I use AI to build
better sourcing tools.
This working prototype turns a supplier comparison into a repeatable decision process. Adjust the brief, inspect the tradeoffs, and see the questions I would take to finance, quality, and operations.
AI helped create this tool. Its calculations use visible formulas and rules; each recommendation still needs commercial and operational review.
Choose chemical sourcing or industrial equipment, then test the cost, speed, and risk tradeoffs. All suppliers, prices, risk ratings, and outcomes are fictional and separate from Joshua Palacios’s career results.
Source the material. Protect the supply.
Compare delivered material cost, inventory commitments, and qualification against supply needs.
Set the brief
Adjust relative importance. The model normalizes the weights to 100%.
Eligibility comes first
A weighted score cannot override these requirements.
Only approved suppliers are eligible.
Supplier A
Supplier A leads among 2 eligible suppliers. Its largest weighted contribution is lower supply risk. The model supports moving to commercial and operational review.
This option costs $71,388 more than the lowest-cost eligible quote. The selected priorities favor its speed and lower illustrative risk.
See what drives the ranking
Weighted points / 100Qualification pending
Eligibility is checked before ranking. Excluded quotes remain visible for comparison.
How I would take this forward.
Fictional supplier scenario · AI-assisted build · Transparent, rule-based calculations
Assumptions600,000 kg / year · maximum 60 days · approved suppliers only · cost 35% / speed 25% / risk 40%
Supplier A leads among 2 eligible suppliers. Its largest weighted contribution is lower supply risk. The model supports moving to commercial and operational review.
This option costs $71,388 more than the lowest-cost eligible quote. The selected priorities favor its speed and lower illustrative risk.
| Cost component | Supplier A | Supplier B | Supplier C |
|---|---|---|---|
| Material | $1,008,000 | $912,000 | $864,000 |
| Freight | $48,000 | $66,000 | $30,000 |
| Carrying | $3,168 | $9,780 | $1,788 |
| One-time qualification | $0 | $0 | $30,000 |
| Total | $1,059,168 | $987,780 | $925,788 |
| Eligibility | Eligible | Eligible | Excluded: Qualification pending |
Validate the business case
The incumbent remains the front-runner, with no modeled cost reduction against its own quote. Validate the baseline, omitted transition costs, and timing with finance. This is a scenario comparison, not realized savings.
Confirm supply readiness
Confirm Supplier A's capacity commitment and delivery performance with operations. Validate the shorter replenishment route and maintain a credible backup source.
Set the negotiation agenda
Use competitive offers to challenge incumbent pricing. Explore formula pricing and supply commitments while developing qualified alternatives.
Three offers. Different commitments.
| Compare the offer | Supplier ARegional incumbent | Supplier BApproved distant source | Supplier CNew local candidate |
|---|---|---|---|
| Eligibility | Eligible | Eligible | ExcludedQualification pending |
| Material price / kg | $1.68 | $1.52 | $1.44 |
| Freight / kg | $0.08 | $0.11 | $0.05 |
| Landed price / kg | $1.76 | $1.63 | $1.49 |
| Modeled first-year cost | $1,059,168 | $987,780 | $925,788 |
| Quoted lead time | 14 days | 45 days | 10 days |
| Minimum order | 18,000 kg | 60,000 kg | 12,000 kg |
| Annual available capacity | 900,000 kg | 1,200,000 kg | 700,000 kg |
| Payment terms | Net 45 days | Net 60 days | Net 30 days |
| Qualification | Approved | Approved | Pending |
| Illustrative risk index0 = lower risk · 100 = higher risk | 20/100Established relationship and a short replenishment route; highest landed price. | 45/100A longer replenishment route and larger order commitment increase exposure. | 65/100Fast and competitively priced, with limited performance history and qualification still open. |
Inspect the model
First-year cost
Annual volume × (material price + freight) + cycle-stock carrying cost + one-time qualification allowance.
Average cycle stock = half the minimum order. Carrying cost = average cycle stock × landed unit price × 20% per year.
| Component | Supplier A | Supplier B | Supplier C |
|---|---|---|---|
| Material | $1,008,000 | $912,000 | $864,000 |
| Freight | $48,000 | $66,000 | $30,000 |
| Carrying | $3,168 | $9,780 | $1,788 |
| Qualification | $0 | $0 | $30,000 |
Where AI fits
AI assisted in building this prototype. The displayed scores and review brief come from explicit rules, with no live AI model generating recommendations. The fictional risk ratings are inputs to inspect, not AI predictions.
Scoring
Cost: lowest modeled quote = 100; highest = 0; other quotes fall linearly between them. All three quotes define this scale, including excluded candidates.
Speed: 100 × (60 − quoted lead days) ÷ 60. Resilience: 100 − the illustrative risk index. Each score is multiplied by its normalized priority weight.
Scores within 0.01 points are treated as tied. Risk ratings are scenario assumptions, not measured probabilities.
Scope of the model
Assumes stable annual demand, full MOQ consumption, one qualified source for all volume, and no capacity reservation. Excludes taxes, duties, safety stock, in-transit inventory, disruption losses, and other transition costs. Payment terms are shown for context and are not monetized. Candidate C’s quoted lead time starts after qualification; qualification duration is unknown.
No orders are placed and no suppliers are contacted.
From a business question
to a working tool.
The value is a repeatable analysis that gives stakeholders a clear starting point for a decision.
- 01
Define the sourcing problem
I set the brief: compare the cost to serve, expose supply tradeoffs, and keep qualification and capacity visible.
- 02
Use AI to build the workflow
AI assists with the implementation: calculations, scenario controls, comparison views, and the review brief. This prototype demonstrates that approach.
- 03
Test the logic, then challenge the answer
The model is checked for cost accuracy, ties, invalid inputs, and cases where no supplier qualifies. In practice, I would validate source data and assumptions with finance, quality, and operations before an award.
Read the Ecolab sourcing case. Its $3M result is separate from this AI-assisted prototype.