Identify the buyer
Shopper, authenticated buyer, and internal associate are different tokens. The wrong one is how account B sees account A price.
Sales and B2B
Sales & B2B Agent
Quote requests, reorders, and "what is my price" questions from buyers who will not tolerate a chatbot. The agent drafts the next step. Sales still owns the account.
Last updated: August 31, 2026
The work today
B2B buyers ask for net price, availability, and last-order repeats. Retail chat that reads the public PDP will quote list price into a contract account. That breaks margin and most-favored-customer clauses in one turn. The other failure is an agent that places the PO or captures payment because it was told to be helpful.
What the agent does
A sales agent answers from permissioned customer price, inventory, and order history. It drafts a quote or a reorder cart. The commerce system takes the order after the buyer confirms. High-value quotes and any price exception go to an associate. Account management — "what is open, what is late, what renews" — nests here. It is not a separate family in v1.
Workflow
How a turn runs
Reads, tools, then a stop. Skip the stop and you have a demo.
Read contract context
Customer price, availability, and order history from named tools. If there is no account-matched price, say so and hand off. Do not interpolate from the PDP.
Draft the next step
A quote, a reorder cart, or an account brief. The buyer or associate confirms. The agent does not capture payment.
Escalate exceptions
Off-book discounts, new ship-to, and credit questions leave the tool catalog and enter a human queue.
Systems
What it is allowed to touch
Named tools only. Anything undeclared is unreachable, regardless of the prompt.
Price book and ERP
Customer-specific price, not the public catalog API. Fail closed when the account does not match.
Inventory and ATP
What can actually ship, including reserved and inbound — not a single "in stock" flag.
CRM / account record
Assigned associate, open quotes, and the accounts that associate is allowed to see.
Approval
No payment capture in agent tools. High-value quotes and any override of contract price wait for a named associate. The agent makes the decision cheap to take correctly; it does not take it.
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.
Faster first response on RFQs
The buyer gets a drafted quote from their price book, not a "we will get back to you" holding pattern — or a retail price that is wrong.
Reorders without a treasure hunt
Last-order repeats become a cart the buyer confirms, with availability called out before anyone promises a date.
Associates keep the relationship
The agent is a junior on the account, not a replacement. Exceptions still land on a person who owns the book.
This page is the commercial summary for sales and B2B agents. Quote construction, identity (shopper vs buyer vs associate), and account-manager workflows are documented in the field-guide cluster linked below. Those posts remain the long-form ranking URLs.
Account management nests here
“What is open on this account, what is late, what is up for reorder” is the same permissioned context as a quote. We did not create /ai-agents/account-management/. If that is the pain, start with the account-manager guide, then the engagement at eCommerce AI Agents.
What this is not
- A retail site chatbot that “also does wholesale.”
- An agent that captures payment or silently places a PO.
- A generic outbound SDR product. We go deepest where there is a price book, an order history, and an associate who still owns the relationship.
How we start
If you cannot call customer price without leaking another account’s book, you are not ready for a buyer-facing agent. That is a data and identity problem — see the knowledge agent — not a prompt problem.
Field-guide depth
These posts are the long-form canonicals. This page does not replace them, and they are not redirected here.
AI Sales Agent for B2B eCommerce (2026)
B2B sales agents must call permissioned customer price — not the public PDP. Reuse the B2B CRM Gateway canary (~180 ms → ~95 ms, 12 tools, ~8k turns/day). Shopper vs buyer vs associate JWT.
AI Quote Agent: Automating B2B eCommerce Quote Requests (2026)
B2B quote agents draft from customer price and ATP — they do not auto-issue above your cap. Reuse Gateway ~180 ms → ~95 ms on the B2B CRM canary. Artifact thresholds are fixtures, not client SLAs.
AI Account Manager Agent for B2B eCommerce (2026)
An AM agent should cap "accounts requiring attention today" at 5 rows with evidence_tool on each. Reuse Gateway ~180 ms → ~95 ms on the B2B CRM canary — not a book-of-business KPI. Demo table is fixtures.
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 →
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 →
Inventory
What to reorder today, what is about to stock out, and what has quietly become dead stock — answered against live data, not last month's report.
Inventory 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
Can this sit on the public storefront?
Not if the only catalog API returns list price. A retail copilot that quotes the PDP to a contract buyer will break the price book. Customer price stays behind a buyer or associate token.
Does the agent place the purchase order?
No. It drafts the cart or quote. The commerce system takes the order after the buyer confirms. Payment capture is out of the tool catalog.
Where does account management live?
Here. Open orders, late shipments, and renewals are the same permissioned account context. There is no /ai-agents/account-management/ in v1.
Do you have B2B agent case studies?
Not yet. The B2B sales, quote, and account-manager field-guide posts are the published depth. Do not read AWS case studies on this site as agent results.
Scope a sales agent
Tell us how quotes and reorders move today. If the price book cannot be called safely, we will say so before anyone builds a demo.