Earlier this week, our CEO talked about the illusion of departmental silos. He explained why enterprise AI will fail if it relies on isolated bots that can only do one job. To truly accelerate a supply chain, specialized AI agents must be able to securely communicate with each other across departments.

Today, we are thrilled to show you exactly what this neural network looks like in practice, with a deep dive into our new Signal Mesh, the RFQ Agent, and the Vendor Selection Model.

The No Stock Brick Wall 

In a traditional ERP, operations grind to a halt the moment you run out of parts. If a salesperson builds a quote and realizes there is zero stock, the software simply flashes a warning. The system stops. It is then up to the salesperson to switch screens, email the purchasing team, and wait for them to start the manual sourcing process from scratch.

With the AvSight Signal Mesh, that brick wall is gone. As the Quote Agent processes a quote, it runs our Source of Supply model to search every internal source such as warehouse inventory, in-transit purchase orders, and repair pipelines. When that model determines there is truly zero internal stock for a line item, it doesn’t just warn the user—it publishes a secure signal into the Signal Mesh, routing it instantly to the right team.

Proactive Procurement: The RFQ Agent 

Sitting in the purchasing department, our new RFQ Agent is constantly listening for that signal. The second it hears “No Stock,” it springs into action, often while the quote is still being processed.

To find the right parts, the RFQ Agent utilizes our new Vendor Selection Model. Finding a vendor isn’t just about finding the lowest price on an external marketplace; it is about weighing lead times, historical reliability, active contracts, and strict quality certifications.

The Vendor Selection Model acts as a master procurement assistant. It instantly synthesizes external market data (like ILS) and cross-references it against your native AvSight ERP history—past RFQs, purchase orders, vendor responsiveness, and contract pricing—to rank the most reliable suppliers for that specific part and condition.

Shadow RFQs and Absolute Human Control 

So, does the AI just start firing off emails to vendors and spending your company’s money? Absolutely not.

True to our philosophy of keeping humans in command, the RFQ Agent performs this heavy lifting as shadow work. It prepares a complete, drafted solution in a secure background staging area we call the Agent Augmented Context (AAC).

Instead of starting with a blank screen, your Purchasing Manager opens a finished draft. They see the requested part, the AI’s top-ranked vendor recommendations, and the exact reasoning behind those suggestions. The manager reviews the research, selects their preferred vendors, and clicks “Approve” to officially create and send the RFQ.

A Dial, Not a Switch 

We know every purchasing department has different comfort levels with AI. That is why the RFQ Agent is a dial, not a switch.

Every signal subscription comes with configurable conditions and action chains. If you prefer, you can set strict, rules-based conditions that control exactly when and how the agent acts, for example, only triggering vendor sourcing for specific part categories or above a certain dollar threshold. You don’t need generative AI for every scenario. You can rely on deterministic business logic for your routine purchasing, and only leverage the AI’s Vendor Selection Model for complex or rare part sourcing

Looking Ahead 

By allowing the Quote Agent and the RFQ Agent to talk to each other through the Signal Mesh, we have bridged the gap between Sales and Purchasing in seconds

This magic is made possible by the background staging area: the Agent Augmented Context (AAC). Next Thursday, we will take a closer look at how we are rolling out AAC across the entire platform, turning every empty form into a finished, reviewable solution.

(Missed the strategic breakdown of the Signal Mesh? Read our CEO’s piece on the Neural Network of the Supply Chain here: The Neural Network of the Supply Chain: Why Your AI Agents Must Talk to Each Other)