What the New AI Execution Layer Race Is About
The current wave of enterprise software M&A around AI workflow automation refers to large platforms acquiring specialist startups that build AI agents able to plan, take actions, and complete business tasks across multiple systems as an “AI execution layer” on top of existing ERP and CRM data. Rather than stop at recommendations or summaries, these agentic ERP systems aim to coordinate work between humans and AI, connect to core applications, and execute steps with appropriate controls and approvals. Vendors such as Asana, Coupa, Salesforce, and Vertice are not only adding generative features; they are buying proven execution engines, content layers, and domain-specific intelligence. This signals a shift from chat-style copilots toward operational AI teammates that sit inside workflows customers already trust, changing how projects, spend, and procurement are run end-to-end.
Asana’s StackAI Deal: From Tasks to Human–Agent Teams
Asana’s acquisition of StackAI for USD 75 million (approx. RM345 million) shows how work management vendors are turning project tools into human–agent operating systems. StackAI’s no-code AI workflow automation lets teams design, test, deploy, and govern agents that connect ERP, CRM, ITSM, and document tools like Salesforce, AWS, DocuSign, and Oracle. Asana plans to fuse this with its Work Graph, AI Studio, and AI Teammates so agents can pull project context, perform actions in connected systems, and write results back into Asana. CEO Dan Rogers says customers will be able to “agentify” complex processes, moving beyond simple request intake or task routing. For buyers, this marks a shift from bolt-on AI apps to native AI execution layers inside their existing work graph, reducing integration overhead while keeping governance and ownership in one place.
Coupa, Salesforce, and Vertice: Vertical AI Workflow Automation
Alongside Asana, other enterprise software M&A moves show how AI execution layers are becoming highly vertical. Coupa acquired intelligent document processing specialist Rossum to push AI deeper into source-to-pay workflows, using a transactional LLM trained on tens of millions of documents to automate complex invoicing and extend document intelligence across spend management. Salesforce agreed to acquire Contentful to give Agentforce a native, composable content layer, so agents can query, assemble, and deliver structured content without manual publishing. Vertice’s acquisition of Vendr combines more than USD 75 billion (approx. RM345 billion) in indirect spend data, over 2 million pricing points, and 250,000 negotiated contracts into a procurement intelligence engine. According to ERP Today, these deals give AI agents the data, workflow, content, and execution layers needed to complete real work, not only generate suggestions.
Why Vendors Are Buying, Not Building, Agentic ERP Systems
Enterprise vendors are choosing acquisition over greenfield development because AI execution layers demand deep domain models, proven integrations, and governance that take years to build. StackAI already spans major enterprise systems, Rossum is tuned to transactional documents, Contentful powers thousands of digital experiences, and Vendr brings pricing benchmarks and negotiation histories. Buying these platforms gives Asana, Coupa, Salesforce, and Vertice immediate access to agent-ready workflows and datasets while letting them focus on embedding controls, roles, and audit trails in their core products. The result is a new class of agentic ERP systems where AI teammates can handle intake, orchestrate steps, and complete actions under constraints. For customers, this lowers the risk of stitching together point AI tools and increases the chance that automation respects existing processes, approvals, and compliance requirements.
What This Consolidation Means for the Next ERP and CRM Cycle
The consolidation wave points to a future where ERP and CRM platforms are judged by the quality of their AI execution layer as much as by their data models. Agentforce, Asana’s human–agent OS vision, Coupa’s document intelligence, and Vertice’s negotiation agents all signal the same direction: systems that can plan, act, and learn inside defined guardrails. For buyers, the opportunity is to gain AI workflow automation that feels native and safe rather than experimental. The risk is tighter vendor lock-in as AI agents depend on proprietary execution stacks and datasets. Over time, success will likely favor platforms that expose their agents, workflows, and controls through open APIs and clear governance, letting enterprises mix and match capabilities while keeping a single system of record. Enterprise software M&A is now as much about owning AI execution as owning the underlying data.