Read position, not a single stock number
On-hand, reserved, and inbound separately. Treating reserved as available is how you advertise a SKU you already promised.
Inventory
Inventory Agent
What to reorder today, what is about to stock out, and what has quietly become dead stock — from live position, lead times, and open POs. Not an unattended purchaser.
Last updated: August 31, 2026
The work today
Buyers still rebuild "what needs a PO" from a spreadsheet, last month's velocity, and a promo calendar in someone's head. Auto-send is worse: a flash-sale day becomes the run rate, reserved units look like a stock-out, and you stack a PO on inbound that lands next week. A reorder-point formula that already posts the right qty does not need an agent wrapped around it.
What the agent does
An inventory agent ranks SKUs as reorder, wait, stock-out risk, or excess from on-hand, reserved, inbound, lead time, and open POs. It drafts a purchase order. A named buyer sends it. Demand forecasting and PO construction are part of this family. Merchandising and margin sit here as a section — not a sixth top-level SKU.
Workflow
How a turn runs
Reads, tools, then a stop. Skip the stop and you have a demo.
Score the risk
Velocity, lead time, open POs, and the promo calendar you can actually attach. Rank reorder vs wait vs excess.
Draft the PO
Vendor, qty, and the evidence the buyer needs. The agent does not submit the order.
Buyer sends
A human with buying authority approves. Tokens that are not buyer or admin cannot create a purchase order.
Systems
What it is allowed to touch
Named tools only. Anything undeclared is unreachable, regardless of the prompt.
WMS / inventory API
Live position with as-of timestamps. A nightly dump without as-of is how the brief lies after 10 a.m.
Purchasing and vendors
Preferred vendor, lead time, and open POs so the agent does not stack a second order on inbound.
Demand inputs
Sales velocity and any forecast you trust enough to put behind a named tool — not a model guessing from the PDP.
Approval
Every purchase order is a proposal. Cedar should deny createPurchaseOrder for any token that is not a buyer or admin. If you cannot name the person who must approve every PO, do not build this agent yet.
What good looks like
Outcomes we will stand behind
Qualitative on purpose. We do not have published agent case studies, so we will not invent a percentage.
A risk brief, not an auto-PO
Buyers see reorder / wait / excess with evidence. The formula you already trust can keep posting; the agent covers the cases the formula cannot see.
Reserved and inbound stay visible
The brief fails if it collapses position into one "stock" number. That is a product rule, not a hope.
Margin is in the same picture
Dead stock and giveaway pricing show up here as a section and a linked guide — not a separate commercial family in v1.
This page is the commercial summary for an inventory agent. Reorder ranking, purchase-order drafts, and demand inputs each have a field-guide post. Those posts keep their canonicals. This page tells you what we sell and what we will not automate unsupervised.
Merchandising and margin
Which SKUs are quietly destroying margin, and which need a merchandising pass, is the same live inventory and price picture. We did not add /ai-agents/merchandising/. Read the margin intelligence guide, then come back here or to operations if the pain is the morning triage rather than the buy.
What this is not
- An agent that auto-sends POs.
- A single “stock” number the model is allowed to treat as truth.
- A replacement for a buyer who understands vendor politics and promo risk.
There is no native Shopify AgentCore inventory connector. Shopify, Magento, and custom stacks look the same at the tool boundary — we attach to your WMS or admin APIs.
How we start
Name the buyer. Expose on-hand / reserved / inbound as separate fields. Then the eCommerce AI Agents engagement. If join keys across SKU, warehouse, and open PO are fiction, start with the knowledge agent.
Field-guide depth
These posts are the long-form canonicals. This page does not replace them, and they are not redirected here.
AI Inventory Agent: What Should We Reorder Today? (2026)
An AI inventory agent ranks SKUs as reorder, wait, stockout risk, or excess — a risk brief, not an auto-PO. Reuse Gateway ~180→95 ms and ~$791/mo at 50K, not a client turns-of-cover KPI.
AI Purchase Order Agent for eCommerce (2026)
An AI purchase order agent drafts qty, vendor, and need-by from inventory risk and demand — a human still sends the PO. Reuse Gateway ~180→95 ms and ~$791/mo at 50K, not a client buying KPI.
AI Demand Forecasting Agent for eCommerce (2026)
An AI demand forecasting agent explains the number a stats model already owns — likely demand, why, and what to do. Reuse Gateway ~180→95 ms and ~$791/mo at 50K, not a MAPE claim.
AI Margin Intelligence Agent: Which eCommerce Products Are Actually Profitable? (2026)
Revenue is not profit. Rank SKUs only when cost, discounts, shipping, returns, ads, and fees exist — else unknown. Reuse Gateway ~180→95 ms and ~$791/mo at 50K sessions.
AI Agent vs Workflow Automation: What Should eCommerce Businesses Use? (2026)
AI agents do not replace Shopify Flow or OMS state machines. Hybrid is the default: rules own money movement; the agent returns a structured decision. Reuse Gateway ~180 to 95 ms and ~$791/mo at 50K sessions — not store conversion KPIs.
Human-in-the-Loop AI Agents for eCommerce: Where Automation Should Stop (2026)
HITL is a queue with session id and tool trace — not a prompt. Reuse Gateway ~180 to 95 ms and ~$791/mo at 50K sessions. Do not auto-approve on SLA timeout.
The eCommerce AI Agent Readiness Assessment: Is Your Business Ready? (2026)
Score org-wide AI agent readiness /30 — data, integration, process, governance, priority. Below 16, skip writes. Reuse Gateway ~180→95 ms and ~$791/mo at 50K sessions.
The engagement
eCommerce AI Agents on AWS
Production AI agents for eCommerce on Amazon Bedrock AgentCore — support and WISMO, inventory, merchandising, margin, returns and B2B. Tool boundaries, evals before launch, and a human on anything that moves money.
Agentic Commerce Readiness
Make your store sellable to AI shopping agents. ACP and UCP protocol surfaces, an MCP server over your catalog, and product data that survives an agent comparison — built on AWS.
Where this sits
Family page → industry or decide tree → the engagement. AWS cases on this site are not agent results.
Other families
Five kinds of work
Customer support
The same WISMO, tracking, and returns questions every day. An agent looks up the order, answers from live data, and hands off the moment it is out of its depth.
Customer support agent →
Sales & B2B
Quote requests, reorders, and account questions from buyers who will not tolerate a chatbot. An agent drafts the next step; a human still owns the relationship.
Sales & B2B agent →
Operations
The morning pile: exceptions, stalled orders, and the work nobody wants to triage. An agent ranks what needs a human today — it does not silently rewrite the warehouse.
Operations agent →
Knowledge & catalog
If orders, customers, products, and inventory cannot be joined, an agent will guess. This is the data layer that makes every other agent honest.
Knowledge agent →
Common questions
Before you start
Will the agent send purchase orders?
No. It drafts. A buyer with authority sends. An agent without a named approver is an unattended purchaser — we will not scope that.
When should we not build this?
When a reorder-point formula already posts the correct qty to the preferred vendor and nobody is reading promo calendars or open POs by hand. Do not wrap a working formula in an agent for the slide.
Where does merchandising and margin live?
As a section on this page, plus the margin-intelligence field-guide post. It is not /ai-agents/merchandising/ in v1. Heavy ops triage stays on the operations family.
Do you have inventory-agent case studies?
Not yet. The reorder, PO, demand, and margin field-guide posts are the published depth. Do not treat AWS cost or IoT cases as inventory-agent results.
Scope an inventory agent
Tell us how POs get made today. If you cannot name the buyer who must approve every one, we will stop there.