Article
Forward deployed engineering in practice: real cases
Forward deployed engineering through EcoToGo, BaseCamp Studio and Gazzanet via Fincons: real work on integrations, AI and enterprise constraints.
Published 17 November 20266 min readFDEReal casesIntegrations
Forward deployed engineering is easier to understand through work than through a title. Here I follow three experiences from my background: EcoToGo, BaseCamp Studio and Gazzanet through Fincons Group. They were not engagements presented as FDE roles, and I am not renaming them now. They help explain building within a real context and what I take from it for product work. The lessons are my interpretation, not announcements of undocumented outcomes or roadmap changes.
EcoToGo: the context between shop and business software
EcoToGo is a B2B and B2C e-commerce business selling ecological disposable packaging for catering and events. I follow technical work as a part-time CTO, from conversations with the owner to everyday corrections. The shop uses TeamSystem Commerce, formerly Storeden, and Danea handles business management. The need extends beyond the storefront: information and orders must reach logistics. A connection that appears to work may still fail to carry everything the next step needs.
EcoToGo: what I built and connected
I redesigned and developed the product page, restructured the cart and worked on integrations between the shop and Danea. For logistics, daily orders are exported and processed by an internal tool that sends the information needed to create orders in GSPED through its API. Scripts also send documents. Logistics creates the shipments, not our script. That boundary matters when describing the automation without claiming responsibility for work carried out by another system or operator.
EcoToGo: what comes back to product decisions
With the supplier's support, I added API fields for internal needs, including priority shipping and priority preparation. I updated the shop export and the Danea-Storeden synchronisation script to enrich the information destined for logistics. This is a concrete example of an operational need becoming a change in an integration. I also follow the evaluation of a move to Odoo, considering integrations and scripts. I do not present the migration as completed or publish an unmeasured saving.
BaseCamp Studio: the DAM context
For BaseCamp Studio I work as a developer in the DAM, the software organising photos, video and other digital content. I also work on integrations and production modules. I am not the platform's founder. Unlike a shop integration, this context involves making information across different sources searchable and introducing capabilities into an existing product. An interesting model response is not enough. The response needs an understandable place within the system.
BaseCamp Studio: search, tags and enriched data
I developed a package indexing video, images and text for search by meaning. I worked on automatic tags and AI asset enrichment, extracting and categorising information from images, text, PDFs and videos with audio. The audio here belongs to video, not a claimed standalone audio-file capability. A vector database makes approved semantic information searchable. I use existing AI tools rather than claiming to have built a model from scratch.
BaseCamp Studio: review as a product decision
I also managed and developed the platform's PIM, starting from DAM assets. From the customer's master data I export structure and mapping, enrich the data and reimport it mapped onto photos, video and other assets. Content remains connected to product records: enrichment should not create information detached from the product it belongs to.
Extracted information is proposed enrichment. A confidence percentage supports review but does not establish truth: human checking comes before information becomes asset data. The product lesson I take from this is to distinguish generation, approval and search. Connecting a model is not enough if the data's meaning in the workflow remains ambiguous. I explain more in the articles on semantic search in the DAM and AI asset enrichment.
Gazzanet through Fincons: the enterprise context
I worked on Gazzetta.it's Gazzanet network in 2024-2025 through Fincons Group, which had Gazzanet as a client. That relationship matters: I do not present myself as a direct Gazzetta.it employee. The environment was enterprise and high traffic. In my background it illustrates another side of product engineering: a change working on its own is not enough. Existing behaviour and the risk of breaking something other people rely on must also be considered.
Gazzanet: the engineering work and its boundaries
The engineering environment included CI/CD through Jenkins, Jira, Twig and React and Node integrations. Those names describe the technical context, not a list of systems I claim to have created. I do not attribute a traffic outcome or an undocumented standalone feature to my contribution. For the embedded aspect, the constraint matters: working inside an existing product rather than treating the task as an isolated new project starting from nothing.
Gazzanet: the lesson I bring to product work
My reading is that existing behaviour belongs to the problem to understand, not an afterthought. Useful feedback can mean explaining which dependencies need clarification before a decision. I am not saying a particular request changed Gazzanet's roadmap: that is not an outcome I publish. I am describing how enterprise experience influences the way I assess a technical proposal and the questions I would bring to a product team.
How I would bring these experiences to an FDE team
I would start by asking which task needs to become possible for the customer and which constraints the product must preserve. EcoToGo provides a concrete example of information crossing systems; BaseCamp of generated content needing review; Gazzanet of existing behaviour that matters. These are not solutions to copy. They are contexts for explaining why I would ask to see the workflow, clarify responsibility and separate what we know from what still needs checking before deciding what to build.
The common thread is not a framework. It is the relationship between what gets built and the context that makes it useful. If your team needs this approach, my Forward Deployed Engineer profile explains my location, working approach and contacts. For a broader definition, read what an FDE does.