Inside most steel plants today, two teams run the same facility on two different maps. The OT team owns the PLCs, the SCADA screens, and the millisecond-level control loops that keep a caster or a rolling mill from tearing itself apart. The IT team owns the servers, the ERP, and the data lakes that finance and planning depend on. AI in steel needs both maps at once — sensor-level timing data from OT and production-context data from IT — and that is exactly the seam where most steel AI projects quietly stall. Book a demo to see how a converged data layer avoids that stall from day one.
OT/IT LEAD PLAYBOOK · STEEL PLANTS · 2026
OT and IT Have Run in Parallel for 30 Years — AI Is the First Thing That Needs Them to Merge
Converging OT and IT isn't a network diagram exercise. It's a cultural, technical, and security realignment that determines whether your steel plant's AI initiatives ever get past the pilot stage.
60-70%
Steel AI Pilots That Stall on Data Access Alone
3
Historically Separate Network Layers Per Plant
12-18 Mo
Typical Time to Mature Convergence Governance
WHY CONVERGENCE STALLS
The Real Barrier Isn't the Network — It's Who Owns the Decision
Every steel plant already has a network topology capable of moving data between the control floor and the data center. What's usually missing is agreement on who is accountable when that data moves — and what happens if a change on the IT side ever touches a control loop. OT engineers are trained to protect uptime and safety above everything else; IT teams are trained to patch, update, and iterate quickly. Neither instinct is wrong, but left unreconciled, they produce exactly the friction points below.
01
Change Control Mismatch
IT patch cycles run monthly; OT systems are validated once and left untouched for years to preserve safety certifications.
02
Conflicting Uptime Priorities
A five-minute IT maintenance window is routine; a five-minute OT interruption on a live caster is a production incident.
03
No Shared Data Vocabulary
OT tags reference equipment IDs and sensor codes; IT systems reference orders, batches, and cost centers — with no mapping layer between them.
04
Security Ownership Gaps
Firewalls between zones often exist on paper only, with neither team clearly accountable for monitoring the boundary in practice.
SIDE-BY-SIDE COMPARISON
Where OT and IT Priorities Actually Diverge — and Where AI Needs Both
Convergence doesn't mean making OT behave like IT, or the reverse. It means building a layer where each side's priorities are respected while AI gets a clean, continuous feed from both. The table below is the starting point most plant leadership teams use to frame the conversation between department heads before any architecture gets drawn.
| Dimension |
OT Priority |
IT Priority |
What AI Actually Needs |
| Update Cycle | Change only when necessary | Continuous patching and updates | Stable read-only data taps, no live control changes |
| Latency | Millisecond-level determinism | Seconds to minutes acceptable | Near-real-time for alerts, historized for trends |
| Access Model | Physically isolated, air-gapped where possible | Centralized, cloud-connected | One-way historian replication into a shared layer |
| Primary Metric | Safety and uptime | Data availability and integration speed | Trustworthy, timestamped, contextualized data |
iFactory Sits Between Your Historian and Your ERP — Not in the Middle of Either
Our platform reads from your existing OT historian and IT systems without inserting itself into a live control loop, giving your AI models a converged data feed while your change-control processes stay exactly as they are.
THE CONVERGENCE ROADMAP
Four Stages Steel Plants Move Through on the Way to Real Convergence
Convergence is rarely a single project — it's a sequence of trust-building steps between two departments that have historically had little reason to coordinate closely. Plants that succeed tend to move through the same four stages, in the same order, regardless of how large the facility is.
Stage 1 — Joint Data Inventory
OT and IT jointly catalog every data source across both domains, identifying overlaps and gaps before any architecture decisions are made.
Stage 2 — One-Way Data Bridge
A read-only replication layer moves historian data outward to a shared platform, with zero write-back into control systems.
Stage 3 — Shared Governance Charter
A joint OT/IT committee formally documents who approves what change, closing the accountability gaps that stall most projects.
Stage 4 — AI Models on Converged Data
With trust and data flow established, AI models for quality, yield, and predictive maintenance can finally draw on both domains at once.
SECURITY BY DESIGN
Convergence Without a Widened Attack Surface
The single biggest objection OT leaders raise to convergence is security — and it's a legitimate one. A converged architecture done correctly should reduce risk, not increase it, by making every data flow explicit, monitored, and one-directional wherever possible.
Unidirectional Data Diodes
Physical or logical one-way gateways ensure data can leave the OT network for analytics without any path back in.
Segmented DMZ Architecture
A demilitarized zone between OT and IT holds the shared platform, so neither network directly touches the other.
Joint Incident Response
OT and IT security teams run tabletop exercises together, so a real incident doesn't surface a coordination gap for the first time.
Role-Based Data Access
Every user and every model gets scoped, auditable access to only the data layers relevant to their role or function.
FREQUENTLY ASKED QUESTIONS
Questions OT/IT Leads Ask Before Starting a Convergence Project
Does OT/IT convergence mean giving IT direct access to our control systems?
No. A properly designed convergence architecture never gives IT direct write access to PLCs, SCADA, or DCS systems. Data flows outward from the OT historian through a one-way gateway into a shared analytics layer, while control-system access remains exactly as restricted as it is today.
Book a demo to see this one-way architecture mapped against your current network diagram.
How long does a typical convergence project take from data inventory to a working AI use case?
Most plants complete a joint data inventory and initial data bridge within eight to twelve weeks, with governance charters and first AI use cases following over the next two to four months. The timeline depends heavily on how many historians, DCS platforms, and ERP instances are already in place.
Contact our support team for a scoping conversation specific to your plant's systems.
Who should own the converged data platform once it's built — OT, IT, or a new team?
Most successful convergence programs establish a joint governance committee rather than assigning full ownership to either side. OT retains authority over anything touching control systems, IT retains authority over infrastructure and integration, and the committee jointly approves changes to the shared data layer itself.
Book a demo to review governance models other steel plants have adopted.
Does convergence require replacing our existing SCADA or historian software?
No. Convergence architectures are designed to read from your existing historian and SCADA systems as they are, without requiring a rip-and-replace of infrastructure your OT team has already validated and trusts. The goal is to add a data bridge, not to disrupt systems that are working reliably today.
Contact our support team to confirm compatibility with your specific historian platform.
What's the biggest reason convergence projects fail even after the technical architecture is in place?
The most common failure point is skipping the governance stage — building the data bridge without first establishing joint accountability for changes, incidents, and access decisions. Technically sound architectures still stall when neither team feels ownership over the shared layer.
Book a demo to see how the governance charter stage is structured before any technical rollout begins.
Start With the Data Inventory, Not the Network Diagram
iFactory helps OT and IT leads run the joint data inventory that every successful convergence project starts with, then builds the one-way bridge your AI initiatives can actually run on.
