Skip to main content
Import Intel

Company

We followed the data problem into the plant.

Import Intel began by centralizing fragmented commodity information. The deeper problem appeared inside the operations of the companies producing those commodities: embedded systems generated valuable data, but employees still had to reconcile machines, legacy software, spreadsheets, business systems, and market conditions by hand. We pivoted from a trading analytics destination to an agentic operating layer for commodity producers.

What changed

From market screens to machine centers

Centralized market data still matters — but only when it sits next to what the plant can actually make this week. The company that needs both is the one physically producing the commodity.

What we learned that shaped the platform

  • The data usually exists. The shared meaning does not.
  • Detected downtime without operator context does not improve anything.
  • Answers without source evidence get argued about instead of acted on.
  • The work after the answer — the report, the follow-up, the purchase request — is where the hours actually go.
  • Nobody in a plant wants software that can change things on its own.

Focus

Why embedded systems and commodity production

Base-commodity plants run PLC-heavy processes, keep decades of legacy data, name the same concept differently at every site, and make production decisions inside moving markets. That combination needs an operating layer, not another dashboard.

Built against real plant constraints

Optimizer data, machine downtime, grades, parts, and daily reports from working production environments — real constraints instead of assumptions. That work is the evidence for the architecture.

One pattern across materials

Lumber, aggregates, metals and recycling, and petrochemical plants share the same embedded systems, fragmented records, and operating questions. The platform is shared; the source mappings and ontology are site-specific.

Founders

Who is building it

Tejas Kalpathi, co-founder of Import Intel

Tejas Kalpathi

Computer Science at Texas A&M

Software engineering internships at Meta and Google. Works on the product and interface surfaces.

Arjun Madana, co-founder of Import Intel

Arjun Madana

Computer Engineering at Texas A&M

Runs the ML team for AggieSat. Works on the model, agent, and data layers.

Ian Schandlbauer, co-founder of Import Intel

Ian Schandlbauer

Finance & Business Honors at Texas A&M

Internships at AMD and Banco Santander. Works on the commercial model and market analysis.

[Founder photos and biographies pending brand review] — current photos and updated role descriptions for the pivot will replace this section once approved.

Talk to the people building it.

Operators, partners, and engineers who want to work on industrial agents are all welcome to write to us directly.

Book a plant workflow review

Contact