London-based Bosun Labs has emerged from stealth with a platform designed to stop companies flying blind on AI security.
Founded by second-time founder Aaron Randall, previously a co-founder of climate startup Supercritical and Songkick, Bosun gives enterprises visibility over the AI tools being used across their organisations, including unsanctioned “shadow AI” products often used privately by employees, and flags sensitive data exposure, risky configurations and unmanaged agent connections.
Bosun has raised an undisclosed angel round backed by operators from Spotify, Meta and Stripe, said Randall. The company is initially focused on the UK and Europe, though Randall says the platform will be able to support customers internationally.
Randall told Pathfounders the problem became obvious after six months of discovery calls with senior leaders at more than 40 companies, including executives from Spotify, Datadog, GitLab, Shell and The Guardian.
“AI has been dropped from the very top,” Randall said. “The CEO has said everyone adopt AI at breakneck speed. There are no budget constraints, which leads to this huge problem with tool sprawl.”
That rush, he argues, has left security teams blind. Employees are using personal AI accounts, plugging tools into live systems and feeding company data into products their employers cannot monitor.
“One exec said to me, I was sitting in my office, and I overheard an employee talking about how they’d uploaded a CSV of customer credit card data to their personal ChatGPT account,” Randall said. “In doing so, they created a data breach that needs to be reported to the Information Commissioner’s Office.”
The company cites Microsoft research suggesting 71% of employees use unapproved AI tools at work, while only 12% of companies have a dedicated AI governance framework. Bosun says that gap is already creating hidden compliance risk, and could worsen as AI agents start calling APIs, moving data and acting across live systems.
“Leaders don’t know what tools are being used, who is using them, and what data is going into them,” Randall said. “That’s where the idea of Bosun came from.”
Bosun’s platform works in two layers. First, it connects to approved AI providers such as ChatGPT, Claude and Gemini through read-only APIs, pulling usage, spend and adoption data into a central dashboard. Second, an endpoint agent deployed through mobile device management software detects shadow AI usage in browsers, apps and embedded tools.
Randall says the product can surface risky behaviours such as engineers enabling permission-bypassing modes, employees connecting MCP servers to live systems, or users exposing personally identifiable information to unsanctioned AI tools.
“You have a single hub to basically see those issues and respond to them,” he said.
Other similar firms are entering the market, such as Geordi AI and WitnessAI, as well as incumbents including Microsoft Purview and CrowdStrike.
However, Bosun is not pitching itself as a blunt hammer to crack a nut. Randall says some customers are using it to understand where useful grassroots AI adoption is already happening, then bring those tools formally inside the company with the right terms, controls and oversight.
“We actually don’t want to just lock down and say you can’t use ChatGPT or Gemini,” he said. “A lot of this is about how we create a safeguard around teams so they can experiment with these AI tools in a safe way.”
“Today the risk is human-focused,” he said.
Randall was previously co-founder and CTO of carbon removal startup Supercritical, whose customers included Microsoft, McKinsey and BlackRock. Before that, he was CTO at Songkick, the live music platform acquired by Warner Music.
