The trajectory points toward deeper AI and ML integration, enabling more sophisticated and autonomous decision-making. IBM frames this as building AI-first enterprises through a combination of hybrid cloud and nascent quantum capabilities, with Arvind Krishna fronting that vision. Read it as vendor direction-setting; the autonomy it promises is exactly the layer consumers are already pushing back against.
You already know which of your own tasks are repetitive enough to hand off and which are not. The cobot model assumes that line is worth preserving deliberately rather than erasing.
What Intelligent Automation Actually Integrates
Aligning IA solutions with existing systems is complex and can demand significant adjustments. Specialised expertise is often required to run and support the technology, which means investment in training or new hires. And employee buy-in is not automatic; managing the transition requires deliberate planning and communication.
RPA automation is no longer the frontier of enterprise technology; it is the baseline. Recent technology gains and the COVID-19 pandemic pushed many organizations into automation, and intelligent automation became the mechanism they reached for to streamline workflows. That much is settled.
So how do you reconcile a 50 to 60 percent efficiency gain with a customer base that says you have already gone too far?
The Numbers Vendors Lead With
The concrete example is a Volkswagen engine production plant in Germany, where cobots work with the production team on a strenuous stage of assembling the engine. That is the cleanest illustration in the ledger of semi automated work in practice: the machine takes the strain, the human keeps the judgment.
What is not settled is whether the people on the receiving end actually want it.
Each of those is a single-deployment result, not an industry average; read them as proof of ceiling, not proof of norm.
The stated design intent matters here, and it cuts against the usual industrial automation anxiety. Cobots are not built to replace human workers. They are built to support them.
There is a second framing worth noting, because it shifts one of the three pillars. In that version IA integrates AI, Business Process Management (BPM), and RPA. AI is the analytical core, running machine learning and algorithms across structured and unstructured data. BPM, also called business workflow automation, handles the optimisation of organisational workflows. RPA automation does the unglamorous part: software bots executing repetitive back-office tasks like data extraction and form completion.
Where RPA Automation Is Already Working
UiPath and its peers compete hard on document mining, and Forrester assessed the eight leading vendors in the document mining and analytics platform space. That field exists because IA can now automate tasks once requiring human involvement, including document understanding and unstructured data analysis. Take a business fielding roughly 1,000 customer comments a day across channels; that volume is the kind of unstructured load the category was built to absorb.
The interesting move is at the seam. When RPA is integrated with AI, the bots stop being purely deterministic and start managing more complex scenarios, using AI-derived insights for tasks that demand a higher level of cognitive function. That is the line between automated and semi automated work. A pure RPA bot follows a script. A semi automated process leans on AI judgment where the script runs out.
This is the unresolved tension, and it is the most important one in the piece.
The efficiency case is overwhelming. But while organizations get smarter, consumers are asking for more human interaction, and the survey data is blunt about it: 82 percent of US consumers want human interaction, and 59 percent of all consumers think companies have lost the human touch in delivering customer experience and meeting expectations.
88 percent of small and midsize businesses told Zapier that automation gives them a competitive edge over larger companies. That figure measures perceived competitive advantage among surveyed SMBs; it does not measure realized revenue or margin. The same report found 65 percent of knowledge workers experienced less stress because of automation, which is a self-reported affective measure rather than an output one.
Industrial Automation and the Cobot Question
Start with definitions, because vendors blur them. Intelligent Automation (IA) is the integration of AI, Machine Learning, and Robotic Process Automation applied to business processes, and the pitch is consistent across IT, logistics, and services: more efficiency, lower operational cost, better productivity for medium to large enterprises.
Operational case figures are more concrete. A leading healthcare provider used IA to manage patient data and schedule appointments, cutting administrative workload by 50 percent. A major bank applied IA to loan processing and compliance checks and accelerated approval times by 60 percent. In manufacturing, IA used to track machinery and forecast servicing requirements produced a 40 percent decrease in downtime.
RPA and robotic desktop automation (RDA) are increasingly adopted precisely for the mundane. And the desktop variant produces some of the sharpest gains on record.
The metrics are real, and they are loud.
How Businesses Balance Automation With the Human Touch
On the factory floor, the framing is collaborative rather than substitutive. Collaborative robots, or cobots, work alongside human workers to drive productivity in manufacturing. They handle assembly, packaging, and other repetitive or potentially unsafe tasks.
Sentiment at the executive level is no longer the obstacle. The obstacle is downstream.
The balance, then, is not a dial between human and machine. It is a sorting decision: route the recurring, the unstructured, and the unsafe to the bots, and reinvest the recovered human hours into the interaction customers say is missing. The 82 percent figure measures stated preference for human contact, not refusal of automated service; those are different things, and treating them as identical leads firms to automate the wrong layer. My read is that the consumer complaint is about where automation was applied, not whether it was applied at all.
The ledger points to a single coherent answer, and it runs through the cobot logic rather than the chatbot logic. Automation earns its keep where it removes the menial. In sales, intelligent automation reduces human error and lets workers focus on more complex assignments instead of repetitive ones. Research conducted by Amazon together with Lonergan Research determined that automation and AI had the potential to give Australian workers back 245 hours each year, framed as improving job satisfaction and work/life balance.
The applied surface is broad. Intelligent automation is used to automate software testing, to enhance business resiliency through cloud migration, and to lower marketing spend. In healthcare, AI paired with machine learning detects illnesses, supports diagnoses and clinical decisions, and improves primary care delivery.
Cloud automation sits adjacent to this, helping employees manage cloud-related tasks and services, with 59 percent of organizations planning to focus on cloud migration. That is an internal efficiency play. It does not touch the customer-facing tension directly, which is part of why it is an easier sell.
The Friction the Case Studies Skip
The technology will keep getting more capable. Whether enterprises point that capability at the back office or the customer relationship is the decision that 82 percent of US consumers have already weighed in on.
The deployment figures assume a clean install. Most are not.
One study used RDA to consolidate sales applications onto a single screen. In real estate, that integration cut the time to complete a sales process by as much as 80 percent and raised the rate of units booked by 26 percent. The first figure measures process duration; the second measures conversion. They are not the same lever, and a vendor citing the 80 percent will not always tell you the booking lift came alongside it.
Low-code and no-code tooling, built on AI-powered automation, now lets non-developers design websites, build apps, and wire up workflows. AI and ML translate data into identifiable patterns across image, speech, handwriting, and face recognition. Chatbots and bot automation have reduced the manual, recurring load on employees without compromising ROI.
Weigh those against the 30 percent cost reduction Gartner projected, and the calibration is straightforward. The savings are real but back-loaded behind integration cost, a skills gap, and change management. A firm that budgets for the tooling and not the transition will underperform the case studies, possibly badly.
By Gary Bernstein
