TL;DR

Basware has introduced AI agents to its Invoice Lifecycle Management platform, aiming to push accounts payable from basic automation toward autonomous operation. The agents handle dispute resolution and real-time processing queries while maintaining full audit trails.

Bridging the gap between automation and autonomy

Invoice management company Basware has launched two AI agents as part of its InvoiceAI solution, targeting what CEO Jason Kurtz calls “the AI delegation chasm” — the gap between experimenting with AI and actually trusting it to act on a company’s behalf.

A survey conducted by FT Longitude for Basware found that 61% of 200 finance leaders across the US, UK, France and Germany said their organisations had deployed AI agents largely as experiments. One in four admitted they still don’t fully understand what an AI agent looks like in practice.

How the agents work

The two new agents operate within Basware’s existing platform rather than as standalone tools:

AP Business Agent provides contextual, real-time guidance on actions when handling invoices, suggesting next steps to reduce friction in the approval process.

AP Data Agent responds to natural language queries — such as “show me all invoices awaiting approval in Germany” — delivering instant answers from invoice data to support faster decision-making.

Every agent action flows through a central policy engine with what Basware calls “autonomy gates” that apply business rules, compliance requirements and risk thresholds before any action is executed.

Billerud, a paper and packaging manufacturer and early adopter, reported that invoice quality improved considerably from day one, with efficiency gains translating into cost savings and return on investment within months.

Looking forward

Basware has outlined plans for additional agents through 2026, including a Supplier Agent that will automatically contact suppliers about disputes and summarise outcomes, and an AP Pro Agent for solving processing questions through natural language. The company says its approach prioritises explainability and governance, with every AI decision auditable through a single controlled execution path.