An AI-First Company Finds Its Supply Gap

The Company That Bet on AI Early

A few years ago, Slava Syrota, chief technology officer at Ecentria, the parent company of OpticsPlanet, Inc., made a call that most engineering leaders would have considered reckless. He told his engineering team that they were allowed to use any AI tool they wanted, no guardrails, no approval process, no defined use case, no token limits. Just experiment, and report back on what worked and what didn’t.

It worked. By the time AI adoption became mainstream in other companies, Ecentria was already in the stabilization phase. Engineers at the company had been doing agentic engineering for several years and had experience using different tools and models. They developed a good understanding of the pros and cons of agentic engineering well before others. In 2025, the company adopted Claude Code as its primary agentic engineering tool and Claude Cowork as the AI of choice for the business, though ChatGPT and other frontier labs remain available for certain individuals and use cases. In 2024 and 2025, the company deployed several internal AI chatbots that boosted efficiency across the business, including AI-assisted code writing and code review tools, an internal customer service chatbot, an employee-facing chatbot with visibility into internal systems and tools, and other AI tools. Adoption of AI transformed Ecentria’s operations, streamlined data exploration, and enabled automated AI workflows to take over many responsibilities that were previously handled manually.

Syrota’s current focus has expanded from engineering to business use cases. Having standardized AI across his technical teams, he continues building the infrastructure to give every department in the company access to AI tools, appropriate for their roles and responsibilities. Currently every employee has access to a shared AI layer, connected to internal data sources through MCP integrations, gated by individual OAuth credentials. This gives visibility into most systems, the ability to explore the data, and the ability to take actions on a user’s behalf.

Ecentria’s internal AI environment has access to a meaningful portion of the company’s operational data. What it does not have is full visibility into the supply chain layer.

For a business that works with a large supplier network, that gap is consequential. This means the accuracy of what OpticsPlanet shows as available and in what timeframe, and the reliability of what actually arrives, depends heavily on the quality of data flowing from suppliers in real time: inventory feeds, advance ship notices, invoices, and the transaction compliance that keeps all of it synchronized.

That data existed. It just lived outside the AI environment the company had built.

“I can’t easily see what has been invoiced and is in transit but hasn’t been received yet. That data exists, but I’m not going to see it cleanly in our systems until that product gets received. So, there’s always this hole of visibility that exists in the supply chain.”
– Alex Royzen, chief supply chain officer, Ecentria

Royzen had built workarounds. But workarounds require time, and the manual effort of pulling data from multiple interfaces to answer a single supply chain question was exactly the kind of friction the company was trying to eliminate everywhere else.

What AI Revealed About Ecentria’s Supply Chain

When Royzen’s team got access to SPS MAX, connected to their supply chain transaction data, the first thing he did was run audits he had never been able to run before.

The most revealing was inventory feed frequency. Suppliers send inventory data to communicate what they actually have in stock. For retailers, the freshness of that signal determines whether a customer gets what they ordered and how quickly. OpticsPlanet’s standard is at least one feed per day. The actual state of compliance across the supplier network, before this audit, was simply unknown.

Checking it manually would have meant reviewing transaction timestamps one supplier at a time. Instead, Royzen asked a single question and got a table back. Of the 35 suppliers actively sending feeds in the prior week, four or five had fallen below the acceptable threshold. In each of those cases, an automation had silently failed on the supplier’s side and no one had noticed.

“There is no easy way for me to get to that data. I would have to go one by one and look at every date of the transaction in SPS Fulfillment to see how often it was being sent. It was able to just summarize it for me and then I could have our team reach out to address it.”
– Alex Royzen

He followed up with each supplier directly, with specifics. The downstream effect was immediate: fresher inventory data, fewer orders sent against stock that no longer existed, less customer friction on the back end.

A similar audit on PO change acknowledgements surfaced suppliers who had live EDI connections but were not operationalizing the compliance workflows on their end. And a straightforward question about invoice totals for a specific supplier revealed that the supplier had stopped submitting invoices through the system entirely for two months. Product had shipped. Product had sold. The invoices simply were not there. The gap surfaced in seconds.

Royzen distributed MAX access to a small group beyond himself: finance leadership, warehouse personnel, and a few others across supply chain operations. He watched what they did with it.

What he did not expect was where it landed hardest.

In the warehouse, receiving discrepancies are a constant. A packing slip says five units. The box has four. Someone has to figure out whether the invoice matches, whether this is a short ship or a paperwork error, and what to do about it. The old path ran through accounting: call them, wait for them to pull the invoice from the supply chain system, compare it, and relay the answer back. Two teams, multiple steps, and both of them paused on a question neither could answer alone.

That whole chain collapsed into a single question. The warehouse team could ask directly, get the invoice, and resolve the discrepancy without involving anyone else.

Finance leadership had a similar experience. Invoice reconciliation across hundreds of suppliers had required navigating reporting interfaces to pull and format data that could then actually be worked with. Now they simply asked for it.

“In reality, it’s the people who don’t know how to utilize the reports who get the most out of it. They just want to ask and get a straightforward answer.”
– Alex Royzen

The response from both groups was immediate. Royzen described people messaging him to say they loved it. The value was not in sophistication. It was in access: giving people who needed supply chain data a way to get it without needing to become experts in the systems that held it.

Where the Company Is Going Next

Syrota’s vision for Ecentria’s AI infrastructure is not complicated, but it is ambitious. Every meaningful data source the company relies on should be available inside the company’s AI layer, connected through standardized integrations, accessible to anyone who needs it without a custom build project standing in the way.

Supply chain data is on that list. The current setup works. The next step is making it native: a direct connection between the SPS network and the AI infrastructure Ecentria already runs on, so a question about what is in transit, or which suppliers have gone quiet, or what the invoice record shows for a given PO, can be answered from the same place everything else is answered.

“If you marry a bot that knows its own intent with the data that you have externally, it’s superior. That would supercharge any company.”
– Slava Syrota, chief technology officer, Ecentria

That integration is in development. When it ships, supply chain data will stop being something Ecentria’s teams have to go find and start being something the AI already knows.

For a company that has spent years building toward that kind of infrastructure, the last gap is closing.