LYON -- November 19, 2026 -- Esker, the Lyon-headquartered finance automation vendor, unveiled its Synergy Agentic Framework on September 9, 2026. The framework is a layer of AI agents that execute work across its Source-to-Pay and Order-to-Cash products. The performance figures in the announcement, including more than 90% touchless invoice processing and over 40% lower DSO, were reported by customers of Esker's existing automation, not of the new framework.
Business Agents take over invoice exceptions, approvals and collections
Esker says Business Agents move repetitive execution out of human task queues. Per the company, they can identify invoice exceptions, route approvals, prioritize collections, assess credit risk, support cash application, resolve claims and help customer service teams answer inquiries. Humans keep judgment, policy and exceptions.
The customer-reported results are more than 90% touchless invoice processing, over 40% decreases in days sales outstanding (DSO), and more than 3x faster approvals, dispute resolution and order processing. A touchless rate above 90% means fewer than 1 in 10 invoices still needs manual handling, by our calculation. Esker says the framework is designed to extend this automation from individual tasks to broader sequences of governed work. It has not published results for the framework itself or said when it becomes available.
Esker opens its finance workflows to external AI agents through MCP and A2A
The framework rests on three capabilities: Act, Talk and Connect. Under Talk, Conversational Finance lets users state requests in natural language. The platform routes each request to the right workflow or Business Agent. It can also contact users when an approval, clarification or confirmation is needed, for example an urgent invoice approval or a goods receipt against a purchase order.
Under Connect, Esker exposes selected finance capabilities through APIs and the emerging Model Context Protocol (MCP) and Agent-to-Agent (A2A) standards. The targets are enterprise AI, large language models, ERP and CRM environments. External agents consume controlled capabilities rather than gain direct access to protected systems, the company says. For customers, an action started in another vendor's interface would still run, and be logged, inside Esker under its identity, permission and policy rules.
Esker anchors AI reasoning in rules-based execution and audit trails
Esker pairs probabilistic AI reasoning with deterministic execution. According to the company, workflows, business rules, permissions, validations and audit trails govern every approved action. Chief Product and Technology Officer Jean-Jacques Bérard said, "The value of an AI agent does not come from the model alone." He said it comes from combining intelligence with transaction context, business rules and governed execution.
Chief Marketing Officer and Head of Product Management Eric Bussy said finance leaders want solutions that coordinate work across functions while preserving control, rather than disconnected AI tools. Esker's Data Lake and Benchmark Intelligence feed the agents with KPIs, performance history and anonymized peer benchmarks, used to identify gaps and measure results.