Sunny Ray sits down with Matt Bloomberg, a four time tech CEO now running Markup AI, a multi agent platform that helps enterprises safely scale AI generated content. Bloomberg traces the throughline of trust across his career, from twenty years at Return Path building trust infrastructure for email, through Bolster, which brought transparency to executive recruiting, to Markup AI's Content Guardian agents that check brand, legal, SEO and accuracy standards before content gets published. He explains how the company is pivoting a legacy natural language processing business into a native AI platform, why he abandoned an early developer first approach in favor of serving content teams, and how removing engineering as the bottleneck has upended traditional product roadmaps. Bloomberg also shares hard won lessons on scaling mistakes, like hiring senior executives before a startup is ready for their infrastructure, drawn from his time at MovieFone and beyond. Throughout, he argues that as AI generated content becomes infinite, trust becomes the scarcest and most valuable asset a company can protect.
A four time tech CEO explains why trust, not content volume, will decide who wins the AI era.
What is Markup AI and what are you building?
We call ourselves content guardian agents, a multi agent, multi surface platform that helps companies safely scale AI content production. Companies have poured money into AI coding tools, but content infrastructure hasn't caught up, so writers become editors fixing mediocre AI output. We help companies define standards for brand, legal, search visibility and accuracy, then score and check content against those standards with human in the loop review so teams feel comfortable hitting publish.
You spent twenty years at Return Path deciding whether email was trustworthy. Is Markup AI the same company at a different layer?
Couldn't be more different, though trust is a real throughline. Return Path built trust in email through data. Bolster, a company in between, added trust to executive recruiting through transparency. Markup AI is about trusting AI and trusting content. There's also a through line around culture, building high performing workplaces through trust between management and employees. So trust and transparency run through my career, even though the businesses themselves look nothing alike.
Do you deliberately gravitate toward invisible infrastructure businesses?
I don't, actually. Of my four companies, two were invisible and two were very visible. B2B infrastructure companies can be great businesses that touch consumers without them realizing it, and there's nothing wrong with that. But I also worked early in my career at MovieFone, a hugely visible consumer brand millions used daily to find showtimes and buy tickets. There's no magic to whether a business is behind the scenes or out front.
What's the biggest scaling mistake you now design against from day one?
It's easy to think you're ready to scale before you are, and to hire the wrong person too early. Founders often think hitting a million in ARR means they need a CRO, when really they need a good sales rep or a sales manager. I saw a company hire a super senior CMO who literally asked where the typing pool was on day one. It's easy to get three steps ahead instead of staying one step ahead.
What changed in what you wanted out of a company between selling Return Path and starting Markup AI?
By late 2024 I was excited about AI and had been using it heavily myself, and I knew it was the future, so I wanted to build something in AI. When I found this company, it was already AI adjacent with real revenue and marquee customers, just needing a more contemporary solution. That let us have all the fun of building an AI startup while having an established customer base giving us feedback as they migrated to the new product.
You draw a hard line between AI native and AI bolted on. What's the concrete test to tell them apart?
A lot of SaaS companies say they're an AI company now, but really they've just added a chatbot or a small agent onto an existing product, which isn't wrong, just misclassified for hype or investor appeal. When you build something native to AI from scratch, it functions differently and you care about different things than when you retrofit AI into an existing workflow. That's probably true of any technology disruption, not just this one.
Can you walk me through what a Content Guardian agent actually does inside a large enterprise?
It rapidly evaluates large bodies of content against a fixed pool of rules and standards, then fixes low stakes issues or flags high stakes ones for a human. That covers brand voice and terminology dictionaries with hundreds of thousands of terms, legal and regulatory rules like drug specification compliance for pharma marketing, AI search visibility for SEO, AEO and GEO, and accuracy against a defined source of truth. It requires multi agent orchestration to work at enterprise scale.
Why go API first and MCP powered rather than build a destination app?
Our philosophy is that we need to live where content already lives, in Google Docs, Word, Figma, PowerPoint, GitHub, rather than forcing a whole new workflow on people. So we take a multi surface approach: API, MCP, a Chrome extension, and bespoke integrations for quirky applications. We still need an administrative dashboard for account owners to set rules and link repositories, but for end users it should be as simple as connecting from wherever they already work.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview