Poka AI Studio allows manufacturing teams to build and deploy governed AI agents without coding, while a new MCP server connects frontline data and workflows with enterprise assistants including ChatGPT, Microsoft Copilot and Claude.
14 October 2026 — Orlando, Florida — Poka Inc., an IFS company focused on Industrial AI for Connected Work, has launched Poka AI Studio, a no-code environment designed to help manufacturing teams create and deploy AI agents directly on the shop floor.
Announced at IFS Unleashed in Orlando, the platform is aimed at plant managers, continuous improvement teams and industrial operations staff who want to build AI agents without relying on data scientists or large consulting projects.
Poka says the system is built on 12 years of frontline workflow data and has been tested through an Early Access Program with manufacturers including Mars and Nestlé.
Alongside AI Studio, Poka has also introduced a Model Context Protocol (MCP) server that can make Poka content and operational data available to enterprise AI assistants such as Microsoft Copilot, ChatGPT, Claude and proprietary models.
Industrial AI agents without coding
Poka AI Studio gives operations teams a no-code environment in which they can build agents from scratch or start with predefined templates.
According to the company, customers can create an initial custom agent prototype within days and expand from there.
Use cases already available inside the platform include a Scheduled Agent that reviews open operational issues and escalates them to the appropriate people, as well as an agent that identifies abnormal values in forms and converts them into actionable insights.
The aim is to make AI automation more accessible to the teams responsible for running factories rather than limiting development to central IT or specialist AI functions.
Antoine Bisson, co-founder and CEO of Poka, said:
“AI Studio puts the power of AI creation directly into the hands of plant managers, continuous improvement leads, and operations teams, without requiring a team of data scientists or a costly consulting project.”
AI grounded in frontline procedures
A major part of Poka’s positioning is governance.
The company says agents built through AI Studio are grounded in actual frontline Standard Operating Procedures, allowing their responses and actions to reflect established plant processes.
Bisson said this is intended to give operators actionable answers while providing compliance teams with greater oversight.
“Every agent is grounded in real frontline Standard Operating Procedures and governed by default, so operators get actionable answers and compliance gets complete peace of mind.”
For manufacturers, that distinction is important because industrial environments often involve strict safety, quality and compliance requirements.
AI systems working inside those environments need to operate within defined processes rather than simply generating generic recommendations.
MCP connects factory data with enterprise AI assistants
Poka’s new MCP server extends the platform beyond its own AI environment.
The Model Context Protocol allows enterprise assistants to access Poka data and workflows while maintaining authentication, permissions and governance.
Poka gives the example of a plant manager asking an enterprise AI assistant:
“Which lines missed their changeover target this week, and why?”
The assistant could then retrieve the relevant information directly from Poka rather than requiring the manager to manually gather and analyse the data.
According to Poka, every request through the MCP server is governed by authentication and permission controls.
The company says the production-ready server can work with Microsoft Copilot, ChatGPT, Claude or proprietary models already used by the organisation.
Bringing frontline data into the wider AI stack
Manufacturers often operate with fragmented data across production systems, workforce tools, maintenance platforms and quality-management systems.
Poka’s approach is to make frontline operational data available to the wider enterprise AI environment without forcing organisations to replace the assistants they already use.
The MCP server effectively connects industrial workflows with a company’s broader AI stack.
That could allow operational data to be combined with other enterprise information while keeping access controls in place.
For companies already investing in general-purpose AI assistants, the approach creates a path for those tools to answer questions about what is happening on the factory floor.
Off-the-shelf agents for manufacturers not ready to build
Poka is also offering pre-built Poka AI Agents for organisations that do not yet want to create their own.
These agents are designed to work alongside operators, supervisors and quality engineers, as well as with the existing Poka AI Assistant.
According to the company, they can gather context, guide users through multi-step processes and take action around areas such as troubleshooting, quality and continuous improvement.
The company’s broader strategy is therefore split between configurable no-code agents and ready-made industrial use cases.
Tested with Mars and Nestlé
Poka says AI Studio and its MCP server have been validated at plant level through an Early Access Program involving manufacturers including Mars and Nestlé.
The company positions that testing as an important part of ensuring the tools work in production environments rather than remaining experimental AI prototypes.
Industrial AI is increasingly moving from standalone demonstrations into operational systems connected with production data, workforce processes and established plant procedures.
Poka’s launch reflects that shift by focusing on agents that can monitor, escalate and act within real manufacturing workflows.
From connected work to agentic operations
Poka’s existing platform focuses on Connected Work, helping manufacturers manage frontline knowledge, instructions and workflows.
AI Studio extends that model by allowing autonomous or semi-autonomous agents to operate on top of those processes.
The company’s long-term vision is to make frontline workers more augmented and autonomous through AI while maintaining governance around how the technology operates.
Rather than positioning agents as replacements for industrial workers, Poka is focusing on tasks such as issue escalation, abnormality detection, troubleshooting and process support.
That puts the emphasis on managing human processes with AI, rather than fully automating the factory floor.
About Poka
Poka is an Industrial AI platform for Connected Work and has been part of IFS since 2023.
Founded in 2014, the company says its platform supports more than one million frontline workers across 3,000+ sites in more than 70 countries and 17 languages.
Its customers include Nestlé, Bosch, Mars, Danone, L’Oréal, Tetra Pak, Coty, Hitachi and Flora Food Group.
Poka says its governed AI agents are built on 12 years of real-time frontline workflow data.
More information: https://www.poka.io/en/ai-studio/
