific assignments. They establish the required inputs, expected outputs, decision points and quality standards that guide how AI agents perform their tasks.
How much did AMP spend on Claude Code in September?
Woods reported that AMP spent $8,510 on Claude Code for background automation in September. The amount excluded individual AI subscriptions and should not be interpreted as the company’s total AI spending.
