Pomo applies agent ideas to marketing
Ex‑Google DeepMind and Databricks engineers launched Pomo, a startup applying AI-agent principles to decision‑dense marketing functions. (theaiinsider.tech) The pitch frames marketing as a repeatable, decision‑heavy domain where agentic systems can automate many small optimizations while leaving higher‑risk choices to humans. (theaiinsider.tech)
A two-person marketing team can now get a machine to watch rivals, scan ad channels, and draft campaign moves before breakfast. That is the pitch behind Pomo, a San Francisco startup that announced $4.5 million in seed funding on April 8, 2026. (finance.yahoo.com) The round was led by Kindred Ventures, with Databricks Ventures, Seven Stars, SV Angel, Timeless Partners, and 645 Ventures also investing. Angel backers included Scott Belsky, Mehdi Ghissassi, and Massimo Mascaro, all veterans of Adobe or Google’s artificial intelligence groups. (finance.yahoo.com) Pomo was started by Praneet Dutta and Joe Cheuk, who met at Google. Dutta previously worked on applied generative artificial intelligence and reinforcement learning at Google DeepMind, while Cheuk held engineering roles at Databricks, Meta, and Google Cloud. (tmcnet.com) Their bet is that modern marketing is less like writing one big television ad and more like running an air-traffic control tower. Teams now juggle budget shifts, creative tests, compliance checks, and competitor reactions across many channels at once. (finance.yahoo.com) Most software in that stack still waits for a human to ask a question. Pomo says its system runs continuously instead, monitoring competitor moves, demand signals, creative trends, and channel performance without waiting for a prompt. (finance.yahoo.com) That is where the “agent” idea comes in. Instead of acting like a chatbot that answers one request at a time, the software is supposed to notice changes, rank the few actions that matter that day, and sometimes carry them out inside limits set by the team. (finance.yahoo.com) The company says it combines a brand’s own data, including customer relationship management records, commerce data, and advertising platform data, with outside market signals. That mix is meant to stop the system from suggesting generic tricks that ignore a company’s budget, brand rules, or sales goals. (finance.yahoo.com) On its website, Pomo says its agents “monitor your market 24x7,” surface gaps competitors miss, and generate both strategy and marketing assets. The homepage also frames the product as “Fortune 500 marketing for all,” which shows the customer target is smaller firms that want large-company tooling without a large-company staff. (usepomo.ai) Business Insider reported that the product can help marketers fine-tune ads and react to rivals, including spotting when a competitor’s product is trending on TikTok or Reddit. Dutta told the outlet the goal is to do the backend work “before they even wake up.” (africa.businessinsider.com) Pomo is launching into a crowded market filled with Adobe, Salesforce, and a long list of advertising tools, so its real test is not whether it can generate copy. Its real test is whether marketers trust a system to make dozens of small calls correctly, while humans keep the expensive calls on brand, budget, and legal risk. (finance.yahoo.com)
