Skip to content

Xfuze Order Intake Agentic Application

Up to 5x faster order processing than manual entry, with a confidence rate of 99%.

Operating within a multi-agent ecosystem, three agents turn B2B orders sent by email, spreadsheet or PDF into draft orders that a customer service team reviews and submits.

What the agents do

Decides whether an email is an order, matches each line to a current product, confirms the customer details, scores its own confidence, and flags lines for a person to review.

Brand
B2B CPG suppliers (clients not named)
Built by
Xiatech, on Google Cloud
Contributed by
Xiatech (Agentic order intake application)
Industry
Manufacturing & Industrial, CPG, Grocery, Retail
Audience
B2B
01
The problem

Why this exists

Many B2B suppliers still take most of their orders by email. Customers send them in no standard format: in the email body, in Excel or PDF attachments, sometimes as screenshots, and often with their own product codes or codes that have since been replaced.

Read more

Customer service teams key each order into the order management system or ERP by hand. Volume peaks before the daily cut-off, so the work is slow and error-prone exactly when time is shortest, and skilled people spend the day on data entry instead of on customers and exceptions.

02
Outcomes

What changes when this is in place

  • Up to 5x faster order processing than manual entry.
  • An order confidence level of 99%.
  • Order intake runs 24/7, with a person still approving every order.
Read more
  • Every order sent is confirmed by an order number from the receiving system, every failure raises an alert, and each correction becomes a matching rule that applies to every later order. Targets are agreed with each customer against its own baseline.
03
How it works

Architecture

The application can run standalone in any cloud environment, connecting through MCP and REST APIs to a business's existing inbox, product and customer data, and order management system, with its own review queue, audit log, and monitoring. It can also run on the Xfuze platform, where Xfuze handles intake, egress, and data, and the agents are governed in the Xfuze AI Control Tower.

Read more

The Order Processing Agent decides whether an email is an order, reads the body and attachments as one order, identifies the customer and ship-to, matches each line to a current product through customer codes, legacy codes, and substitutions, and scores every line. The Order Review Agent opens the queue for operators, goes straight to flagged lines and shows the customer's own wording beside its reading. Only this agent can release an order. The Order Monitoring Agent gives supervisors KPIs, full order history, live alerts, and control over matching rules. The three agents coordinate with one another over the Agent-to-Agent (A2A) protocol.

Approved orders go to the order management system through MCP, a REST API or a structured file. Sends are queued for peaks, retried on failure, and alerted, and every match, score, human action, and submission is logged.

B2B order emails, with the order in the body, Excel, PDF or screenshots, and product and customer data covering customer codes, legacy codes, and substitutions feed the Order Processing Agent. It decides whether the email is an order, identifies the customer and ship-to, matches each line, and scores confidence. The draft goes to the Order Review Agent, a review queue that shows flagged lines beside the customer's wording, and operator corrections flow back as matching rules. When the operator accepts or amends, the order goes to the order management system or ERP through MCP, a REST API or a structured file. An order number confirms receipt to the Order Monitoring Agent, which gives supervisors KPIs, order history, live alerts, and rule changes.

B2B order emails, with the order in the body, Excel, PDF or screenshots, and product and customer data covering customer codes, legacy codes, and substitutions feed the Order Processing Agent. It decides whether the email is an order, identifies the customer and ship-to, matches each line, and scores confidence. The draft goes to the Order Review Agent, a review queue that shows flagged lines beside the customer's wording, and operator corrections flow back as matching rules. When the operator accepts or amends, the order goes to the order management system or ERP through MCP, a REST API or a structured file. An order number confirms receipt to the Order Monitoring Agent, which gives supervisors KPIs, order history, live alerts, and rule changes.
Open standards used
Model Context Protocol (MCP)Agent-to-Agent (A2A)REST APIs
Tech stack
Google CloudGemini Enterprise (optional)Xfuze platform (optional)Xfuze AI Control Tower (optional)
Next step

Request an intro

Xiatech is behind this solution. Ask about taking order entry off the customer service team, and the MACH Alliance will put you in touch with the person who can answer.

The MACH Alliance passes your details to a named contact at Xiatech who knows this work and can answer your questions. We aim to reply within 2 business days. Your details are handled under the MACH Alliance privacy policy.

Not ready to talk to anyone? MACH Insider is the Alliance mailing list. New solutions, program openings, and the research behind them, once a month.

Everyone credited on this solution

  • XiatechAgentic order intake application

Every organization named here has a page listing the rest of its work on the Hub.

Keep reading

Related solutions

Verification

Contributor submission

The detail on this page comes from the contributor's own account of their own work. The MACH Alliance reviews a submission before publishing it and does not independently verify the figures in it.

Where this came from

Solution detail is drawn from Xiatech's submission to the hub. The application is listed on Google Cloud Marketplace.

Status: Live in productionAdded Oct 26, 2026A listing is an editorial decision, not a MACH Alliance certification.

Built something worth showing?

Put your agentic solution in front of the enterprise leaders who use the hub to decide what to build next. Submissions are open to MACH Alliance Members, Collaborators, or Ambassadors.