Most enterprise AI experiences don’t fail because they lack AI tools; they fail because those tools never leave the lab. Demos impress, pilots excite, but when it’s time to scale, workflows break down, governance is missing, and trust evaporates. That’s the AI paradox facing CIOs today: plenty of proof-of-concepts, but very little proof of value.
AI adoption in enterprises has been enthusiastic but not transformative. According to MIT’s "State of AI in Business 2025" report, 60% of organizations have evaluated AI tools, 20% reached pilot stage, but only 5% made it to production. The failure rate is staggering: brittle workflows, lack of contextual learning, and the inability to embed AI into day-to-day operations leave most initiatives stranded. This has created what MIT calls the GenAI Divide, a gap between companies that remain stuck in pilots and those who scale AI into real workflows.
Frends 6.1, the platform's latest launch — available from Sept. 2, 2025 to current and new customers —, addresses this divide head-on with its major new feature: the Intelligent AI Connector. Instead of treating AI as an isolated chatbot or demo, the connector allows organizations to embed reasoning models directly into automated workflows, governed by transparency, compliance, and measurable ROI.
“The benefits don’t come from pilots or ChatGPT. The benefits come when we integrate AI into business processes and actually get AI to do the thing — not just be a chat,” explains Asmo Urpilainen, Chief Product Officer at Frends.
The Intelligent AI Connector makes this integration practical. It introduces a new AI shape inside Frends’ BPMN 2.0 workflow engine. Developers and business teams can drag and drop the connector, define input and prompts, and let the model process information with context. Crucially, the connector produces not just an output, but also a reasoning log with a step-by-step explanation of why the AI arrived at its result.
This matters because enterprises are struggling with trust. As one MIT interviewee put it: “If it doesn’t plug into Salesforce or our internal systems, no one’s going to use it.” Another added: “I can’t risk client data mixing with someone else’s model.” The Intelligent AI Connector directly tackles both by ensuring seamless integration with existing systems and supporting on-premises or local AI deployment for maximum data control.
The Frends Intelligent AI Connector solves the practical blockers that keep AI stuck in the lab:
Asmo summarizes it best:
“What’s unique is not just that we call an AI model. Anyone can do that in a minute. What matters is all the orchestration and governance around it, making AI usable in production, not just in a demo.”
MIT’s research notes that while 90% of enterprises have seriously explored buying AI solutions, only 5% have achieved scaled adoption. The paradox is striking: large firms lead in pilot volume but lag behind smaller companies in scaling.
Meanwhile, the study found that some of the fastest ROI came not from sales and marketing — where 50% of AI budgets flow — but from back-office automation, where integrations delivered measurable cost savings. This aligns directly with Frends’ approach: embedding AI into integration workflows where repetitive, cognitive tasks eat up resources.
Frends 6.1 marks the first stage of its Agentic AI roadmap. The Intelligent AI Connector is designed to get organizations past pilots and into production-grade AI workflows. Later releases will introduce the AI Toolbox and advanced orchestration features, allowing even more autonomy while maintaining governance.
By positioning AI not as a sidecar but as a native building block of enterprise automation, Frends enables organizations to:
The message is clear: AI value is created when orchestration meets transparency.
AI initiatives fail when they stay siloed, brittle, or unexplainable. With Frends 6.1, enterprises get an integration-first approach where AI is not a gimmick, but a trusted part of the automation stack.
Or, as Asmo puts it:
“Customers don’t want magic. They want AI that integrates, scales, and explains itself. That’s exactly what we built into the Intelligent AI Connector.”