[{"data":1,"prerenderedAt":40},["ShallowReactive",2],{"article-en-what-is-fde-forward-deployed-engineer":3},{"article":4,"related":17,"prevArticle":35,"nextArticle":36,"availableLocales":37},{"id":5,"slug":5,"title":6,"excerpt":7,"content":8,"image":9,"date":10,"published":11,"status":12,"category":13,"keywords":14,"meta_description":15,"og_image":16,"custom_css":16,"created_at":10,"updated_at":10},"what-is-fde-forward-deployed-engineer","What is an FDE? On-site engineers deliver systems, FDEs deliver results","How is an FDE (Forward Deployed Engineer) different from an on-site engineer? One delivers a system; the other delivers a result. Drawing on three weeks one of our engineers spent working alongside a trading company's cashier and sales team, we look at how an FDE works and when a business needs one.","\u003Cp>FDE stands for Forward Deployed Engineer: an engineer deployed to the front line. The role appeared at Palantir in the early 2010s, but it only became widely visible once AI companies started hiring FDEs.\u003C/p>\n\n\u003Cp>We won't get into how the FDE role has changed since AI arrived. The role came into view alongside AI in the first place, so that question doesn't get us very far. What we want to look at is what an FDE actually does, how that differs from the on-site engineer most people are familiar with, and when a business needs one.\u003C/p>\n\n\u003Ch2>How Palantir uses the term\u003C/h2>\n\n\u003Cp>At Palantir, FDE refers to engineers who don't sit back at the office waiting for requirements, but go and work directly at the customer's site. Palantir has since gone as far as naming the AI agent in its Foundry platform \u003Ca href=\"https://www.palantir.com/docs/foundry/ai-fde/overview\" target=\"_blank\" rel=\"noopener\">AI FDE\u003C/a>, which its documentation describes as:\u003C/p>\n\n\u003Cblockquote>\u003Cp>\"an interactive agent that operates Foundry for you through conversational commands\"\u003C/p>\u003C/blockquote>\n\n\u003Cp>According to the documentation, it works through data transformations and builds and maintains the ontology based on the conversation, then checks the results and corrects itself. To us, the fact that even an AI agent borrows the FDE name says a lot about what the role is at its core: understand the business first, then make the system work the way the business does.\u003C/p>\n\n\u003Ch2>How an FDE differs from an on-site engineer\u003C/h2>\n\n\u003Cp>In a sentence: \u003Cstrong>an on-site engineer delivers a system (or a tool); an FDE delivers a result.\u003C/strong>\u003C/p>\n\n\u003Cp>The real dividing line is \u003Cstrong>who defines the problem.\u003C/strong>\u003C/p>\n\n\u003Cp>Say a company tells you: \"It takes us ten days every month to reconcile accounts receivable across our banks. We'd like your help automating it.\"\u003C/p>\n\n\u003Ch3>What an on-site engineer typically does\u003C/h3>\n\n\u003Cp>The client hands over the reconciliation rules, the fields, the system APIs and the requirements documents, and the engineer takes care of integration, development, testing and deployment. Success means the system meets the spec and works as intended.\u003C/p>\n\n\u003Ch3>What an FDE typically does\u003C/h3>\n\n\u003Cp>An FDE gets deep into how the finance staff actually do the work and digs into why it takes ten days. For example:\u003C/p>\n\n\u003Cul>\n  \u003Cli>Which data arrives late, is missing, or can't be matched up?\u003C/li>\n  \u003Cli>Which reconciliation rules have never actually been written down?\u003C/li>\n  \u003Cli>Which exceptions need a human to make the call?\u003C/li>\n  \u003Cli>Which human decisions could be handed off to rules, AI or workflow automation?\u003C/li>\n  \u003Cli>Could the solution carry over to other banks, other ERPs, or even other clients?\u003C/li>\n\u003C/ul>\n\n\u003Cp>What gets delivered in the end isn't just a reconciliation program. It's a reconciliation capability that can keep growing, covering exception handling, rules, data mapping and a process for human review.\u003C/p>\n\n\u003Cp>An FDE still writes code, integrates APIs and solves deployment problems, but those are the means to a business result, not the whole job.\u003C/p>\n\n\u003Ch2>Interviews aren't enough: you have to be where the problem is\u003C/h2>\n\n\u003Cp>The way we work, people have to be directly involved where the problem actually happens. Interviews alone come nowhere near enough.\u003C/p>\n\n\u003Cp>Here's an example. At a trading company, one of our engineers worked side by side with the cashier and sales team for three full weeks. Throughout those three weeks, they did the work while continually revising their understanding of the business and where it got stuck. In the end, it took them less than a week to put forward a plan that could improve overall efficiency by nearly 35%.\u003C/p>\n\n\u003Cp>That plan didn't come out of a requirements document. It came out of those three weeks of real work.\u003C/p>\n\n\u003Ch2>Where ProCheck fits in\u003C/h2>\n\n\u003Cp>The FDE goes on site and gets the problem clearly defined. \u003Ca href=\"/en/procheck\">ProCheck\u003C/a> handles managing and training the AI agents, which takes a great deal of work off people's plates. Things like cost management and model management are already built in.\u003C/p>\n\n\u003Cp>Because so much is built in, ProCheck has hundreds of parameters, and realistically only trained professionals can use it to train an AI agent. Once the agent is trained, though, using it day to day doesn't take that kind of expertise: a business can simply treat the AI agent as another employee.\u003C/p>\n\n\u003Cp>In other words, the expertise is concentrated in the rollout and training stage, and the FDE carries it. Everyday use stays at a level the business can handle on its own.\u003C/p>\n\n\u003Ch2>When does a business need an FDE?\u003C/h2>\n\n\u003Cp>From our perspective, the test is straightforward: \u003Cstrong>if certain delays, whether you sense them or see them in the numbers, are seriously dragging down operating efficiency, and fixing them would save significant cost or help win revenue, it's worth considering.\u003C/strong>\u003C/p>\n\n\u003Cp>One very common sign is a manager thinking, \"This looks simple. Why is it taking so long?\"\u003C/p>\n\n\u003Cp>When that thought comes up, it may be time to bring in one of our consultants, or to \u003Ca href=\"/en/consultation\">have a conversation with us first\u003C/a>. Often an hour or two is enough to tell whether you need an FDE.\u003C/p>\n\n\u003Ch2>Closing thoughts\u003C/h2>\n\n\u003Cp>Engineers' logical rigor has always been a valuable strength. What an FDE does is put that rigor to work not just on getting the spec right, but directly on helping the business grow.\u003C/p>\n\n\u003Cp>To learn how we embed with clients and roll out solutions, see \u003Ca href=\"/en/services\">our services\u003C/a>.\u003C/p>","https://cdn.sqa.tw/articles/what-is-fde-cover-2026-11-21.jpg","2026-11-21T12:00:00+08:00",true,"published","perspective","what is an FDE, Forward Deployed Engineer, FDE, on-site engineer, AI adoption, AI agent, ProCheck","What is an FDE (Forward Deployed Engineer)? On-site engineers deliver systems, FDEs deliver results. How FDEs work, ProCheck's role, and when you need one.",null,[18,23,29],{"id":19,"slug":19,"title":20,"excerpt":21,"image":22,"date":10,"category":13},"ai-sdlc-transformation-with-teams","What we learned working alongside teams through an AI SDLC transformation","Since late last year, we have worked alongside 8 teams through a full AI SDLC transformation, and 6 of them are now fully up and running. Looking back, the framework itself wasn't what took the time. Taking each project live, one at a time, and adjusting it again and again was.","https://cdn.sqa.tw/articles/ai-sdlc-teams-cover-2026-11-21.jpg",{"id":24,"slug":24,"title":25,"excerpt":26,"image":27,"date":28,"category":13},"ai-development-sdlc-release-verification","AI builds it in two hours, but verification still takes three days","AI can speed up development, but spec sign-off, testing and release don't speed up with it. Drawing on a booking system SaaS we worked with, we explain how we reworked the SDLC to bring full release verification down from two or three days to under two hours, and how the SQA role shifted as a result.","https://cdn.sqa.tw/articles/ai-sdlc-verification-bottleneck-cover.jpg","2026-11-05T12:00:00+08:00",{"id":30,"slug":30,"title":31,"excerpt":32,"image":33,"date":34,"category":13},"sme-guide-to-choosing-ai-agent","With AI agents everywhere, how should small businesses choose?","We were recently brought in to do a full review of an AI agent after deployment, and found it had quietly caused five times too much raw material to be ordered from customer order emails, and nearly created a liquidity risk in finance. The trouble with agents is rarely that they aren't smart enough. It's that nobody asked the right questions.","https://cdn.sqa.tw/articles/sme-agent-black-box-inspection.jpg","2026-09-01T15:00:00+00:00",{"slug":24,"title":25,"date":28},{"slug":19,"title":20,"date":10},[38,39],"en","zh",1795282790111]