Zaro.ai has emerged from stealth with USD $5.1 million in pre-seed funding, led by Cherry Ventures.
The London-based company was founded by Michael Bajwa and Qian Zheng, who previously worked on AI agent products at Convergence and later at Salesforce after Convergence was acquired. Zaro aims to give businesses a single workspace that connects company data, AI agents, and custom applications.
Other investors in the round include Thomas Wolf of Hugging Face, Thomas Dohmke of GitHub, Mandeep Singh of Trouva, Charlie Songhurst, and Convergence co-founders Marvin Purtorab and Andy Toulis.
The startup enters a crowded market for enterprise AI software, where tools often sit in separate layers for application building, data context, and agent workflows. Zaro argues that this fragmented approach spreads automation across multiple products instead of keeping it within a single internal system.
Bajwa and Zheng built their case for the new company on their experience developing agents before the technology became widely adopted. Bajwa was also Convergence's first hire and helped grow the business before its sale to Salesforce, where the team worked on Agentforce.
The product is designed so that interactions, workflow history, files, and operational decisions remain within a company's own context layer rather than being held mainly in a software vendor's infrastructure. The platform is also model-agnostic, allowing businesses to move between providers without becoming tied to one vendor.
According to Zaro, the system routes simpler tasks to lower-cost AI models and uses more advanced models for more complex work. It says this approach can cut costs by about ten times compared with deployments that rely only on frontier models.
"We built agents that worked flawlessly in isolation, and watched them fail collectively. The intelligence never compounds because the context never carries over. Zaro is the platform that fixes that," said Michael Bajwa, Chief Executive Officer, Zaro.
Zheng summarised the company's technical view in brief terms: "Context compounds. Models commoditise. The platform does not."
The founders also argue that many software tools do not adapt closely enough to how organisations actually work. "Historically, operators have been sold ready meals. SaaS tools that do not match their diet, do not adapt to their needs, and leave them picking out the ingredients they never wanted. Zaro inverts that. Your organisation brings the ingredients: every decision, file, and piece of operational history accumulated. From those ingredients, you have a personal chef with the intelligence to know your business needs at any given moment and the autonomy to build it. As you scale from fifty people to fifty thousand, what that chef produces changes with you. Because it was built by you in the first place," said Bajwa.
Early use
The platform includes a built-in app store with pre-built projects and workflows across functions such as HR, finance, and facilities. Zaro already uses the system for its own internal operations.
One early beta user is Seb Johnson, Founder, Scaling Europe, who used the platform to build a press release tracker and a news monitor linked to X and selected technology publications. The tools were meant to replace a manual process of checking social media and websites throughout the day.
"I haven't missed a single press release this week, which is, like, unheard of," said Johnson.
Cherry Ventures backed the company before the product was built. Dinika Mahtani, Partner, Cherry Ventures, framed the investment around the broader challenge of making AI systems useful over time inside large organisations.
"The Chat GPT moment for enterprise context hasn't arrived yet because no one has closed the full loop. Zaro does. It's the first platform where the AI genuinely gets smarter the longer it's in an organisation, and that compounding effect is what makes it defensible," said Mahtani.
Zaro has eight employees, five of whom previously built AI agents at Convergence before joining Salesforce. The company says that background gave the team a close view of how businesses deploy agent software and where those systems fall short when data, context, and outputs remain scattered across separate tools.