Solera Energy AI Procurement Agent
A Brisbane solar EPC contractor let an AI agent draft, compare, and route AUD 4M of monthly supplier RFQs.
Impact
What changed.
Cycle compression
Average RFQ-to-PO cycle dropped from 8.4 working days to 2.1. Project engineers stopped chasing procurement; procurement stopped chasing suppliers.
Cost capture
The firm identified AUD 1.9M in annualised savings in the first six months — mostly from better-priced alternatives surfaced by the comparison tool that the team had been too time-poor to find.
Audit posture
Every PO now ties back to a documented comparison with reasoning. The CFO described it as the first time procurement could survive a finance audit without scrambling.
The challenge
Before
Solera Energy is a Brisbane-headquartered solar EPC contractor putting roughly AUD 4M a month through procurement across modules, inverters, mounting, cable, and balance-of-system. The procurement team of four was drowning — RFQs arriving as PDFs, quotes returning in inconsistent formats, comparisons hand-built in Excel, and recommendations written in long email threads that the project managers admitted they often skimmed. The team was missing better-priced alternatives simply because the analysis was too tedious to do thoroughly.
- Inbound RFQs arriving as inconsistent PDFs from project engineers
- Supplier quotes returning in mismatched formats and line-item naming
- Bid comparisons hand-built in Excel taking 4-6 hours per package
- Procurement recommendations sent as long email threads, often skimmed
- No memory of past supplier performance influencing new decisions
- Better-priced alternatives missed because thorough analysis was slow
- Compliance documents (certificates, warranties) re-collected per quote
- No audit trail tying final PO back to the comparison that justified it
The solution
What we built
We built a procurement agent that handles the work the team had been doing manually. When a project engineer raises an RFQ, the agent parses the requirement, identifies eligible suppliers from a vetted panel, drafts the outgoing RFQ in the firm's template, and sends it. When quotes return — in whatever format — the agent extracts line items, normalizes units of measure, applies known supplier discount structures, and produces a side-by-side comparison with total landed cost including freight, duties, and lead-time penalties. The agent flags compliance gaps (missing certificates, expired warranties) and pulls in supplier performance history (on-time delivery, reject rates, payment terms compliance) from the firm's history. A recommendation is generated with the agent's reasoning, but a human procurement lead approves before any PO is issued. Approval is one click; every decision and its justification is preserved as an audit trail. The agent runs on a private deployment with the firm's data — supplier price files never leave the environment.
Core workflow connections
How the system flows.
- RFQ IntakeRequirement ParsingSupplier Panel MatchRFQ Drafted
- Quote Received (PDF/email/XLS)Line ExtractionUnit Normalisation
- Comparison BuildLanded CostCompliance CheckPerformance History
- Recommendation DraftedReasoning CapturedProcurement Lead Review
- ApprovalPO GeneratedAudit Trail LinkedProject Cost Updated
- Supplier Performance FeedbackDelivery + Reject CaptureHistory Update
- Compliance Document RefreshCertificate Expiry WatchSupplier Nudge
- Private deployment with supplier price files inside the firm environment
- One-click human approval preserving every reasoning step
- Vetted supplier panel maintained by procurement, not by the agent
Process
How we built it.
RFQ Intake → Requirement Parsing → Supplier Panel Match → RFQ Drafted
Quote Received (PDF/email/XLS) → Line Extraction → Unit Normalisation
Comparison Build → Landed Cost → Compliance Check → Performance History
Recommendation Drafted → Reasoning Captured → Procurement Lead Review
Approval → PO Generated → Audit Trail Linked → Project Cost Updated
Supplier Performance Feedback → Delivery + Reject Capture → History Update
Compliance Document Refresh → Certificate Expiry Watch → Supplier Nudge
Private deployment with supplier price files inside the firm environment
One-click human approval preserving every reasoning step
Vetted supplier panel maintained by procurement, not by the agent
Start a project
Drowning in supplier quotes and PDF comparisons?
We build procurement agents that respect human approval — they do the grinding analysis, you keep the decision and the audit trail.
No retainer lock-in · Month-to-month · Full transparency
More work
Up next.
A Hong Kong boutique hotel deployed a multilingual AI concierge that answers 79% of guest queries before a human is paged.
Harborlight Hotel AI Concierge Agent
A Singapore-headquartered regional insurer cut first-touch claims triage from 6 hours to under 8 minutes.