Suplari Canvas is an agent-built visualization surface inside the Suplari Procurement Intelligence Platform. A user describes the view they want in plain language and the Suplari Agent builds it, or they copy a prebuilt element from an example gallery, or assemble one in a visual editor. Every canvas reads the spend, supplier and contract data Suplari has already unified and classified from the customer's source systems, stays live as that data changes, and is shared by link with security that filters what each viewer sees. Suplari Canvas is in preview now.
Suplari Canvas turns a conversation into a connected operational surface, agent-built visualizations that filter and link across your entire spend, suppliers, and contracts. Not a report you run. A view that thinks with you.
How Suplari Canvas builds a view from a prompt
The demo below shows a canvas being created from a single request, then changed by asking for one more element. What it reads is the customer's own unified spend, supplier and contract data.
A request such as "add another chart showing spend by category" produces a new saved version of the canvas with that element in it. The agent reads and writes the canvas definition directly, so the element becomes part of the canvas and survives the session. The next person who opens that link sees it.
What Suplari Canvas does
The chart was never the hard part
Most procurement teams are not short of charts. Somebody builds the view everyone works from, it is right for about a month, then the data moves and it gets rebuilt, usually by whoever is next in line behind the one person who knows the reporting tool. The version history on that file tells the story on its own.
Underneath that loop sits a problem no visualization tool solves. Suppliers that do not match across systems. Categories nobody maintains. A number everyone quietly double-checks in a side spreadsheet before it goes in front of a CFO. A confident, well-designed chart sits on top of all of that and never flinches, which is what makes it expensive.
Both halves have to be fixed together. A view anyone can build is worth little if the data underneath it is still reconciled by hand, and clean data is worth little if changing the view takes three weeks and a specialist. Suplari's AI Readiness 2026 benchmark, based on 121 procurement teams, put 10.6 hours per procurement professional per week into work that could be automated, most of it assembling data rather than interpreting it.
What separates this from a business intelligence tool
General business intelligence platforms are built to visualize any dataset, which means the procurement work happens before the tool is opened. Someone models the data, reconciles suppliers, maintains the taxonomy and rebuilds the model when a source system changes. That work is what creates the specialist bottleneck, and it survives every change of visualization tool. We wrote about the trade-off in more depth in spend analytics vs business intelligence.
Suplari Canvas starts after that work. The spend, supplier and contract model is already unified and classified by the Suplari AI Data Platform and kept correct by Suplari Data Assistant as source systems change. What Canvas adds is that the person with the question can build and change the view, in the same place the data already lives.
What separates this from general-purpose AI
A general-purpose model can take a spend export and draw a chart from it. What comes back is a one-off: an answer inside a session, with no classification, no filtering, no access control, and no view of the years of supplier and contract history sitting in the systems behind that export.
Canvas writes into a governed platform. The version is saved, the filter domain is shared, the drill-through resolves to real records, and the same security model that governs the underlying data governs the view. The difference shows up on the second question, which is also the argument for a procurement semantic layer underneath any AI that touches spend data.
Where it fits in the Suplari platform
Suplari Canvas joins the Suplari platform capabilities alongside the AI Data Platform, the AI Operating System and the Suplari agents, including Worker, Assistant and AI Studio.
The division of labor is straightforward. The AI Data Platform is the unified, supplier-centric foundation procurement data lives in. Suplari Data Assistant puts data there and keeps it correct as source systems change. Canvas is where a person asks that data a question and keeps the answer.
Why we are shipping this now
Two things had to be true before this was worth building. The data model had to be reliable enough that a view assembled in thirty seconds was not a fast way to be wrong, which is the work AI-ready procurement data describes. And the agent had to be good enough to write a working definition from an ordinary sentence, with no prompt engineering in front of it.
Both are now true in production environments, so the constraint moved. It is no longer what procurement can see. It is how long it takes to see something new, and who has to be free to make it happen.
Availability
Suplari Canvas is in preview with a selected group of customers. Customers interested in the preview cohort can contact their Suplari account team. Canvas is part of the roadmap story at the Suplari Executive Customer Roundtable.
Frequently asked questions
What is an AI-generated procurement dashboard?
It is a view of procurement data that an AI agent assembles from a request written in plain language, rather than one a person builds by hand in a reporting tool. In Suplari Canvas the agent writes the elements into a saved canvas that stays live, filters as one, and drills through to supplier, contract and transaction records.
Do you need to know a query language or a BI tool to build a canvas?
No. A canvas can be built by describing it to the Suplari Agent, by copying a prebuilt element from the example gallery, or by assembling elements in the visual editor. The three routes produce the same kind of canvas, and none of them requires a query language.
What can a canvas show?
Ten element types are available on the same governed data: bar, table, waterfall, donut, map, bubble, stacked bubble, tree map, multi-stacked bar and flow. Each one reads the spend, supplier and contract model Suplari has unified from the customer's source systems, so two elements on one canvas cannot disagree with each other.
Can a canvas be shared outside the team that built it?
Yes, by link. Domain-based security filters what each viewer sees, so a category manager and a CPO can open the same canvas and see the data each is entitled to. Role-based access control, SSO via SAML 2.0 and OIDC, and MFA through the customer's identity provider apply to every canvas.
How is this different from asking ChatGPT to chart a spend export?
A general-purpose model works on the file it is given and returns a one-off answer in a session. Canvas works on classified, unified data inside the platform, saves each change as a restorable version, applies the customer's access model to every view, and links each element back to the records behind it.
