This is your assignment from the July 24 daily, turned into a concrete, fast-to-execute plan: redesign the Intake → Item Matching → Supplier Sourcing part of the Bid flow in Figma, on top of your Smart Basket base, with AI “killer features” and the corrected process order. The clickable prototype now covers all three screens.
01What you were actually asked to do
Alex: “The current UX is terrible. Shelby’s system confuses more than it helps. If Magalí can give her expert opinion to improve Shelby’s design in Figma, that’d be great. We need killer features. Focus on what AI can automate — if something can be pre-filled, describe it… use realistic supplier data (ratings, history) to beat the other consultants.”
Improve the design in FigmaRedesign the confusing Bid flow. Base = your Smart Basket prototype from Nicoleta.
Show what AI automatesAnything the system can pre-fill or auto-match → make it visible and explainable.
Realistic supplier dataRatings, on-time %, win history — the differentiator vs the rival consultants.
Fix the flow orderMatch items first, source suppliers after. See section 03.
02What the three uploads are for
- Whiteboard “BID AI Discovery” — the 11-step process map. Two rows matter for you: “Include in Demo” (which steps to design) and “Opportunity” (the AI improvements per step). Your two steps — Item Matching (#6) and Identify Suppliers (#4) — both say “include in demo.”
- 30 “Bid Genie” screenshots — the IT-built demo (
localhost:7001). This is the “built by IT without domain knowledge” version. Use it as the reference for what exists and what to fix, and reuse its clean visual shell.
- Transcriptions — the business rules that make your design look domain-aware: Sysco Brand (margin), Active Stocked, CN Labels (K-12), By-American law, sodium limits, A/B/C alternates, RBSM final decision.
03The flow fix — the one thing you can’t get wrong
Alex: “In Eugene’s scheme, supplier matching happens before item selection. That makes no sense… If we show a wrong flow on Thursday they’ll say ‘you don’t understand our business’ and that’s the end of the demo.”
| Today (Bid Genie) | Your redesign |
|---|
| Suppliers → then items | Search catalog → match items → then find suppliers who carry them in-region |
In the prototype this shows as the stepper order (Item Matching before Supplier Sourcing) and the green “correct order” banner. Keep this order in every Figma frame.
04UX audit of Bid Genie — what to improve
| Problem in the demo | Fix in your design |
|---|
| Supplier outreach happens before items are matched | Reorder: matching first (section 03) |
| “Not in catalog” shows as a dead red status with no next step | AI suggests the closest SKU + a one-click “Source supplier” action |
| Match is keyword-based (“black magic”, per Alex) with no reasoning | Semantic match % + a “Why this?” explanation on every candidate |
| Domain rules (Sysco Brand, CN, By-American) aren’t visible | Auto-flagged attribute badges on each candidate SKU |
| No A/B/C alternates, though the manual process uses them | Show ranked alternates A/B/C per bid item |
| Suppliers are a flat list with no quality signal | Rank by rating, on-time %, volume, win history, price + rule-set sort |
| Intake makes you type contact info manually, then AI overwrites it | Upload-first: AI pre-fills; flagged/missing fields → “Ask the customer” |
05The killer features (priority order)
1 · Semantic match + %
2 · Auto domain badges
3 · A/B/C alternates
4 · “Why this?” explainability
5 · Rich supplier data
6 · Rule-set pre-selection
- Semantic match candidates. Vague line like “BBQ sauce, no brand/size” → ranked Sysco SKUs with a match %. Beats the keyword AS tool.
- Auto-flagged attribute badges. AI checks each candidate for Sysco Brand (margin), Active Stocked, CN Label (K-12), By-American, Low-Sodium. This is the domain knowledge IT’s UI lacks.
- A/B/C alternates auto-generated when the exact item isn’t in catalog.
- Confidence + “Why this?” — keep Bid Genie’s High/Medium/Low review triage, add a one-line AI reason.
- Supplier sourcing with rich data. Alex was explicit (Jul 24): go past the star rating in the current demo — show volume/revenue with Sysco, orders delivered per year, on-time %, past winning bids, price, pulled from the product graph. This is the differentiator vs the rival consultants.
- Rule-set pre-selection + AI pre-fill. The AI pre-selects the best SKU/supplier by a rule set (e.g. Sysco Brand → cheapest) so the human only reviews — “the less routine, the better” (Alex). Same idea on Intake: upload the RFP first and let AI pre-fill company/contact/items, instead of manual entry.
06How to execute fast in Figma
- Open
prototype.html next to Figma — it’s already built on your Smart Basket design system (solid Sysco-blue header + real white logo, blue category bars, product cards with the “Sysco” blue brand prefix, pill buttons, footer action bar) using the exact brand blue #008CD2 from Sysco Logos.svg. Type is Myriad (Smart Basket’s font): the mock’s font-stack prefers Myriad Pro if it’s installed on your machine (via Adobe), and falls back to a close humanist sans otherwise — in Figma, set the real Myriad.
- Start from your Smart Basket file; reuse its components (cards, chips, category bars, footer bar) — the prototype maps 1:1 to them.
- Build the 3 frames in the prototype: Intake (upload-first + AI pre-fill + “Ask the customer”), Item Matching, and Supplier Sourcing. Don’t redesign the whole 11-step flow yet.
- Make the badge set a Figma component with variants (Sysco Brand / Stocked / CN / By-American / Low-Sodium) — you’ll reuse it everywhere.
- Add the corrected stepper at the top so the order reads at a glance.
- Populate with the real item + supplier names from the demo (fries, cereals, cheese; General Mills, McCain, Bongards’) so it feels authentic — Alex’s “realistic data” note.
- Alex wanted a short meeting to map the demo ↔ Figma before you commit a date — use this prototype as the artifact for that call, then aim a first pass Monday-ish.
- Prep one line for the daily: “I redesigned matching to be catalog-first, with AI match %, auto domain badges, and ranked suppliers — here’s the click-through.”
More opportunities (beyond the 3 frames — mention, design later)
- Bid Verification tab. Alex wants a step to compare the original documents vs what AI extracted, side by side. The Intake frame stubs this as a “Compare to source” button — turn it into a full side-by-side step.
- Per-supplier subset outreach. Don’t email every supplier all 22 items — ask each only for the lines they can supply (dairy supplier → dairy lines).
- Extract twice / Consolidation. Items are matched once to pick who to contact, then again after suppliers confirm what they can ship — consolidation is where the final per-supplier pick happens.
Ask the team to confirm (open questions)
- Priority rule conflict: Alex said Sysco Brand first, then cheapest; Natalie understood Trisha as cheapest only. Confirm the real rule set — it drives pre-selection.
- Are they legally obliged to contact all matching suppliers (audit), even if only the top ~3 are chosen?
- Do they contact suppliers for all RFP items or only a subset? (subset → only after consolidation)
- Does By-American / CN labeling apply to all K-12 items, or only protein? (which badges are mandatory)
- Is the region driven by the customer’s delivery region automatically, or chosen by the RBSM?
- For “not in catalog” — contact a supplier to add the SKU, or always offer the closest alternate?
Open the clickable prototype →