Imagine onboarding a new employee solely through written materials—emails, documents—without any conversation. Even the smartest person would miss critical context and information, especially if excluded from meetings or one-on-ones, leaving a significant information gap. This challenge inspired Co-founders David Gu and Amanda Zhu to create Recall. ai, an API infrastructure company enabling thousands of developers to capture and process conversation data. Despite AI agents’ growing sophistication, they face the same hurdle as that hypothetical employee: without access to the conversations where most company context resides, even the best AI struggles to navigate tasks as smoothly as humans. Recall. ai addresses this vital infrastructure need for companies like HubSpot, ClickUp, monday. com, and PagerDuty by solving a painful problem the founders experienced firsthand. Our case study series explores journeys of SaaS and AI leaders who successfully built AI-native solutions, distilling their insights into actionable frameworks. Based on our belief in Recall. ai, we spoke with CEO David Gu to explore how they turned infrastructure pain into a business—a vendor they wished they had. Recall. ai’s journey: From pain to platform Before Recall. ai, David and Amanda ran a video conference recording product for user researchers and product managers. Though demand grew, they spent about 80% of their time managing the tedious backend work of reliably capturing meeting data. They faced a “no-win situation”: good infrastructure goes unnoticed while poor infrastructure causes catastrophic data loss and unhappy customers. David explained, “When you’re a product company, you can’t win with infrastructure. If it works perfectly, customers want features, not backend improvements; if it fails, customers are mad. ” The technical challenges were immense. Meetings cluster tightly (e. g. , 9 AM daily standups on the West Coast), requiring infrastructure that scales instantly, processes data in real-time, and ensures zero data loss. Any failure meant losing a client. David realized they could neither over-invest (invisible returns) nor under-invest (risk churn), leaving them stuck in painful frustration. Yet this pain became their signal. The inspiration: Infrastructure expertise meets AI opportunity In 2022, as GPT-3. 5 and other large language models (LLMs) advanced, two-person startups began building products impossible only years before. David noted, “LLMs don’t solve the painful build and maintenance of infrastructure. ” With more companies needing conversation data for AI products, infrastructure pain was about to be widespread. Recognizing their top-tier expertise in meeting infrastructure, David and Amanda pivoted to build an infrastructure-as-a-service company. They reached out to former competitors—responses ranged from silence to relief: “This problem really sucks. ” Thus, Recall. ai was born. Building for builders: From infrastructure to product David and Amanda began selling their extracted infrastructure before perfecting the product, launching an alpha program within a month.

Their experience running an end-user product gave them unique insight: they understood not only developer customers but also those customers’ end users. David emphasized, “As an infrastructure company, you need to understand the whole chain of causality from your infrastructure to end user value. ” They advised developers on user needs, cost structures, and product optimization, leveraging lived experience. Recall. ai’s product evolved in waves: - Meeting Bot API: Core API for recording video conferences, addressing immediate needs. - Desktop Recording SDK (late 2025): Embedded AI note-taking style recording for any app, gaining hundreds of thousands of installs rapidly. - Mobile Recording SDK (coming soon): Extending capabilities to phone calls and in-person meetings. Usage-based pricing for vendor-customer alignment Recall. ai naturally adopted usage-based pricing, aligning costs with customer success. David explained, “If your product thrives and processes more data, costs increase; if you’re early-stage with low volume, it costs less. ” This model ensures: - Transparent, predictable pricing scaling with value. - No sticker shock from surprise overages. - Avoidance of seat-based fees that limit adoption. - Simplicity without complex tiering, naturally segmenting customers by scale. Their pricing works for both seven-figure contracts and self-serve clients, harmonizing elasticity with predictability. Contrarian sales-led growth Though developer infrastructure products typically adopt community-first, self-serve strategies, Recall. ai chose the opposite. For three years, every customer engaged directly with founders David or Amanda, building deep market understanding unattainable through analytics or support tickets alone. “The real value isn’t revenue; it’s learning, ” David stated. They rigorously asked customers about their goals, motivations, and how Recall. ai could help. This engagement across diverse industries—telemedicine, finance, sales, customer success—uncovered regulatory nuances and varying user comfort with recording. In early 2025, Recall. ai launched self-serve sign-ups and now gains thousands of weekly customers. This upfront investment in customer conversations paid off, attracting industry-leading clients and confirming a shift in how software harnesses conversation data. The no-win situation that became opportunity Back in 2020, their video recording tool saw surging demand during COVID-induced remote work, but infrastructure consumed 80% of their efforts. David characterized their dilemma: infrastructure excellence is invisible, yet inadequacy leads to failure and churn. The real challenge was managing massive concurrent meeting loads that scale like daily “Black Friday” events, demanding flawless, real-time processing with zero data loss. This unsolvable tension felt like a "no-win situation, " fostering frustration but also clarity on where the real problem—and opportunity—lay. Conversation data: The foundation of AI’s future David’s closing insight reframes AI agent development: despite advances, AI still requires access to the same conversation context humans rely on for automation. Without capturing and processing conversation data effectively, automation falls short of its promise. As AI becomes integral across industries, mastering conversation data infrastructure is critical. Companies that excel here will build genuinely helpful AI products, transcending current limitations. About Recall. ai Founded in 2022 by David Gu and Amanda Zhu, Recall. ai offers a unified API to capture audio, video, transcripts, and metadata from video conferences and in-person meetings. Now powering infrastructure for over 3, 000 companies, Recall. ai is at the forefront of enabling AI products that truly understand and leverage conversation data. Learn more at Recall. ai.