Business and technology leaders continue to use automation to augment and sometimes replace human labor. Enthusiasm has surged for AI agents in software and humanoid robots in the physical world. There is clear potential for both, but hype and hope have outpaced thoughtful deployment and sustainable returns on investment (ROI). In 2027, many of the more overenthusiastic early bets on automation will be rolled back, while physical AI’s promise to combine elements of hardware and cognitive automation will create new opportunities and challenges. Forrester’s automation and robotics predictions for 2027 point to three shifts:

  • Enterprises revert 20% of agentic workflows to deterministic processes to save money. In 2027, companies will admit that some of their early AI experiments haven’t delivered ROI. Infrastructure, documentation, and data automation (seen as important by 84% of automation decision-makers in Forrester’s Automation Survey, 2025) don’t always need AI to cut costs and boost productivity. To save money, enterprises will return some workloads to basic infrastructure and digital process automation platforms. This will curb rising AI costs while restoring more predictable workflows.
  • 10,000 humanoid robots will be doing useful work, with 75% in China, by the end of 2027. Humanoid robots commanded attention in 2026, sprinting, performing kung fu, and dancing their way through the year’s events calendar. Humanoid robots continue to advance, with some machines already doing useful work. But hype outpaces operational reality. Chinese firms could build as many as 100,000 humanoid robots in 2026 and far more in 2027, but most will end up in huge training labs rather than operational workplaces. There, robot fleets repeatedly perform basic tasks to train hardware, software, and AI models. Operational deployment of humanoid robots will grow in 2027, but it’s not (yet) at the scale the hype suggests. Don’t let visually impressive robotic sprints, martial arts, or dancing determine your physical AI strategy. Focus on deployments that create lasting value.
  • Twelve percent of firms move process intelligence from afterthought to core capability. Token cost overruns, inaccurate AI agent outcomes, and agent sprawl are pushing tech leaders to rein in initiatives. In 2026, process intelligence software vendors prepared for AI enablement, implementing broad support for agent grounding, monitoring, and emerging interoperability standards such as MCP. In 2027, process intelligence will become core to the agentic layer, enabling better agentic outcomes with higher conformance rates at lower cost. Tech leaders must stop treating process intelligence as retrospective business analytics. Instead, they should use it as an always-on guardrail for AI behavior, cost, and outcomes.

Forrester clients can read our full Predictions 2027: Automation And Robotics report to get more detail about each of these predictions, plus two more. Clients can also join a webinar on January 19 or set up a guidance session to discuss these predictions and the ways they might influence your plans for 2027.

If you aren’t yet a client, you can download one of our complimentary Predictions guides, which cover our top predictions for 2027 across a variety of themes. Get additional complimentary resources, including webinars, on the Predictions 2027 hub.

Share