Why StackOverflow usage is down 50% — with David Hsu of Retool

Episode Summary

Executive Summary: David Su, cofounder/CEO of Retool, discusses the company’s origin, fundraising philosophy, developer-first positioning, and AI strategy. He argues that startups should optimize for customer value and team ownership rather than hype, and that AI’s biggest near-term impact will come from embedding automation into internal business workflows—not just chatbots. He also shares survey findings on AI adoption, sentiment, and hiring impacts.

Main Topics: Retool’s origin and founder philosophy (Priority: 5/5): Su explains how Retool skipped its first YC demo day because the product lacked traction and he preferred to underpromise and build something real before presenting. He frames this as a commitment to authenticity over social validation. Capital efficiency and fundraising strategy (Priority: 5/5): He argues that raising less money at lower valuations can be better for employees and long-term company health because it reduces dilution and avoids valuation traps that make future fundraising harder. Developer-first product positioning (Priority: 5/5): Retool is positioned as a tool for developers building internal software, not as “low-code.” Su says developers want building blocks, not finished solutions, and that internal tools are a huge but underappreciated software category. AI product strategy at Retool (Priority: 5/5): Su says Retool’s AI focus is less about making coding marginally faster and more about helping developers build AI-enabled applications and workflows. He sees automation, not chat, as the major productivity unlock. Open source and infrastructure choices (Priority: 4/5): He prefers open-source or off-the-shelf infrastructure where Retool lacks a unique advantage, citing vector databases and AI tooling as fast-moving areas where partnering is safer than building everything in-house. State of AI survey findings (Priority: 4/5): Su reviews survey results showing mixed sentiment, limited production adoption, and strong interest in AI’s impact on jobs and workflows. He emphasizes that most current value is in internal use cases. Philosophical reflections on AGI and intelligence (Priority: 3/5): The conversation closes with a philosophical discussion about intentionality, AGI, and whether machine intelligence should be judged by outputs or by internal mechanisms, drawing on analogies from evolution and Hofstadter.

Key Arguments: Startups should focus on creating real customer value before seeking social proof or fundraising optics. Raising more money at a higher valuation is not always better; it can hurt employees through dilution and create future financing constraints. Retool’s core audience is developers, who prefer reusable primitives and control over a turnkey solution. AI’s biggest business impact will come from automating workflows end-to-end, not just improving chat interfaces or coding speed by a small percentage. Most AI value today is in internal enterprise use cases, where hallucinations are more tolerable because humans can verify outputs. AI adoption will likely be engineering-led, with business leaders identifying use cases and engineers implementing them. Open-source infrastructure is likely to win in fast-moving AI tooling categories because it is inspectable, adaptable, and less risky than betting on one vendor. Hiring should test fundamentals and critical thinking, while still rewarding competent use of AI tools like Copilot and ChatGPT.

Data Points: YC demo day delay: 6 months - Retool skipped its initial YC demo day and spent six months finding traction before presenting. Initial team size at Series A: 3–4 people - Su says Retool raised a Series A while still extremely small. Revenue at Series A: $1 million - Retool raised its Series A around $1M in revenue. Early funding runway: 5–6 years - He says the company had long runway because founders paid themselves $30K–$40K/year. Founder compensation: $30K–$40K/year - Used to illustrate how far early funding went. Survey sample size: ~1,600 people - Retool’s State of AI survey conducted last August. AI is overrated: 52% - Survey respondents who said AI is overrated. Stack Overflow usage among respondents: 58 people - Survey respondents who reported using Stack Overflow. Stack Overflow usage attributed to AI tools: 94% - Of Stack Overflow users, 94% said they used it because of Copilot and ChatGPT. AI impact on jobs: operations: 8/10 - Operations workers rated AI as highly likely to affect their jobs. AI impact on jobs: designers: 6.8/10 - Designers were the lowest-rated group for expected AI job impact. Companies with AI in production: 27% - Survey respondents saying AI was in production use. Internal AI use cases among production adopters: 66% - Most production AI use was internal rather than customer-facing. Companies making engineering interviews harder: 45% - To account for candidates using Copilot/ChatGPT. Retool customer AI usage: 95% developers - Su says most Retool customers are developers. GPT-3.5 NPS: ~14 - Su cites a low NPS for GPT-3.5 in Retool’s survey results. GPT-4 NPS: ~45 - Su cites a much higher NPS for GPT-4 than GPT-3.5. OpenAI usage share: 80% - He says roughly 80% of customers use OpenAI models. Potential developer base: 25–30 million - Su estimates the number of developers Retool wants to reach for broad ubiquity. Amazon Retool adoption: 11 business units - He says Retool is used across 11 Amazon business units. AI workflow output example: 10,000 designs/day - A clothing manufacturer uses Retool and AI to generate and review large volumes of design ideas.

Pivotal Quotes: "We really didn't want that. And so we chose actually not to present the demo day, mostly because we felt like we didn't have anything substantial underneath." — David Su: Explaining why Retool skipped its initial YC demo day. "The point of starting a startup is basically you have to create value for customers." — David Su: On why Retool prioritized customer value over startup theater and rapid hiring. "What we think the next big breakthrough in AI is, is actually automation. It's not just, like, oh, I have a problem, let me go to a chatbot and solve it." — David Su: Describing Retool’s AI strategy and why workflows matter more than chat.

Implications: For founders, the episode argues for disciplined fundraising, small teams, and customer-first execution. For builders, it suggests AI’s biggest near-term opportunity is internal workflow automation, especially in enterprise software, with engineering-led adoption and open infrastructure choices.