Map the invisible – The graph at a glance
Every modernization project seems to start the same way. Before anyone is allowed to change a thing, someone has to spend days, sometimes weeks, piecing together what actually depends on what. Which PI Vision display stops working if you retire a tag? Which AF attribute, Seeq workbook or TrendMiner view quietly reads a point you are about to rename? For most teams that knowledge lives in a few people’s heads, a handful of spreadsheets and a lot of careful trial and error. We built the Amitec Industrial Knowledge Graph, or IKG, to put an end to that.
One graph. Every system connected.
IKG takes your entire AVEVA PI and AF landscape and turns it into a single property graph that you can actually query. It scans your PI Data Archive, AF Server and PI Vision, and now Seeq, TrendMiner and our own Amitec reVision as well. It resolves the references between all of them and writes the complete picture into a graph database. Every node knows which system it came from. Every connection tells you what depends on what. When a reference cannot be resolved, it does not quietly disappear. It becomes a visible, explicit node, so you can find it and fix it instead of discovering it the hard way in production.
The questions that took days now take seconds
Once the graph is in place, the questions that used to take days take seconds. Before you retire a PI tag, you can see everything that depends on it, across AF, PI Vision and Seeq, in one view. You can rank every Vision display by how many broken references it carries and fix the worst offenders first. You can expose complexity hotspots and orphaned assets before a migration, rather than halfway through it. And you can produce real landscape documentation, every server, database, display, owner and dependency, ready for handover, an audit or a cybersecurity review, in hours instead of weeks.
What it looks like at real scale
To see how that looks at real scale, take one customer estate we mapped recently. In a single scan, IKG pulled together more than 43,000 PI System nodes, around 2,900 PI Vision nodes and over 320,000 Seeq nodes into one connected graph. It then classified every one of the 73,000-plus cross-system references by health, separating the genuinely broken links from the ones that are merely pending a source sync, and flagging the weak, name-only links that tend to fail silently later. Thousands of dependencies that nobody could previously see were suddenly listed, ranked and exportable. From there, asking “if I change this tag, what breaks?” returns the affected attributes, displays and Seeq analyses with a single click.
The demand has been hiding in plain sight
None of this is a guess about what the market wants. The demand has been sitting in plain sight on AVEVA’s own feedback forums for years. The single most requested item is AF to tag traceability, which has gathered 496 votes. It is administrators asking for a simple way to find which PI Points are actually used by AF attributes, without exporting everything through PI Builder, dropping it into Excel and stitching it back together with custom scripts. Close behind, with 257 votes, is a request for a PI Vision dependency map that lists every PI tag and AF attribute used on every display, so teams can find broken references and judge the impact of a change before they rename anything. AVEVA has marked that second request as not planned for PI Vision 2025, which tells you how long customers are likely to keep waiting for it.
Around those two headline requests sit several more open threads, and they all describe the same underlying gap from slightly different angles. One asks for proper change management visibility, a single place to see which displays, reports and analyses use a given tag or attribute, so a change can be made without quietly breaking something downstream. Another asks for analysis lineage, a way to know which PI Analyses already read from or write to a tag, so engineers stop creating duplicate or conflicting calculations on top of each other. A third asks for bulk find and replace across PI Vision displays when an instrument or a data source changes, instead of opening and editing each display by hand. Add up the votes across all of these threads and you are looking at hundreds of customers asking, in their own words, for the very same thing: a reliable way to see the hidden dependencies in their PI estate before they touch it. IKG is our answer to exactly that, built by a team that has lived inside the PI installed base for thirty years. And we are not stopping at the graph. We are already adding a Model Context Protocol layer on top, so an AI assistant can answer those same impact and dependency questions about your estate in plain language, securely and against your own live data.
