
Nowadays we are subject to an enormous AI hype. Use AI agents – and they will do a great deal of your work just if you tell them so. This is to some extent true. If the work is web native, sure, you can ask AI agent to reserve you the tickets for your favorite club's next game.
But what if on other side is hard core business software such as SAP, with tight security rules and access restrictions? The problem is, at work you deal with exactly such solutions. You don't reserve concert tickets – you use ERP, CRM, HR software…
So, can you tell your ERP to check the level of an article on the stock? And order some more if the level is low? The answer is yes, but AI agents can't do it on their own. You will still need good old RPA to click and type into ERP front-end.

But how will the agent communicate with RPA?
Nowadays there is a standard which is getting more and more dominant – Model Context Protocol, or MCP for short. MCP is an open standard used in communication between AI and old fashion stuff such as APIs, files or databases. And yes, RPA too.
In other words, you can use MCP to get structured and exact instructions to RPA: check me the level of an article. If the level is below threshold, order some exact quantity.
We believe this is exactly how the future of work and modern workplace will look like. You will click the ERP forms less and less, manage your own digital worker more and more. In that way, you will be able to focus on running the business, not entering a myriad of transactions.
Tags
You might also be interested in
View all articles
BLOGMigrate ODI / DataStage to dbt
BLOG

