ML training workstations
PR-7002 · Tom Okafor · Oct 15, 2026
What it does
This page is the full record for one purchase request: what was asked for, how policy rules applied, and each approval step. When the last required approver signs off, ProcureAI issues a purchase order (PO) to the supplier with the same lines and payment terms.
What you can do here
- Review line items, the original wording, and policy check results
- Approve or reject if you are the person waiting on the timeline (demo: act as that approver)
- See the issued purchase order and success banner once the request is ordered
Behind the scenes
Each approval or rejection is stamped on the timeline with an optional comment. Rejection stops the request. When every step is approved, the agent creates a new PO in the system, links it to this request, and logs the activity for the command center.
Example: After Tom Okafor (manager) and Lena Fischer (IT category manager) approve Dev Kapoor's hardware request, ProcureAI issues a PO to Apex Devices—for example PO-1049—with Net 60 terms on the total shown here.
Line items
| Description | Qty | Unit price | Line total |
|---|---|---|---|
| Quanta GPU Workstation (2x RTX) | 3 each | $11,900 | $35,700 |
Original request
We need 3 GPU workstations for the recsys team, ideally within a month
Justification: Recommendation-model training is queueing 2 days for shared GPUs
Policy checks
- Contracted catalogItem is not on a negotiated contract; consider sourcing event
- Approval thresholdRouted to Finance for spend above $25,000
Approval timeline
- TOTom OkaforManagerapproved1d ago
- LFLena FischerIT category managerpending
- RIRaj IyerFinance (> $25k)pending