Avent wants your reps back on the phone.

Every growth conversation seemed to end with another person. In the electrical and industrial distribution business where Abhay Kalra grew up, there were customers to pursue and opportunities to chase. But the sales team was occupied with the customers it already had: preparing quotations, entering orders, finding information. Adding business meant adding people, then waiting for those people to learn the business.

The growth conversation that kept ending in hiring

Kalra’s father built the family’s distribution business in India. Kalra and his brother kept returning to its hiring constraint when they discussed expansion. The predicament was mundane enough to be familiar and expensive enough to matter: experienced people were busy processing demand, leaving less time to create it.

“Growth was always limited by how fast we could find and train good people.”

ABHAY KALRA / MDM, 2025

Avent, the company he founded that year, took that observation as a product brief. Make the existing team’s knowledge usable without requiring the existing team to perform every repetitive step.

The part number is only the beginning

Avent sells AI software to industrial distributors and manufacturers. Its original proposition follows a request through quoting, product search, and order entry. Customers can keep using familiar channels while the software does the translation into structured commercial work.

A request for quotation, or RFQ, may describe a product using shorthand rather than its exact catalog identifier. Avent says it interprets that language, matches the item to a SKU, and applies the relevant pricing rules. Customer agreements, discount tiers, availability, and configuration details can all affect the answer. A plausible product match is therefore the beginning of the quote, rather than its completion.

If stock is unavailable or vendor pricing is needed, Avent describes checking alternatives, costs, and lead times under the distributor’s sourcing rules. When a purchase order arrives, the software can match it to the quote, check quantities and terms, and prepare the sales order for the ERP. Each handoff carries commercial consequences.

A useful place to put the brakes

The order-entry interface places extracted information alongside the original email and attachments. A person can inspect the line items and correct missing or incorrect details. The workflow determines what validation and approval must happen before an order reaches the ERP.

That is a consequential design choice. A system that reads a quantity correctly can still apply the wrong customer terms. A system that drafts a quote should not automatically gain permission to approve it. Avent describes separate permissions for viewing, editing, approving, and executing work.

Its Forge interface makes workflows editable. Arc, its AI copilot, can prepare proposed changes for review. In a homepage example, a manager asks for approval before sending quotes below a specified margin. The demonstration is labeled sample data. Its point is control: the business determines where discretion belongs.

What the quote counter actually counts

At its 2025 YC launch, Avent said it generated more than $200,000 worth of quotes daily for distributors across North America and the UK. That measures the value of offers prepared. It does not establish booked sales, customer savings, or Avent’s own revenue.

MDM reported a $500,000 investment from Y Combinator and customers in the United States, Mexico, and Britain. Avent’s commercial route is a sales conversation and a demonstration. For a buyer, the useful cost calculation includes the integration and review effort alongside any software fee.

A sensible trial would count minutes spent preparing quotes, corrections per order, and the requests that still need expert attention. It would also track whether faster preparation produces a faster customer response. Moving a queue from a salesperson’s inbox to an approval screen merely relocates the waiting. The business benefit depends on what happens after the draft appears.

The company memory problem

Kalra studied electrical engineering and computer science at UC Berkeley, including computer vision research on manufacturing defects. Avent addresses another kind of operational fragility: important knowledge sitting with a few people.

Its Knowledge Base brings contracts, pricing agreements, and policies into designated workflow steps. Staff can ask Arc questions about connected business information within their access permissions. Expert corrections can become proposed improvements for the team to approve, keeping exceptions available when employees change roles.

The surrounding tools address the same bottleneck from different directions. Inbox Management routes requests and extracts tasks. Voice Agents transcribe calls and prepare draft quotes for review. The common ambition is to give a conversation somewhere useful to go after it ends.

The inbox has company

Avent occupies a busy corner of enterprise software. Distro also markets AI quoting, catalog matching, customer pricing, and ERP submission to distributors. Conexiom offers order and invoice automation for manufacturers and distributors. Buyers have alternatives beyond another hire and another spreadsheet.

Avent’s particular emphasis is the connected operating model: customers, products, transactions, rules, and people working together. Its engineering recruitment reinforces that approach. A forward-deployed engineer is expected to understand customer constraints, write integrations, and own deployments. The small team works in person in San Francisco, close to the operational problems it is trying to encode.

Start with the awkward order

There is a practical lesson here for any operator considering automation. Pick a real request with an ambiguous description, a special agreement, or a missing item. Follow it through the quote and the purchase order. Watch where someone intervenes, what gets corrected, and whether that correction survives the next transaction.

Avent’s April 2026 ISA26 announcement put supplier data quality on the agenda beside live demos and a custom-agent workshop. That pairing identifies a real constraint. Incomplete product information, inaccessible agreements, and undefined approval authority limit what connected automation can reliably do. The work begins with making those things usable. Then a salesperson might finally have time to make the next call.