Love this breakdown on FDEs by Angela Strange. At Valon, our FDEs are deeply embedded with customers and are mission-critical to scaling ValonOS across regulated industries. We’re growing fast. 👉 https://bit.ly/49ro2hE
FDE has become both the most important and the most overloaded term in a very short period of time. My favorite definition from Kevin Bai Leo Mehr Priya Khandelwal: Your FDE org is how you close the gap between the potential of your platform and the needs of your customers. When framed this way, it is easier as a founder or prospective FDE to figure out the best fit. Does bridging this platform<>customer gap require: 1️⃣ Prompt engineering + Evals The gap = your agent needs to follow customer specific workflows and adhere to brand guidelines. e.g., your customer service agent needs to respond to each situation "appropriately" , your medical scribe needs to take long transcripts and output in standard SOAP format. Here an FDE needs to understand business requirements & translate these into prompts & eval sets (how should the agent behave in each situation, what policies it must follow). While these roles require an understanding of how agentic systems work, they can have lighter engineering skillset requirements. 2️⃣ Integrations & product extension (often to move up market) The gap = integrations into enterprise specific systems & feature extensions of an existing product. e.g., a billing product requires integration into a custom ERP and additional payment flows. An FDE needs to work with the customer to define the requirements *and* have the engineering chops to build into the existing product. 3️⃣ Entirely new products/ features The gap = customer has a product need (adjacent to your existing platform) and will pay for the value you deliver e.g., An insurance company asks for a claims document ingestion product while your core product is an underwriting workbench. Your FDEs are strong engineering and product deliverers. Often former founders. This category was the most debated. Do you: - Focus, and avoid customer requests that are too far afield of your core area. Or... - Not overanalyze the initial product. Assume most requests will get asked for multiple times and are eventually baked into the product. Solving one problem, builds trust to solve more problems. h/t James da Costa Ankur Rastogi,Tanay Padhi, Sashank Gondala, Michael Starin, Marcello Pedersen Sharan T. for contributing to the conversation! (&check out Leo's post in the comments) There is a LOT of nuance here. Many teams have personas in all 3 categories. What would you add? And if any of these roles seem exciting to you -- we have many exciting AI companies that are hiring!
