Gusto is a B2B payroll & HR platform used by Small and Medium-Sized Businesses (SMBs) in the United States. The platform currently has an AI assistant (Gus) that is built on a multi-agent architecture. Gus incorporates specialized sub-agents that handle user queries related to Reporting, PTO Approval, and FAQs. Each specialized agent added increases Gus' capabilities and improves the customer experience. This project explores the development of a new specialized agent by researching customer pain points and defining product requirements for an agentic solution for a specific pain point.

We identified poor customer support responsiveness for payroll issues as a critical customer pain point through data analysis of user reviews. When payroll incidents occur—such as delayed direct deposits or ACH holds—users are stressed and require urgent clarification and resolution, but existing support processes are too slow. We propose the addition of a Payroll Support Agent for handling payroll-related user queries. This agent aims to provide instant, automated, and contextual resolution for these issues, reducing response times and improving customer trust.

This repository contains all the project files and prompts used to define the product requirements and technical specifications for the proposed Payroll Support Agent. You can find more details on this project here.

• gusto_research.md — Overview of the Gusto platform, its payroll product ecosystem, core features, and typical customer workflows. Provides foundational context on how users interact with Gusto's payroll functionality.
• gusto_agent_research.md — In-depth analysis of Gus, Gusto's existing AI assistant. Documents its multi-agent architecture, orchestration layer, user interaction model, and technical design patterns relevant to integrating the new Payroll Support Agent.
• gusto_pain_point_analysis.md — Comprehensive analysis of customer pain points derived from 397 payroll-related reviews across App Store, Google Play, and Trustpilot. Identifies "poor customer support responsiveness" as the top issue (50.1% of reviews) and maps pain points to potential AI intervention opportunities.
• gusto_reviews_analysis.csv — Raw dataset containing customer reviews from App Store, Google Play, and Trustpilot, with sentiment scores and categorized feedback. Used as the foundation for pain point identification.
• reviews_scraper_gusto.py — Python script for extracting and aggregating customer reviews from multiple platforms. Includes sentiment analysis and data preprocessing functionality.

• gusto_payroll_agent_requirements.md — Comprehensive requirements document covering product requirements, user flow specifications, agent orchestration logic, and high-level system architecture. Serves as the foundation for PRD and technical specification development.
• gusto_payroll_agent_prd.md — Product Requirements Document (PRD) defining what the Payroll Support Agent will accomplish and why. Includes problem statements, goals, success metrics, user scenarios, and functional requirements.
• gusto_payroll_agent_tech_spec.md — Technical Specifications Document detailing how the agent will be built. Covers system architecture, orchestration flow, API integrations, data models, and implementation details.
• gusto_payroll_agent_user_flow.png — Visual diagram illustrating the end-to-end user experience when interacting with the Payroll Support Agent through Gus. Shows entry points, intent detection, routing logic, and resolution paths.
• gusto_payroll_agent_architechture.png — High-level system architecture diagram showing the technical components, data flows, and orchestration logic that enable the Payroll Support Agent to function within Gus's multi-agent ecosystem.
• gusto_payroll_query_examples.json — Collection of 50-75 realistic user queries representing different intents (initial issues, follow-ups, escalations). Used to train Gus's intent detection and routing logic.
• gusto_payroll_query_keywords.json — Structured list of 30-40 keywords with relevance scores that signal payroll support queries. Helps the orchestration layer identify when to route requests to the Payroll Support Agent.
• Asana Spec Template.md — Template used to structure the PRD, ensuring consistency with standard product documentation practices.

Note: There are also prompt files with the initial and optimized prompts (_prompt_initial.txt, _prompt_optimized.txt) used to generate the files above. Refer to the Workflow section for more details

This repository demonstrates how AI can enhance the process of distilling product requirements and technical specifications. It serves as a practical example for product managers and product teams looking to leverage AI assistance in their product development workflow.

Key Takeaways

• Requirements Definition: We can use AI to transform raw user data into structured PRDs and technical specifications, significantly accelerating the requirements gathering process.
• Data Source Flexibility: While this example uses customer review data, the same methodology applies to other qualitative data sources such as user interviews, survey responses, support tickets, or usability studies.
• Critical Considerations: When building AI agents or introducing AI capabilities into their products, product teams must consider the following:
  • Multi-agent orchestration and routing logic
  • Intent detection and context management
  • User experience consistency across agent handoffs
  • System architecture for scalable AI integration
• Human Review: All project files represent edited and refined versions of raw AI-generated output. Since AI can hallucinate or misunderstand instructions, human review is critical to ensure accuracy, feasibility, and alignment with business goals.
• Structured Documentation: All documents are in Markdown format for easy reading, version control, and team collaboration.

You can find a detailed version of this workflow here. This is just an overview of the process.

1. Platform Research: Conducted comprehensive research on the Gusto platform to understand its payroll product capabilities, user workflows, and how customers interact with payroll features. This established a foundational context for identifying where an AI agent could provide value.

2. Existing Agent Research: Analyzed Gus, Gusto's current AI assistant, to understand its multi-agent architecture, orchestration patterns, user interaction model, and technical infrastructure. This research informed integration requirements and ensured the new agent would seamlessly fit into the existing ecosystem.

3. Customer Review Extraction: Extracted customer reviews from App Store, Google Play, and Trustpilot using a Python-based scraper.

4. Pain Point Analysis: Analyzed customer reviews to identify and prioritize pain points related to payroll functionality. Discovered that "poor customer support responsiveness" was the top complaint, making it the primary target for an agentic solution.

1. Requirements Definition: Synthesized research findings to define comprehensive product, user flow, orchestration, and system architecture requirements. These requirements provide the foundation for creating the PRD and technical specifications.

2. User Flow and Architecture Visualization: Created visual diagrams to illustrate the end-to-end user experience and the underlying system architecture. These diagrams help stakeholders understand both the customer-facing interaction and the technical implementation.

3. PRD and Technical Specifications: Generated a Product Requirements Document (PRD) defining what the agent will accomplish and why, along with a Technical Specifications Document detailing how it will be built.

4. Query Examples and Keywords: Developed structured datasets of example queries and keywords to train Gus's intent detection and routing logic. These resources help the orchestration layer identify when to route requests to the Payroll Support Agent.

For more information on the methodologies used in this project:

• Meta Prompting Guide — Learn how to optimize prompts using meta-prompting techniques
• Extracting App Reviews — Guide to extracting and analyzing customer reviews from app stores