Why We Built ArchAgents

That is my realization after working with customers small and large across my time at Microsoft, Stripe and Meta. Product teams ship fast, but customers take months, sometimes years, to deploy what they buy. Making things easier for customers is of course the right answer, but the reality is that all software is hard to integrate into increasingly bespoke customer environments, each with its own requirements, business models and configurations.

The stakes could not be higher: delayed deployments mean real revenue loss, unhappy customers churn, and product teams burn out, spending more time debugging individual customers than building what’s next.

To solve this, our industry invented Forward Deployed Engineers (and TAMs, Professional Services, Implementation Engineers, and Solutions Architects). These are some of the most critical roles in B2B companies today. They take on the lion’s share of the hard problem of customer success. But these roles only scale to so many customers at a time, are often smaller in number than what’s needed, and are constantly squeezed between product velocity and customer debt.

We know all this firsthand. Our team has seen this story play out at Microsoft building Exchange Server and Office 365; at Stripe with Billing, Connect and Payments. We saw it again at Statsig (now OpenAI), at Atlassian and even at Meta, where Partner Engineers glued the platform to its consumers.

My team and I are building Forward Deployed Agents to help these heroes and their companies scale to thousands of happy customers. While AI tooling has progressed rapidly, nothing has focused on where the work actually happens: across companies. This is a very difficult problem. You have to build an extensible system that respects privacy and control while still providing rich context, memory and tools. It has to deliver deterministic outcomes while orchestrating LLMs from multiple providers. Most importantly, it has to meet our heroes where they live: across company boundaries.

With our cross-company runtime, customer-facing teams get the solutions they have always wanted. A customer-specific agent that a TAM can pair with on a shared Slack with their customer. An embedded agent that knows how to properly integrate your product, available right inside your customer’s coding harness (Codex, Claude, Cursor). And finally, Forward Deployed Agents that sit inside each company’s boundary and collaborate over shared threads: building test cases, fixing breaking changes, and working with their FDE peers to speed up deployments and revenue.

This is ArchAgents, designed hand in hand with our Fortune 500 design partners. This is how enterprise software should always have been built and delivered. We’re simply making it real.

Backed by early believers and investors: 20VC Fund, Kyber Knight, Will Gaybrick, Claire Hughes Johnson, Deb Liu, Vijaye Raji, Homebrew (Satya Patel, Hunter Walk), Sheila Gulati, Vlad Fedorov, Karandeep Anand and Rajeev Rajan. We were lucky to work with Soma Somasegar as well.