Suplari

A named integration catalog on Suplari's own Data Integration page: SAP, Oracle and NetSuite ERPs, contract management platforms, AP systems, supplier networks, P-card providers, Power BI and Tableau, with custom APIs and flat-file sources supported for any other system.

Action in outside systems is asserted rather than documented: the AI Agents page has the Tail Spend Consolidation Worker triggering sourcing events or catalog updates with approval and the Contract Renewal Manager initiating workflows, and the Data Integration page has Application Agents embedding intelligence into the customer's ERP and procurement workflows, but the receiving system and the action mechanism are not named. The catalog documents the inbound data direction, and Suplari positions itself as an intelligence layer rather than a transactional system.

SourceSuplari, suplari.com/spend-intelligence/procurement-data-integration and /products/ai-agentsread 2026-09-08

Suplari Worker deploys agents that detect, then execute workflows with or without human approval.

The Procurement Intake Triage Worker analyzes incoming requests, routes them to the right process or team and enriches them; the Contract Renewal Manager recommends actions, prepares summaries and initiates workflows; Suplari Studio lets the customer build custom agents, workflows and automations without code from templates or their own logic; and Value Orchestration is a named product turning identified opportunities into coordinated execution.

Together these run multistep work across a catalog of cooperating workers under one named platform, though no BPMN style designer or runtime documentation is published, so orchestration is described only at product level.

SourceSuplari, suplari.com/products/ai-agentsread 2026-09-08

A maintained, correlated data foundation.

Suplari's Data Integration page documents a unified procurement intelligence foundation into which data from ERPs, P2P, contracts, AP, general ledger, cards, expense, catalogs and supplier networks is ingested continuously via API, harmonized into a unified schema where vendor codes become vendor identities and PO numbers link to contracts and invoices, enriched with supplier context, category and business rules, and kept as an intelligence layer the agents answer questions from and act on.

The foundation persists past any run, scales past a context window and stays queryable, and the AI Agents page has every insight grounded in that data with sources cited.

SourceSuplari, suplari.com/spend-intelligence/procurement-data-integration and /products/ai-agentsread 2026-09-08

The customer sets each agent's autonomy, with an approval gate on Suplari's own surface. The AI Agents FAQ states the customer controls the level of autonomy, some agents execute automatically while others require human approval before triggering workflows, and the customer sets the rules; the Tail Spend Consolidation Worker can trigger sourcing events or catalog updates with human approval; and on the Data Integration page low-confidence contract matches are flagged for review rather than applied. No screenshot or documentation of the approval queue itself is published.

SourceSuplari, suplari.com/products/ai-agents FAQ and /spend-intelligence/procurement-data-integration FAQread 2026-09-08

Access control is documented; a compliance attestation is not. Suplari's AI Agents FAQ states agents operate within a governed data environment with role based access controls, and the subprocessor list names Okta as the identity provider authenticating Suplari account holders, so RBAC is a named access model.

Customer SSO or SAML is not itself documented, only Suplari's own Okta layer. On compliance, the same FAQ says only "compliance with enterprise security requirements", which names no certificate, auditor, report or date. The site has no security or trust page, and no SOC 2 or ISO 27001 attestation is published.

SourceSuplari, suplari.com/products/ai-agents FAQ and /utility-pages/list-of-subprocessors, probe recordedread 2026-09-08

A per-action record from detection to outcome, with the reasoning behind each recommendation.

The AI Agents FAQ states Suplari tracks every agent action from detection through execution to realized outcome, so the customer can see which savings were identified, which risks were mitigated and which workflows were automated, with clear attribution to agent activity; the same FAQ says all agent actions are documented for audit purposes and names audit logging; and every recommendation cites its sources and explains its reasoning so the team can verify before acting.

That is a record of what the agent did and why rather than reporting on the customer's data, and the Value Orchestration product line carries the outcome side. No log or trace surface, retention period or export is documented, and the description sits in an FAQ rather than product documentation.

SourceSuplari, suplari.com/products/ai-agents FAQread 2026-09-08

Agents learn from corrections and adapt to the organization's logic over time, but that learning is absorbed into the product rather than kept as memory with a stated scope and lifetime, and no such memory is documented for the AI Operating System or Studio. The customizable thresholds, logic and definitions in Studio are configuration, and the unified procurement intelligence foundation is the application's data model; deleting it would delete the business record. Conversation state in the Assistant is not documented either way.

SourceSuplari, suplari.com/products/ai-agents FAQread 2026-09-08

The platform is hosted in the US, with nothing for the customer to choose. Suplari's subprocessor list names Amazon Web Services as hosting provider for US data centers, with Okta and Pendo also hosted in US data centers.

The Data Integration page confirms the platform pulls data out of the customer's systems via API, extract or direct database connection into Suplari's own unified foundation, so the intelligence layer is hosted by Suplari; data connected in place with no migration describes where the customer's source systems stay, not where Suplari runs. No region list, customer environment option (VPC, on-premises, customer tenant) or selection surface is published. That subprocessor list was last updated 29 September 2021, before the current platform, so the hosting fact may be stale, though nothing on the site contradicts it.

SourceSuplari, suplari.com/utility-pages/list-of-subprocessors and /spend-intelligence/procurement-data-integrationread 2026-09-08

A browsable catalog of ten named prebuilt agents, each with a description, on Suplari's AI Agents page: Spend Anomaly Detection Worker, Savings Opportunity Scout, Supplier Risk Monitor, Contract Renewal and Obligation Manager, Tail Spend Consolidation Worker, Supplier Intelligence Research Worker, Procurement Intake Triage Worker, Market and Tariff Impact Analyst, Executive Reporting and Narrative Builder, and Compliance and Policy Enforcement Worker.

The FAQ states pre-built agent templates deploy in days, and Studio builds new agents from "proven templates". Each agent does a distinct job and none depends on another.

SourceSuplari, suplari.com/products/ai-agentsread 2026-09-08

The agents wake on continuous monitoring rather than on request, with named event classes documented per agent: spend anomalies and variance from norms, supplier risk thresholds with alerts when they are close, contract renewals, expirations and obligation milestones, incoming procurement requests triaged and routed on arrival, and policy breaches checked continuously.

The Data Integration page has APIs pulling data continuously and Digital Workers monitoring data quality in real time. Event coverage is documented across the catalog. No schedule, webhook or inbound queue surface is separately documented, and no channel (email, Slack, Teams) is named.

SourceSuplari, suplari.com/products/ai-agents and /spend-intelligence/procurement-data-integrationread 2026-09-08

Suplari discloses no model provider, no customer or admin model selection and no vendor side routing. The subprocessor list, last updated 29 September 2021, predates the agent products and names no model provider, so the absence is a gap in Suplari's disclosure rather than a statement of a single provider.

SourceSuplari, suplari.com/utility-pages/list-of-subprocessors and /products/ai-agentsread 2026-09-08

Data can be exported via APIs, according to the Data Integration page FAQ, which also says Suplari connects to Power BI and Tableau and integrates with existing procurement applications. That export API lets an outside caller pull from the platform, but it is asserted with no developer portal, reference, endpoint list, SDK or MCP server behind it.

The custom APIs and flat file sources the same page supports are inbound data connections from the customer's systems, and Studio is the customer building inside the platform rather than calling it from outside. No developer surface for Suplari is published, and the API's scope and access path are not documented.

SourceSuplari, suplari.com/spend-intelligence/procurement-data-integration FAQread 2026-09-08

The agents' own performance is not measured in anything published: no evaluation harness, release gate or controlled post deployment experiment is documented for them, in Studio or the AI Operating System. Cited sources and reasoning on each recommendation gate an output for a human reviewer, which is oversight rather than testing.

The FAQ statement that custom agents built in Studio take weeks to configure and test describes the customer's implementation phase and documents no test surface, scored cases or gate. Closed-loop tracking of every action from detection to realized savings measures the customer's procurement outcome, not the agent, and low-confidence contract matches flagged for review are per-output gating.

SourceSuplari, suplari.com/products/ai-agents FAQ and /spend-intelligence/procurement-data-integration FAQread 2026-09-08

Other systems are reached through APIs, extracts and direct database connections, per the Data Integration page, and no browser, desktop or remote computer control is documented. The closest thing is the Supplier Intelligence Research Worker on the AI Agents page, which performs continuous supplier research using internal data and external signals with no mechanism documented. That is external monitoring feeding the intelligence layer rather than an agent operating a web or desktop interface at run time.

SourceSuplari, suplari.com/spend-intelligence/procurement-data-integration and /products/ai-agentsread 2026-09-08