Poka lets factory teams build their own AI helpers without coding

Poka, a company that makes software for factory floor work, has launched Poka AI Studio. It lets plant teams build their own AI helpers, called agents, without writing code. Poka says teams can do it within weeks, without data scientists or a costly consulting project. Poka is part of IFS. Mars and Nestlé tested the tool at the plant level before the global launch.

What it does

An agent is a piece of software that does a task on its own, then hands off to people when needed. In AI Studio, teams can build one from scratch or start from a ready template. Poka says most customers launch a first prototype in just days, then expand from there.

Two examples come from the announcement. A Scheduled Agent reviews open issues, flags them and escalates them to the right people. Another agent spots abnormal values in forms and turns them into useful insights.

The connector to other AI tools

Poka also launched an MCP server. MCP stands for Model Context Protocol. In plain terms, it is a standard way for an AI assistant to pull live information from another system. This one links Poka data to Copilot, ChatGPT, Claude or a company's own AI model.

Poka gives an example. A plant manager asks the company's AI assistant which lines missed their changeover target this week, and why. The assistant answers with live data from Poka. Poka says each request comes with authentication, permission limits and governance, meaning checks on who can see what.

Who it is for

The main audience is plant managers, continuous improvement leads and operations teams. Poka says every agent is grounded in a plant's real standard operating procedures, the written steps for how work gets done. It also says agents are governed by default.

Companies that do not want to build anything can use off-the-shelf Poka AI Agents instead. These work alongside operators, supervisors and quality engineers, and with the Poka AI Assistant. They gather context, walk people through multi-step processes and take action on troubleshooting, quality and continuous improvement.

What is not known yet

• Pricing was not shared in the announcement.
• Availability by region was not spelled out.
• Results from Mars and Nestlé were not published.
• The speed and governance claims come from Poka itself.

Anyone running shifts or a plant who wants to try this should ask Poka for pricing and a demo, and test it on one narrow task first.