New · RetailFinance
Dead stock, spotted early. Not written off late.
Scans sell-through, aging, and margin signals across your full catalog every day and flags at-risk SKUs weeks before they'd otherwise hit a clearance rack — with a cited markdown depth and timing recommendation for every flagged item. The approach that recovered $41,200 in projected markdown loss for ClearanceIQ.
10 fields cited per flagged SKU. Built on Sphere AI Foundry.
markdown_scan — catalog run #4,812 SKUsLIVE
00:00Ingest → 4,812 active SKUs, 6 categoriesread
00:03Sell-through check → SKU 8842-BLU tracking 34% behind planflagged
00:06Aging check → 91 days on hand, 3x category averageflagged
00:09Margin impact → current margin 48%, floor 22%passed
00:11Recommendation → 25% markdown, apply within 9 dayssealed
scan time 41sSKUs flagged 187/4,812planner sign-off required
Built on Sphere AI Foundry·$41,200 recovered for ClearanceIQ·21 years, 300+ clients
Dead stock is a decision made three weeks too late
By the time a SKU shows up on a slow-mover report, it's usually already lost most of its margin cushion. Planners are watching hundreds or thousands of SKUs across categories, and a weekly review cadence catches the obvious laggards but misses the ones quietly falling behind plan in week two or three — right when a modest early markdown would have protected far more margin than the deep clearance markdown that follows once the item is truly stuck.
$41.2K
projected markdown loss recovered for ClearanceIQ
3–4 wks
earlier detection vs. a weekly slow-mover report
10
fields cited per flagged SKU
Catalog in, ranked action list out
Four steps, run daily, every flag traceable back to the underlying sales and inventory data.
01
Ingest
Reads the full catalog directly from your systems
Daily sell-through, on-hand units, receipt dates, and current margin by SKU — read from your merchandising system and POS, not a weekly export someone has to pull.
Source data
4,812 active SKUs — 6 categories, 214 stores + e-commerce
02
Compare against sell-down plan
Every SKU checked against its own plan, not a flat rule
Sell-through and aging are compared to that SKU's planned sell-down curve for its category and season, so a slow week for a long-tail basic isn't treated the same as a slow week for a seasonal item three weeks from changeover.
Example query
"Which SKUs in Outerwear are tracking more than 25% behind their sell-down plan?"
03
Cite every flag
No flag without the numbers behind it
Each flagged SKU carries the exact sell-through gap, days-on-hand, and margin figures it was flagged on, plus the comparable past markdown pattern the recommendation is modeled against — the citation travels with the flag.
Cited finding
SKU 8842-BLU: sell-through 34% behind plan, 91 days on hand (3x category average), margin 48% vs. 22% floor.
cited to POS sell-through report, wk-ending 8/2304
Recommend & route
One ranked list, one recommendation per SKU
The agent returns a ranked action list with a suggested markdown depth and timing window per SKU, routed to the planner for sign-off before it reaches pricing or POS.
Recommendation
MonitorMarkdown 25% ✓Clearance now
10 fields, every one cited
What the agent extracts and produces for every SKU it flags.
Risk classification & ranking
Sell-through rate vs. plan
Weeks-of-supply & days-on-hand
Current margin vs. margin floor
Recommended markdown depth
Recommended markdown timing window
Comparable historical markdown pattern
Projected margin recovery if actioned
Store vs. e-commerce sell-through split
Consolidated ranked action list
Typical inputs
What the agent reads, and what it connects to.
Documents
- Weekly and daily sell-through and sales reports by SKU
- Inventory aging and on-hand reports
- Category sell-down and markdown calendar plans
- Buy plans and open-to-buy positions
- Historical markdown and clearance performance records
Systems
- Merchandising / assortment planning system
- Point-of-sale (POS) platform
- Inventory management system
- Pricing and markdown optimization tool
Is this the right fit?
Built for catalog-wide monitoring — not a one-off pricing decision.
✓ Works best for
- Retailers running 1,000+ active SKUs across multiple categories
- Merchandising teams working to a defined sell-down plan and markdown calendar
- Planning orgs wanting flags surfaced days or weeks ahead of a scheduled review
Too small for
- A small, hand-curated assortment a single planner already reviews daily
- Replacing a planner's final pricing decision
- Categories with no sell-down plan or historical markdown baseline to compare against
Grounded in your plan, not a black box
Compliance-first, the same standard every Sphere agent is held to.
The agent requires a configured sell-down plan and margin floor per category before it will return a recommendation, and every flag sits between your daily sales data and the planner's pricing decision — never past it. A planner confirms every markdown recommendation before it reaches pricing or POS.
Recommendations reference your own sell-down plan and historical markdown baselines directly — see how the same governed, cited-recommendation standard applies across other data-driven deployments in Governed AI for Retail Merchandising and Private LLM Deployment for Retail Inventory Systems.
Prerequisites sell-down plan
A configured sell-down plan and margin floor per category required before a recommendation is returned.
Human review required
A planner confirms every markdown recommendation before it reaches pricing or POS — never fully autonomous.
Regulatory grounding internal plan
Every recommendation references your own sell-down plan and historical markdown baseline directly in the flag.
Foundation Sphere AI Foundry
Access controls and audit trail configured here carry forward to every other agent you deploy.
Frequently asked
Does the Markdown & Inventory Intelligence Agent set prices automatically?
No. It flags at-risk SKUs and recommends a markdown depth and timing window, but a merchandising planner approves every price change before it goes live in your pricing or POS system.
What counts as an at-risk SKU?
A SKU where sell-through rate, weeks of supply, and inventory aging cross your configured thresholds relative to its planned sell-down curve — for example, tracking well behind plan with markdown season approaching. Thresholds are set per category during onboarding, not hard-coded.
How does it integrate with our existing merchandising and POS systems?
It reads directly from your merchandising system, POS, and inventory management platform — no separate upload step required. Integration scope is confirmed during deployment planning.
Is every markdown recommendation traceable back to the underlying data?
Yes. Every flagged SKU cites the specific sell-through, aging, and margin figures it was flagged on, alongside the comparable historical markdown pattern the recommendation is based on — there is no recommendation without a cited data trail.
How is this different from a standard inventory dashboard?
A dashboard shows you the numbers; this agent reads them against your sell-down plan every day, ranks which SKUs need action first, and hands a planner a cited recommendation instead of a report to interpret. It runs on Sphere AI Foundry, so the access controls and audit trail you set up here carry forward to other agents you deploy.
Start here
See it on your own catalog
Bring a recent sell-through export to the demo. We'll show the ranked action list live — flagged SKUs, cited figures, and markdown recommendations, on your own data.
Looking for a different workflow? Browse the full Agent Catalog.
Talk to a solutions architect
Direct to a senior architect — never a sales queue. Replies within one business day.
By submitting, you agree to be contacted about this request. We never share your details.