# Can someone build our AI agent use case with us? Forward-deployed engineering > Yes. Sanlabs' forward-deployed engineers build AI agent use cases with your team on Sanbox, with experience in aviation, pharma, supply chain and logistics, and internal admin. - Source: Sanbox knowledge base, by Sanlabs GmbH - URL: https://sanbox.cloud/kb/teams/forward-deployed-engineering/ - Type: explainer - Section: Agents for teams - Language: en - Published: 2026-11-06 - Updated: 2026-11-06 - German version: https://sanbox.cloud/de/kb/teams/forward-deployed-engineering.md Yes. Sanlabs, the company behind Sanbox, offers forward-deployed engineering: our engineers work with your team to turn a business process into an AI agent that runs on Sanbox Computers, connected to your systems and inside your controls. We have experience in aviation, pharmaceuticals, supply chain and logistics, and internal admin workflows. [Talk to us](mailto:sales@sanlabs.ai?subject=Forward-deployed%20engineering) about the process you have in mind. ## What does a forward-deployed engineer do? A forward-deployed engineer works inside your project rather than behind a support queue. They: - **Learn the process** from the people who run it today: the inputs, the systems, the exceptions and who signs off. - **Build the agent and its tools**, on a Sanbox template made for that role, with the files, packages and browser it needs. - **Connect your systems** through Secret Proxies, so the agent uses your APIs without holding the keys, and through a network policy that allows only the systems the work needs. - **Test with real cases** from your team, and adjust until the results hold up. - **Hand it over or run it with you**, with your people able to see, stop and change what the agent does. ## How does an engagement work? 1. **Pick one process.** A good first candidate repeats often, has a clear owner and has an outcome you can measure, such as hours saved or cases handled. 2. **Scope it.** We map the steps, the systems involved, where a person must review the agent's work, and the data protection questions. 3. **Build it on Sanbox.** One template per role, deny-by-default egress, credentials in Secret Proxies, and snapshots so every new Computer starts from a tested state. 4. **Pilot it.** Your team uses it on real work while we measure the results and fix what doesn't hold up. 5. **Roll it out.** More users or more processes, from the same templates. ## Which processes suit an agent? Processes where people move information between documents, portals and systems, follow rules with judgment, and spend time on cases that repeat. Examples from the areas we work in: - **Aviation:** finding answers in maintenance and technical documentation, preparing compliance and audit material, and operational reporting. - **Pharmaceuticals:** first-pass review of documents against checklists, pulling data out of reports and literature, and preparing drafts for qualified people to approve. - **Supply chain and logistics:** handling order and shipment exceptions, working in supplier and carrier portals, procurement requests, and matching documents such as orders, delivery notes and invoices. - **Internal admin:** back-office forms and portals, reports built from spreadsheets and documents, inbox triage and drafted replies, and onboarding checklists. In regulated work, the agent prepares and checks; the people responsible for the decision approve it. ## What stays under your control? - **Where it runs.** Computers, files and run records stay in EU data centres, under an Article 28 GDPR data processing agreement with Sanlabs GmbH. - **Which models it uses.** You bring your own model provider and key, including EU-hosted endpoints. See [where your agent's model calls go](https://sanbox.cloud/kb/eu/where-do-agent-model-calls-go/). - **What it can reach.** Each template's network policy lists every allowed destination, and nothing else is reachable. - **What it can see.** Credentials stay in Secret Proxies, outside the Computer. ## When is this not the right fit? - **A one-off script.** If the work runs once and has no owner afterwards, a short script is cheaper than an agent. - **Work that needs GPUs or local models.** Sanbox Computers have no GPUs; agents call your model provider. - **Desktop-only Windows software.** Sanbox Computers run Linux. Browser-based applications work through a Browser Computer. ## Related - [How do I give every employee their own AI agent computer?](https://sanbox.cloud/kb/teams/agent-computer-for-every-employee/) - [How do I give each of my users their own persistent agent computer?](https://sanbox.cloud/kb/build/one-persistent-computer-per-user/) - [Can an AI agent create Word, Excel and PDF files in the cloud?](https://sanbox.cloud/kb/teams/agents-that-write-word-excel-and-pdf/) - [How do I give an AI agent API keys without exposing them?](https://sanbox.cloud/kb/security/api-keys-your-agent-cannot-see/) - [Do I need a DPA for AI agents?](https://sanbox.cloud/kb/eu/dpa-for-ai-agents/) ## Next step Talk to us at sales@sanlabs.ai. Sanlabs' forward-deployed engineers build agent use cases with your team, with experience in aviation, pharmaceuticals, supply chain and logistics, and internal admin workflows. How we work: https://sanbox.cloud/kb/teams/forward-deployed-engineering.md Existing customers can use Sanbox from a coding agent: install the CLI with `npm install -g @sanlabs/sanbox-cli@latest`, sign in with `sanbox login`, then run `sanbox skill` for the full agent guide.