The future is agentic – it’s a phrase ERP Today has heard time and time again in recent weeks, repeated almost like a mantra of sorts by vendor chiefs across the enterprise tech spectrum.
Microsoft is the latest business software name to get in the act, releasing its own AI agents last week in a move which some outlets framed as a response to Salesforce unveiling its agentic offering in September. I’d say this is slightly misleading, though, when factoring in three elements:
- Microsoft production cycles are not known to be knee-jerk in reactiveness.
- Microsoft has been ahead in the AI stakes with Copilot’s release and other tools.
- SAP, Boomi, Celonis, ServiceNow and more have also all decided that the future is agentic with their own AI agent releases. The market is wider than simply Microsoft and Salesforce.
To delve into the hype around AI agents warrants looking at case studies of agents being used by companies in the here and now. In other words, is the present agentic? At Microsoft’s AI Tour in London, Charles Lamanna, Microsoft’s Corporate VP, Business & Industry Copilot, gave an example of a large energy provider in the US whose IT service desk was “transformed” by using conversational experiences combined with AI agents working in the background. Imagine ChatGPT on top of an ERP layer where agents mimic human actions, automating complex business processes.
Lamanna also reported that Microsoft AI agents are seeing significant activity in customer engagement centers, alongside back office processes such as financial and operational.
“In areas such as procurement [for example] where if you don’t have the right inventory and assets at hand, you can’t make your products if you’re a manufacturing company,” he adds.
ERP Today also heard the story of British pet care company Pets at Home, an early adopter of Microsoft Copilot Studio. Using the platform, an AI agent was created to help Pets at Home’s retail fraud detection team investigate any fishy transactions.
“The power of it is on a different level – it’s a different level to RPA altogether,” attests William Hewish, CIO at Pets at Home, who explains the agent pulls information from various databases to make a report with rankings for the benefit of his human team, who then make a decision if a transaction merits dogged investigation. Hu-man’s best friend, one might say.
“Speed’s very important in this area of business and this has really sped it up – it blew our socks off as the speed that this comes through at is really quite powerful.”
Speed is the word, considering also it took 10 working days for Pets at Home to build the agent. As Bryan Goode, Microsoft Corporate Vice President for Business Applications and Platform, explains to me, ERP users can enter a prompt to make an agent, just like they would for an action using Copilot and ChatGPT (on the latter, Goode reminded ERP Today that OpenAI is a “super strategic partner” for Microsoft, in a fine counterbalance to recent discussions of the relationship between Redmond and ChatGPT’s owner company).
“It’s very obviously no-code, if you will; you can then configure it […] even extend it with Azure.
“We don’t want you to have to hire professional developers to build these – we want users to be able to get started and get value as quickly as possible,” Goode elaborates.
This ties in with what ERP Today heard at the recent Celosphere event from process mining leader Celonis. Mamta Lamba, SVP – Global Transformation and Process Excellence at PepsiCo, attested that no digital upskilling was needed for use of agentic Celonis tool AgentC, giving way to a vision of it being used by all 150,000 PepsiCo employees.
Other tech leaders such as Kyle Hill, CTO at SI company ANS, meanwhile attest: “These agents can drive organisational wide optimisation and automation, a seismic shift from the AI assistants that have traditionally been used to help with personal time saving and efficiency.”
“Leveraging AI tools enables professionals to focus on strategic and high value tasks, and empowers teams to perform more efficiently at a higher standard.”
At the end of the day, AI in ERP relies on user adoption and assistance, and if AI agents make it as easy as possible for implementation while augmenting current work processes, then it might not be too soon to say the future is agentic.
Additional reporting by Yoana Cholteeva.
