Industrial Knowledge Graph (IKG)

Mapping the invisible

Every PI tag, AF attribute, PI Vision display, Seeq workbook, TrendMiner view and reVision version, connected in one queryable graph

The Amitec Industrial Knowledge Graph (IKG) maps your entire AVEVA PI/AF landscape into a single, queryable graph, so you can see exactly what depends on what before you change anything. It scans your PI Data Archive, AF Server and PI Vision, and connects them with Seeq, TrendMiner and Amitec reVision, resolving every reference into one connected model. With pre-built impact questions, you can answer in seconds what used to take days: which displays break if you retire a tag, which AF attributes and analytics depend on a signal, and where broken or orphaned references are hiding. IKG turns tribal knowledge into a durable, auditable asset, giving your team the confidence to modernize, migrate and maintain without the guesswork.

What is it?

The Amitec Industrial Knowledge Graph (IKG) maps your entire AVEVA PI/AF landscape into a single, queryable property graph — so you can answer operational questions that would otherwise take hours of manual archaeology. It scans your PI Data Archive, AF Server, PI Vision, Seeq, TrendMiner and Amitec reVision, resolves every reference between them, and persists the complete topology into a graph database you can query with one click.

The result: the hidden dependencies that decide what breaks become visible before anyone touches a tag.

The problem: fragmented by default

Your metadata lives in many places, and none of them talk to each other. PI tags, AF structures, PI Vision displays, PI Analyses, Seeq workbooks, TrendMiner views and reVision display versions each describe a piece of the same plant — but the relationships between them live only in tribal knowledge.

Which tag feeds which display? Which AF attribute breaks if you rename a point? Which Seeq workbook or TrendMiner view depends on a signal you’re about to retire? Today, every modernization project starts the same way: weeks of manual tracing to find out what depends on what. Every change carries hidden risk.

IKG makes that hidden topology visible — and queryable.

How it works: Scan. Resolve. Persist. Query.

IKG ingests PI Points from the PI Data Archive via the AVEVA AF SDK, AF Elements and Attributes from the AF Server, displays and folders from PI Vision via REST, analytics from Seeq and TrendMiner, and display version history from Amitec reVision. Every node carries the system it came from; every edge tells you what depends on what. Unresolved references become first-class MissingRef nodes, so a broken link is visible, not silently dropped. The complete topology lands in a graph database, queryable through pre-built impact questions or open queries.

Key Features

  • Scan your whole landscape: Per-scan selection of PI servers, AF databases, Vision folders, Seeq and TrendMiner sources, and reVision repositories, with wildcard PI Point patterns. Reachability is auto-detected — any unavailable source is skipped gracefully.

  • Resolve every reference: Each af:// and pi:// path on every display, workbook and view is walked and matched against AF and PI inventories. Unresolved references become explicit MissingRef nodes.

  • Seeq & TrendMiner lineage: Workbooks, worksheets and views are linked to the exact PI tags and AF attributes they consume — so analytics dependencies are no longer invisible.

  • reVision version history in the graph: Amitec reVision display versions are ingested alongside the live topology, so you can see not just what a display depends on today, but how those dependencies changed over time.

  • Pre-built impact queries: Retirement risk, display health, orphaned tags, dependency chains, missing references — one click per question, no query language required.

  • Open queries when you need them: The graph browser is one click away. Analysts and power users can write their own queries against the same canonical model. The graph is yours.

  • Built-in diagnostics: Live connectivity status for every connected source and the graph database, plus a full inventory of known PI Data Archives and AF servers.

See it in action

A single demo scan produces 959 PI Points, 1000 AF Attributes, 18 Vision displays and zero missing references across nine AF databases in roughly 17 minutes — landing as 2,220 graph nodes and 3,272 relationships. From there, the Explore page answers questions like “If I retire PI tag COMP_001.SPEED, what breaks?” — surfacing the AF attribute that maps to it, the Vision displays that render it, and the Seeq and TrendMiner analytics that consume it, in seconds. Every result exports to CSV.

Pre-built impact questions

IKG ships with ready-to-run questions grouped into four categories — no query language required:

  • Health & Discovery: missing references, display health ranking, orphaned PI points, data sources per display.

  • Retirement Impact: what breaks if I retire a tag; where an AF attribute is used and which tag it maps to; which Seeq and TrendMiner analytics depend on it.

  • Dependency Tracing: what a display depends on, what references an AF database, what references an AF element (recursive).

  • Data Exploration: browse all objects by type.

Where the value lands — priority use cases

  • Retirement risk: Before renaming or retiring a PI tag, see every AF attribute, Vision display, Seeq workbook and TrendMiner view that depends on it. End the rename-and-pray pattern.

  • Display health: Rank every Vision display by broken references. Fix the worst offenders first; document what is deprecated and what is salvageable.

  • Migration readiness: Before a PI Vision modernization or AF restructure, expose complexity hotspots and orphaned assets. Cleanup first, migration second — cost and risk drop.

  • Documentation in hours: Auto-generated landscape documentation — every server, database, display, workbook, owner, timestamp and dependency — exported to CSV for handover, audit or cybersecurity review.

  • Engineer onboarding: New engineers navigate the plant visually instead of through tribal knowledge that survives staff change.

Why choose Amitec IKG?

The market has been asking for dependency intelligence for years. Public AVEVA feedback threads make the gap explicit — 496 votes for AF-to-tag traceability and 257 votes for a PI Vision dependency map, alongside several other open requests for lineage, change-management visibility and bulk replace. IKG is the response, built by a 30-year AVEVA partner that knows the PI installed base by heart — and uniquely able to connect your AVEVA estate to Seeq, TrendMiner and our own Amitec reVision in a single graph.

Benefits

  • Make changes with confidence instead of hidden risk.
  • Cut modernization discovery from weeks of manual tracing to a single scan.
  • See analytics dependencies across Seeq and TrendMiner, not just PI and AF.
  • Turn tribal knowledge into a durable, queryable asset.

  • Produce audit- and cybersecurity-ready documentation on demand.

  • Keep the graph current as the estate evolves.

What’s next: an MCP layer for natural-language impact analysis

We are extending IKG with a Model Context Protocol (MCP) layer on top of the graph — so AI assistants and copilots can answer impact and dependency questions about your estate in natural language, securely and against your own live data. Ask “what breaks if I retire this tag?” in plain language and get a grounded answer straight from your graph.

How we engage

  • Pilot — Two weeks, one PI site: a scoped scan and three answered impact questions on your real estate.
  • Platform — Annual subscription for the IKG application, core connectors and the pre-built query library.
  • Implementation — Fixed-scope discovery and source mapping for multi-site rollouts, plus governance baseline and training.
  • Managed service — Scheduled scans, graph health monitoring, governance reviews and SLA support.
  • Advisory — Migration planning, PI Vision modernization and graph-led architecture consulting.

Want to know more?

Start with a two-week pilot on your own PI estate. We scan one site, build the graph, run three impact questions against your real data, and hand you the results — no commitment beyond the pilot.