Agent Pricing Engine · Data Center Pods

Data center pods,
quoted by conversation.

Say “1 MW, two rows of 20 racks at 25 kW, N+1” — the engine sizes the racks, containment, power train and cooling together and prices the pod as one BOM. Every number comes from your rules — never from a guess.

What it sizes together

→kW per rack drives the rack PDUs and feeds — and everything upstream of them.
→UPS modules and battery runtime sized under the redundancy you choose — N, N+1 or 2N — with step-down transformers per side.
→Generators, switchgear and transfer gear sized to the pod’s load, with interrupting ratings checked.
→Cooling — CRAH, coolant distribution, rear-door heat exchangers — checked against the heat each rack actually sheds, so a 40 kW rack never ships with 25 kW of cooling.
→Containment, row busway and commissioning levels carried into the same BOM.
→Every size and every refusal shows its source — the rule, the rating, the catalog page.

Built for pod, e-house and power-skid builders quoting first-pass sizing on bids where estimating time is the scarcest resource.

The engine is a full MCP server — connect it to Claude and your team quotes data center pods in plain conversation, on a rule pack built from twenty years of configuring these products. Override the pack with your own rules, governed and audited — and you're onboarded.

Try it in Claude — right now.

Connect the engine as a custom connector (two steps, shown here), then paste any of these. They were run on the live engine; prices are illustrative until your price book loads.

“Quote a 1 MW pod: two rows of twenty racks at 25 kW, N+1, rear-door heat exchangers, 480 V utility, 65 kA, 18 °C chilled water, slab.”

What you’ll see: The full power train and cooling sized from one line: UPS, generators, transformers, switchgear, busway, rack PDUs and CDUs — each checked for capacity after redundancy.

“Price an 800 kW GPU pod: two rows of ten 40 kW GPU racks, direct-to-chip liquid cooling, 2N, 480 V, 65 kA, warm-water 32 °C loop.”

What you’ll see: 2N doubles the A and B paths (UPS, transformers, switchgear); CDUs carry the liquid load and CRAH units the air remainder. Every capacity is shown against its limit.

THEN ASK — same chat as the prompt above

“What changes if that GPU pod goes N+1 instead of 2N?”

What you’ll see: The engine re-sizes the whole train for the new redundancy and shows the price difference.

Data center pods,quoted by conversation.

What it sizes together

Try it in Claude — right now.

Skeptical? Send us a real RFQ.

Data center pods,
quoted by conversation.