Paul Koullick: AI-Driven Systems, Beyond Rules-Based Software and Revolutionizing Tax Filings

Welcome to the AI First Business Podcast with Tina, where we show you how teams, companies, and leaders are turning AI hype into ROI. In this episode, we dive deep with Paul Koullick, founder and CEO of Keeper, on his journey, insights, and the remarkable impacts of AI technology on his business model.

Introduction

Paul Koullick kicked off with a brief introduction. Having a background in product management at various San Francisco-based venture-backed startups, Koullick recounted meeting Tina at Amplitude before founding Keeper five years ago. Keeper is an AI-based tax filing software designed to handle individual tax preparation complexities that rules-based software struggles with.

AI in Tax Preparation

In 2019, when Keeper was launched, Paul mentioned grouping AI with other tech novelties like crypto and IoT. Although AI then seemed more of a method to reduce cost of goods sold (COGS) by scaling operations, Paul's co-founder and he built Keeper to solve a non-rules-based complex tax preparation system, leveraging machine learning on the backend without overtly branding it as an AI solution.

Rule-Based Systems vs. AI-Driven Systems

Paul explained the limitations of rule-based systems, which consist of numerous if-then statements working deterministically. While rules-based software peaked in the 2000s, complexities hit a point where they could no longer effectively handle multifaceted scenarios. An example cited is TurboTax, a significant rules-based system in tax preparation.

Keeper was designed to address inadequately served Americans unfamiliar with tax expertise or advice. The goal was to simplify their tax preparation process by automatically identifying and categorizing tax-deductible business expenses through AI.

AI Technology Landscape Today

The podcast moved to how the technological landscape has changed since Keeper's founding. Koullick noted a pivotal shift with the release of ChatGPT and GPT-4, offering broader and faster potential applications of AI. Before, AI applications were very niche and needed extensive infrastructure.

Paul mentioned GPT-4 Leapfrogging the need for extensive if-then statements by applying embedded logic, thus simplifying user interfaces. He discussed how AI could emulate an accountant's intuitive problem-solving approach, reducing user stress and making interaction more efficient.

Product Development During AI Advancements

Paul illustrated Keeper's product evolution through specific iterations influenced by the technological advancements in AI.

Iteration One: Chat-based Onboarding

In March 2023, Keeper experimented by replacing UI with chat-based onboarding but found that it performed worse. Users didn’t prefer typing with thumbs and didn’t trust chatbots.

Iteration Two: UI and AI Collaboration

Keeper then tested combining UI and AI, allowing AI to provide context-specific assistance within UI workflows. This enhanced experience showed improvement but still faced trust issues, partially because of AI’s historic perception as unreliable (e.g., Clippy).

Iteration Three: Integrated Group Chat

The latest iteration presented a group chat concept where an AI agent assists alongside a human bookkeeper. The AI provides quick, reliable answers, while humans step in when necessary. This setup demonstrated a balanced approach of leveraging AI while maintaining user trust without emphasizing AI’s novelty.

Key Takeaways and Advice

Paul shared multiple recommendations on integrating AI technology, team building, and managing user experiences:

AI-First Approach: He advised younger startups to rethink problems that have been unsolved with traditional software by applying AI innovatively.
Balancing AI and UI: Highlighted the importance of integrating AI without overshadowing it, focusing on the problems it solves.
Employment Efficiency: He emphasized that AI will enhance job efficiency and effectiveness rather than replace jobs entirely.

Hiring and Team Recommendations

For those transitioning to AI-focused careers, Paul encouraged skepticism and thorough evaluation of potential employers. He recommended platforms such as AngelList (now Wellfound) and YC's Work at a Startup to find genuine, mission-driven companies.

Conclusion

The session wrapped up with reflections on the real-world challenges and successes of introducing AI into Keeper’s system. Paul reiterated that messy, real-world product development cycles are the norm and that innovations like AI can offer substantial improvements in efficiency and user experience when strategically implemented.

Keywords

AI
Rules-Based Systems
Keeper
Tax Preparation
Machine Learning
GPT-4
ChatGPT
Product Development
User Experience
Iteration
Team Building
Employment Efficiency

FAQs

Q1: What is Keeper? Keeper is an AI-based tax filing software that simplifies the complexities of individual tax preparation by using machine learning.

Q2: How does AI help in tax preparation for Keeper? AI helps Keeper by identifying tax-deductible business expenses and categorizing them automatically, making the process easier for users who lack tax expertise.

Q3: What major change did GPT-4 bring to companies like Keeper? GPT-4 enabled companies like Keeper to bypass extensive if-then logic by embedding complex rules within AI systems, improving user interactions significantly.

Q4: How did Keeper improve user trust in its AI systems? Keeper improved user trust by integrating AI assistance without overtly emphasizing the technology, demonstrating value through effective problem-solving within a familiar UI framework.

Q5: What should new startups consider when integrating AI? New startups should rethink unsolved problems using AI, avoid focusing only on AI's novelty, and focus on actual problem-solving and user benefits.

Q6: What advice does Paul give to job seekers in the AI field? Paul advises job seekers to thoroughly evaluate potential employers, ask detailed questions about a company’s financial health and strategic vision, and use platforms like AngelList and YC’s Work at a Startup to find genuine opportunities.