Connect your AI agent to any machine.

Start free. Run it in production for a flat platform fee, or let our built-in agent do the work. 14 protocols, 18 manufacturers, one universal data layer, governed and attested.

Forge Connect, Free
$0
  • 3 machines, keyless sandbox
  • All 14 protocols
  • All 29 MCP tools
  • Your agent talks to equipment in 30 seconds
  • No credit card, no sales call
Forge Runtime, $999/mo
$999 / month
  • Production deployment, 25 machines included
  • $29 / machine after 25, platform included
  • All protocols, governed agent execution
  • Tamper-evident attestation
  • AWS Greengrass, Azure IoT Edge, or Docker
  • Runs on any 4 GB Linux host
Runtime Enterprise
Talk to us
  • Site license
  • On-prem deployment
  • SSO, audit exports, SLA, multi-site fleet management
  • AWS and Azure Marketplace available
Talk to us
Forge API, Pay Per Call
$0 / to start
  • For developers and AI agents. No subscription, no minimum.
  • Free: identify, get_coverage, correct_mapping, get_agent_card
  • $0.01 normalize, $0.02 anomaly, $0.05 predict / breach
  • $0.25 health index, OEE, energy, machine intelligence
  • $0.50 fleet health / OEE, $1.00 diagnose / shift report
  • Premium actions: work orders, exports, valuations, certificates
  • At ~1,000+ calls/mo, Runtime ($999 unlimited) saves money
Forge Monitoring, $49/machine/mo
$49 / machine / month
  • Our built-in intelligence agent included
  • Failure predictions and real-time OEE
  • Automatic shift reports, energy tracking
  • AI root-cause analysis, parts prediction
  • Intelligence without building your own agent

Runs on any Linux host with 4 GB RAM. No new hardware. Connects via MQTT, Modbus, OPC UA, SNMP, MTConnect, EtherNet/IP, PROFINET, BACnet, ROS2, J1939/CAN, MAVLink, REST, or CSV.

Pilot Rescue, $25,000 fixed
$25,000 fixed
  • Your industrial AI pilot is stalled on the data layer
  • We connect your agent to your equipment in one week
  • 14 protocols, normalized, governed, attested
  • Fee credited against Runtime Enterprise
Rescue my pilot →
Any MCP client
One line to connect
  • Works with Claude, Cursor, and custom agents
  • LangChain, CrewAI, AutoGen compatible
  • Listed on Smithery, Glama, and the MCP directory
claude mcp add --transport http forge https://mcp.foundrynet.io/mcp \
  --header 'Authorization: Bearer YOUR_API_KEY'
View documentation

Forge API, full rate card

For developers and AI agents making occasional queries rather than running a full deployment. No subscription, no minimum. Bring your own agent, pay only for what it calls. Standard tools are unmetered on Runtime and Enterprise; premium actions bill on every tier.

normalize_telemetry, query_machine_history$0.01
detect_anomalies$0.02
predict, predict_breach, remaining_life$0.05
predict_batch$0.10
health_index, calculate_oee, energy, machine_intelligence, verify_record$0.25
fleet_health, fleet_oee$0.50
diagnose_machine, shift_report$1.00
Premium: work order / dispatch / post-repair verify (all tiers)$3 / $1 / $5
Premium: compliance / condition cert / VHaaS export / fleet valuation$49 to $299
identify / get_coverage / correct_mapping / get_agent_cardFREE

Running a real deployment? Forge Runtime is the flat platform plan ($999/mo, unlimited standard calls), and Forge Monitoring is cheaper per machine past a handful of readings a day.

Does the built-in agent pay for itself?

One hour of unplanned CNC downtime costs $300 to $1,000 or more. Forge Monitoring on a 10-machine cell is $490/month. One caught failure per year pays for the decade.

$15,000
Annual downtime cost
$5,880
Annual Forge Monitoring cost
40%
Break-even prevention rate

Get a free Forge Connect key now

No email gate, no waiting. Your keyless sandbox key is issued instantly.