Mid-market companies do the same work Amazon does, with a fraction of the tooling. We started Fask to fix that.

I spent over twelve years at Amazon, eventually becoming a Principal Engineer leading mission-critical payment systems. The B2C payments and checkout platform we shipped processed over $800 billion a year, wired into the systems that ran ordering, fulfillment, accounting, compliance, customer support, and seller operations.

What I remember most clearly is that the visible part of an order, the checkout flow, the delivery email, the receipt, is maybe a tenth of the work. The other ninety percent is what happens behind it. Matching the invoice to the PO. Chasing the vendor when the shipment is late. Reconciling the payment across the processor, the bank, and the ledger. Filing the replacement. Closing the case in the ERP.

At Amazon, we had platforms for all of that. Long-running services that handled the work end to end. Engineers paged when something stalled. Auditable logs of every decision.

After Amazon, I spent two years building an AI startup. That stretch taught me what frontier models can actually do, and where the seam is between a research demo and something an enterprise can run in production.

My co-founder George Koshy was on the same team at Amazon. He shipped the B2C payments and checkout platform with me, then went to Coinbase as an Engineering Manager running payments, fintech, and RevOps engineering. Same back-office instincts, two complementary angles. I leaned into platforms and infrastructure. George leaned into RevOps and the systems customer-facing teams actually live in.

When we started talking about what came next, the same picture kept showing up. Mid-market companies, retailers, wholesalers, distributors, brands, were running the same kinds of operations Amazon ran. Vendor relationships, PO lifecycles, EDI traffic, carrier coordination, monthly close, supplier scorecards. Without Amazon's tooling. Without Amazon's headcount. The gap between what their teams could process and what was coming in was getting wider every quarter.

We ran dozens of conversations with VPs of Operations, Supply Chain, and Finance. The pain was consistent and loud.

A retailer with three hundred vendors reading four thousand emails a week to find the few hundred that actually needed action. POs tracked in a spreadsheet that someone refreshed on Friday by clicking through carrier portals. Replacement approvals stuck in a vendor's inbox for two weeks while a customer asked for an update every day. NetSuite, SAP, Oracle, or Dynamics 365 records lagging the truth by a day or two because data entry happened after the work, not during it. Monthly close taking six to seven business days because partial payments, multi-entity allocations, and processor fees never matched cleanly.

The tools they reached for were either built for a different problem or stopped at the conversation. Front-office AI closes the customer chat but never reaches into the ERP. RPA records clicks and breaks the moment a supplier swaps an email template. Workflow builders work brilliantly until the logic spans an ERP, EDI, carrier APIs, and supplier email, at which point the canvas becomes the work. ERP copilots summarize what is already in one system but cannot read what is about to land in it.

What teams were asking for was an AI version of the operational platforms we had at Amazon. Agents that would log into the ERP, read the email, process the EDI, pull the carrier API, run the SOP, and close the loop. Without ripping out anything they already use.

Fask is autonomous AI agents that execute on the systems your business already runs. Not chatbots. Not flow charts. Agents that log into NetSuite, SAP, Oracle, or Microsoft Dynamics 365, process EDI 850 and 856, pull UPS and FedEx APIs, read vendor email in thirty-plus languages, and run the work your team currently does by hand.

We ship four pre-built solutions for the verticals where the pain is sharpest:

Vendor Ops handles email triage, PO tracking, replacements, billbacks, carrier sync, and supplier scoring for retailers, wholesalers, distributors, and brands managing hundreds of vendors. Procurement Ops handles requisition triage, supplier selection, contract compliance, and spend analysis for teams whose policy gets enforced after the money is committed. Finance Ops handles invoice reconciliation, AP and AR automation, payment matching, revenue forecasting, and monthly close for finance teams running multi-processor and multi-entity reconciliation. Inventory Ops handles stock monitoring, demand-driven reorder, phantom inventory detection, and warehouse coordination for ops teams reconciling WMS, ERP, 3PL, and sales channels.

Underneath, three thousand integrations via Myst, our agent platform, plus custom integrations into the internal APIs and ERPs your team already runs. The agents read the SOPs your team already has, score every action with confidence and a full audit trail, and route low-confidence cases to humans for review. No flow charts to maintain. No hidden hallucinations. No items dropped because nobody chased them.

The numbers we see in the first quarter of a deployment are typical: thirty to seventy percent of routine back-office workload absorbed, eighty percent of vendor email auto-classified, hundred times faster reconciliation, late POs surfaced in real time instead of in weekly batch.

We started with vendor operations because that was our anchor customer's most painful workflow. Procurement, Finance, and Inventory Ops are running on the same platform with the same agent shape. We keep adding verticals and depth based on what customers ask for.

The teams we work with are tired of being asked to grow back-office headcount at the same rate as revenue. The technology to stop doing that finally works. If your team is reading vendor emails, chasing late POs, or closing books in a six-day fire drill, we should talk.