This week, we’re looking at a company going after a problem most teams feel but rarely name: the growing coordination tax of working across humans, tools, and now, AI agents.
Highlight AI just raised a $40 million Series A led by Khosla Ventures to build what it calls a shared intelligence layer for modern work.
The pitch is straightforward. AI has increased output, but not alignment. Teams are buried in messages, tickets, and drafts spread across tools, with employees switching between as many as eight apps a day and spending significant time just keeping context intact.
Highlight sits across those systems as a memory and coordination layer. It captures decisions, tracks changes, assigns ownership, and prepares next steps automatically. In practice, that means a design review in Figma flows into Slack, tasks are created in Linear with context attached, and updates are drafted without someone manually stitching everything together. The goal is not more agents, but less coordination work around them.
Sergei Sorokin, formerly VP of Product at Discord, steps in as CEO. He helped scale Discord from 5 million to nearly 300 million users and built its AI efforts, then spent time advising AI-native startups. That exposure shaped the core insight behind Highlight, that the bottleneck is no longer intelligence, it is coordination across systems.
“The biggest opportunity is in solving coordination as the number of AI agents continues to proliferate across teams and tools,” Sorokin said. “We’re already seeing work spread across a growing set of systems, where people are constantly switching between apps and manually stitching context together just to stay aligned, creating a greater need for our intelligence layer.”
The company plans to double headcount, with a focus on engineering, as it builds out this infrastructure layer.
The bet here is subtle but important. While much of the market is focused on building smarter agents, Highlight is betting that the real unlock comes from connecting them, and the humans around them, into something that actually works as a system.
–
This article was drafted with the help of generative AI using company-submitted details, then manually edited and carefully reviewed by a human editor before publication.