Forward Deployed Engineering
Three phases. One year. AI in every role.
Discover → Build → Adopt & improve. Our engineers run the numbers first, build second, and stay until it sticks.
One-year timeline
What happens each month
Three phases
Each phase: what we do, what you receive
Discover
Month 0–1 (diagnostic 2–3 weeks)
You get: One clear picture: what goes to AI, what it saves, when it pays back, and where the risks are.
What we do
- On-site interviews with owners and department heads; map the org chart and headcount
- Break each role into tasks and time share; score every role for AI impact
- Model each work pool: hours AI can absorb, process changes, capacity and shift coverage
- Cost before vs. after, AI spend, payback and scenarios
What you receive
- AI Transformation Target-Org & Labor-Cost Report
- Role-by-role diagnostic
- AI employee list by department, prioritized
- Pilot scope and acceptance metrics
Build
Months 1–4 (incl. 90-day pilot)
You get: Within 90 days, one department is live — with real data on what it saved.
What we do
- Design AI employees per department: capability, trigger, input, output, human checkpoints, fallback, KPIs
- Run a 90-day pilot in 1–2 departments, in parallel with the old process
- Integrate into the super AI workbench and your existing DingTalk, WeCom, Feishu, Yonyou, Kingdee systems
- Set human gates: payments, approvals and safety calls are never automated
What you receive
- An AI employee spec sheet for every agent
- The super AI workbench (private or dedicated cloud)
- Integration list and permission audit
- Weekly pilot reports on quality, hours and compliance
Adopt & improve
Months 4–12, then ongoing
You get: A year on, AI is part of everyday work — and savings are counted only when spending actually falls.
What we do
- Role-by-role training and AI coaching until people actually use it
- Expand pool by pool: headcount changes only after 8+ weeks of measured results
- Appoint an AI operations & knowledge owner
- Monthly review; roll back if any guardrail is missed
What you receive
- Training kits, question banks and assessment records
- Pool-by-pool acceptance reports
- Monthly AI transformation review
- Iteration and rollback playbook
Deliverables
What you'll walk away with
Every phase ends with something you can hold, read and take into a board meeting.
Diagnostic & cost report
Target org, labor cost before/after, payback, scenarios and sensitivity
AI employee catalogue
Capability, triggers, inputs/outputs, human checkpoints, fallback and KPIs for every agent
Super AI workbench
Owner/sales/data/finance/HR views, connected to your systems
Role-based training kits
Courseware, question banks, AI coaching, assessment records
Iteration & rollback playbook
Monthly reviews, acceptance and rollback conditions
Before any headcount change, all of these must hold
- 8+ weeks of measured results, including a peak period
- No drop in service, accuracy or safety metrics
- Shift coverage and capacity confirmed
- Process changes signed off by department heads
- Labor-law process followed; AI is never a reason to terminate
- Savings booked only when spending actually falls
Roll back immediately if
- AI accuracy stays below the agreed target
- Complaints or service metrics clearly worsen
- Any safety, compliance or data incident
- Adoption stays low and the cause can’t be fixed
- Capacity can’t cover a peak period
Team
Who comes on site
Transformation advisor
Sets goals, runs the numbers and prioritizes with the owner
FDE engineer
Builds AI employees on site, integrates systems, runs them
Data & security engineer
Data governance, permissions, private deployment, audit
Training & ops specialist
Role-based training, AI coaching, monthly reviews
An AI roadmap you can act on — in 14 days
Free diagnostic: on-site interviews, role-by-role AI scoring and a cost model. Run the numbers first, then decide.