From industrial supply chains to the cockpit, the first quarter of 2026 brought a wave of San Francisco startups stepping out of the shadows with real products, and real customers.
Kana: The marketing platform built by people who have actually done this before
Marketing technology is crowded to the point of absurdity. Every major platform, every incumbent, and a long tail of AI startups all claim to make marketers’ lives easier. Kana‘s opening argument is that most of them are built by people who have never run a marketing operation. Co-founders Tom Chavez and Vivek Vaidya have spent more than 25 years building in this space across four ventures, including Rapt, acquired by Microsoft, and Krux, acquired by Salesforce. Kana is their fourth company together, incubated inside their startup studio super{set} before launching in February with $15 million in seed funding led by Mayfield. The platform deploys loosely coupled AI agents that marketers can tailor on the fly and integrate into existing tools, covering data analysis, audience targeting, campaign management, and media planning. The pitch is not that AI will replace marketing judgment. It is that AI can handle everything in between while humans stay in the loop on decisions that matter.
Freehand: The supply chain back office that runs itself
Enterprise supply chains run on manual work that is invisible to everyone until it breaks. Invoices get matched by hand. Purchase orders are reconciled by human teams. Exceptions sit in inboxes. Freehand, formerly operating as Pando, launched at the Manifest 2026 conference in February as a new company with a direct mission: replace that operational grind with autonomous AI teams. Founded by Nitin Jayakrishnan and Abhijeet Manohar and headquartered in San Francisco, Freehand’s platform reads unstructured documents, emails, and internal chats, reasons across contracts and policies, and executes decisions directly inside ERP, procurement, and finance systems. The company launched with Fortune 500 customers already live and reported early results showing accounts payable cycle time reductions of 80 to 90 percent and manual procurement effort reductions of 30 to 50 percent. The technology builds on Pi, a production-proven AI agent for freight and logistics that the same team developed previously, earning recognition from Gartner, TIME, and the World Economic Forum.
Emanate: The AI revenue engine for the industries that build America
Most AI has landed in software companies. Kiara Nirghin wants to bring it to the industries that still run on phone calls and gut instinct. Emanate emerged from stealth in February backed by Andreessen Horowitz’s American Dynamism fund, with additional investment from Peter Thiel, Alexis Ohanian, and other prominent angels. The San Francisco-based company deploys autonomous AI agents for industrial materials companies: distributors, service centers, and suppliers operating in a $5 trillion market that has historically resisted digitisation. Nirghin is a Thiel Fellow, a former Grand Prize winner of the Google Science Fair, a Stanford AI researcher, and the youngest board member of the Google Impact Fund. Her argument is pointed: industrial companies lose revenue every day because they cannot respond to inbound demand fast enough and price on instinct rather than data. Emanate’s agents handle inbound conversion, customer relationship management, and prospect research around the clock. The company projects revenue growing nearly 50-fold as it deepens partnerships with leading industrial distributors.
Whirl AI: The enterprise IT company built by someone who lived the problem
Enterprise AI pilots fail constantly. The reason is usually the same: the AI does not understand the environment it is operating in. Sunny Bedi spent two decades as CIO and IT leader at VMware, NVIDIA, and Snowflake watching this problem compound. Every customization, every workaround, every configuration decision made by someone who left the company three years ago exists nowhere a machine can find. Whirl AI, which he founded and which emerged from stealth on March 31 with $8.9 million led by ICONIQ, is built to fix that. The platform continuously captures and maintains context across enterprise systems, turning it into a living, searchable knowledge base that AI agents can actually use. ICONIQ’s decision to lead a seed round is notable in itself. The firm is primarily known for growth-stage investments. Partner Matt Jacobson, who worked with Bedi during his time at Snowflake, described the problem as one Bedi had lived firsthand at scale, not one he had read about.
Navi AI: The flight instructor that never leaves the cockpit
Pilot training has a data problem. Every training flight generates cockpit audio, aircraft telemetry, weather conditions, and traffic information simultaneously. Instructors currently synthesize all of it manually during post-flight debriefs, a process that is slow, inconsistent, and dependent on individual memory. Navi AI, founded in 2024 and headquartered in San Francisco, emerged from stealth in March with $6.7 million in total funding. Its platform installs a small device in the aircraft, ingests all available cockpit and flight data in real time, and produces detailed, moment-by-moment debriefs automatically, aligned to each flight school’s training syllabus. CEO Nikola Kostic describes the approach as what the National Transportation Safety Board does after a crash, applied at scale before anything goes wrong. The platform has been trained on more than 100,000 real flight hours and is already in use or under evaluation at Embry-Riddle Aeronautical University, the University of North Dakota, Purdue University, and the U.S. Air Force Test Pilot School at Edwards Air Force Base.
All five companies are headquartered in San Francisco.
