SERVICE / FORWARD DEPLOYED ENGINEERING
FDE onsite customisation: testing AI recruiting capability where the work happens
Off-the-shelf AI recruiting tools rarely fit a company's own role rules and collaboration model. Forward Deployed Engineering means engineers go to the site, start from the part worth solving, build a usable workflow, then validate it against agreed metrics.
- Typical use: key-role search, talent review, CV and material handling
- Delivery: onsite co-creation, system integration, pilot validation
- Offices: Shanghai, Shenzhen and across China
When FDE onsite customisation fits
- You already use AI tools, but they cannot follow your own Job Portrait and evaluation rules
- Recruiting spans several systems, data is scattered, and it is hard to form a complete view of a candidate
- You want to validate AI on a small number of key roles before committing further
- You have an internal IT team but lack engineers who understand recruiting and can write code
Three stages of onsite work
- 01
Find the part worth solving first
Work with the business and recruitment teams to map workflows, data conditions and existing systems, and define the problem, available data and validation approach.
- 02
Design around how the work actually happens
Design AI-assisted workflows and the integration they need around the company's role rules, collaboration model and systems, including interface, permissions and data boundaries.
- 03
Validate and adjust in use
Validate with the people on site, collect issues, calibrate outputs, test against agreed metrics, then decide on expansion.
Inputs, outputs and validation
| Stage | What the client provides | What we deliver | How it is validated |
|---|---|---|---|
| Pilot scope | Business goals, priority roles, current process | Pilot plan and validation metrics | Agreed pilot boundary |
| Data and permissions | Data sources, permission scope, compliance requirements | Data handling and permission design | Permission list and de-identification check |
| Workflow design | Role rules, evaluation standards, collaboration model | AI-assisted workflow and interface | Onsite user trial |
| System integration | Existing recruiting systems and interfaces | Necessary system integration | End-to-end run on real role data |
| Pilot evaluation | User feedback and business metrics | Pilot report and scaling recommendation | Review against agreed metrics |
Data access and boundaries
- Data scope, storage location and access permissions are agreed in the pilot plan
- AI output is supporting information; it does not make hiring or rejection decisions automatically
- Data handling at the end of the pilot is agreed between the client and YUNLUE
How it relates to consultative search
FDE onsite customisation takes the Job Portrait, Talent Portrait and interview methods we have refined in consultative search and turns them into AI-assisted workflows the company can use. It works alongside AI-assisted matching such as the candidate relationship graph, so the method can be tested where recruitment actually happens.
Frequently asked questions
How long does an FDE pilot take?
It depends on the problem and system conditions. After the plan is agreed, we typically run one verifiable recruiting step at a time, review against agreed metrics, then decide whether to expand.
Does FDE replace our existing recruiting system?
No. FDE integrates with existing systems first and adds workflow and interface only where needed. Whether to build a new system follows from pilot results and the company's plans.