AI, Automation, and the New Reality of Business Operations
-
By
VireonPress Editorial
A couple of years ago, automation was a way to stand out. This year, it’s a way to stay afloat. Companies that ignore autonomous systems sooner or later find their products and/or services uncompetitive in terms of price and speed.
What Automation Actually Means Today
According to a PwC report for 2025 [1], companies that have transitioned to agent automation demonstrate margins 12-18% higher than those that remain with legacy solutions. This means that less efficient players will soon be simply acquired or displaced through price dumping, which only automated competitors can afford.
Orchestration instead of fragmentation
Old automation was piecemeal (usually implemented separately for CRM, warehouse management systems, accounting systems, etc.), leading to errors and delays when transferring data from one department to another.
Today, if your systems don’t interact without human intervention, you lose up to 20-30% of operational efficiency due to manual data reconciliation. Automation 2.0, aimed at workflow optimization, on the contrary, creates a unified data layer, so changes in one department instantly adjust plans in others, without requiring a single email or phone call.
Eliminating the human factor
Modern companies have grown so large in data volumes that humans are physically unable to detect anomalies or errors in real time. Companies relying on manual verification suffer losses from so-called bad data. Gartner estimates that the average cost of data errors for large companies will reach $12.9 million a year on average [2].
At the same time, self-correcting systems identify bad data and correct it autonomously, alerting your specialists only when a critical conflict happens.
Predictiveness
The traditional just-in-time purchasing model has become too fragile due to the increasing impact of external negative factors such as climate change, freight rates, and strikes. Therefore, reactive management leads to stockouts or, conversely, warehouses overloaded with unsaleable inventory.
While old methods only generate profits in a stable environment, which has ceased to exist in recent years, agent-based AI systems can analyze dozens of signals, enabling procurement weeks before warehouse shelves are empty.
Areas Where Businesses Automate First
In 2026, business automation trends are focused primarily on areas with the highest data density and the highest cost of error, where humans cannot physically compete with algorithms in processing speed.
Logistics and supply chains
In a world where climate change and geopolitical instability impact supply chains on a weekly basis, manual planning has become unprofitable. For example, the global shipping giant Maersk has implemented an autonomous chartering and dynamic routing system, where AI agents monitor port capacity, weather conditions, and fuel price fluctuations in real time, automatically adjusting vessel routes without human intervention.
As a result, Maersk has reduced vessel downtime by 15% and fuel savings by 7% [3].
Fintech and customer service
Customer service today implies comprehensive customer support, including resolving transactional issues. Specifically, Klarna replaced 700 full-time support staff with a single OpenAI-powered assistant, tasked with initiating refunds, recalculating credit limits, and resolving disputes with merchants.
This reduced the average resolution time from 11 minutes to 2, while revenue growth through automation reached $40 million per year [4].
Legal and banking compliance
Banks previously spent billions on manual legal due diligence, but AI automation has transformed this process into a matter of seconds. This is confirmed by the case of JPMorgan Chase, whose NLP-based COiN system automatically analyzes loan and legal agreements.
A year ago, this system was expanded with a real-time module for cross-border tax compliance. Thus, work that lawyers used to spend 360,000 hours on annually is now completed in seconds with 99.9% accuracy [5].
Retail and inventory management
Retailers shouldn’t just buy products based on customers’ purchases – they should anticipate customer needs. To achieve this proactive approach, Walmart decided to use AI in business operations to manage inventory in 4,700+ stores. Their systems consider 50+ factors and automatically place orders with suppliers, allowing them to reduce excess inventory by 12% [6].
Energy and industry
In heavy industry, automation is primarily aimed at preventing breakdowns before they occur. For example, Siemens’ proprietary platform, Senseye Predictive Maintenance, collects data from thousands of sensors in factories, predicting bearing or turbine failure weeks in advance. Thus, thanks to AI, Siemens has reduced unscheduled downtime of its equipment by 50% [7].
Where Automation Fails
Below, we’ve highlighted areas where automation risks causing more harm than good.
Chaotic workflows
You shouldn’t rely on AI to fix a chaotic business process – on the contrary, automating inefficiencies leads to their exponential growth. Zillow’s Offers project confirms this thesis and is considered a classic example of algorithm failure. The company used AI to automatically appraise and purchase homes, but the algorithm failed to account for local defects and market anomalies, continuing to buy properties at inflated prices even as the market began to cool.
As a result, the company lost $420 million, laid off 25% of its staff, and completely shut down the project, simply because its operating model lacked safeguards against market hallucinations [8].
The luxury segment
For brands built on scarcity and prestige, AI algorithms can destroy capital in a matter of hours. Take the case of Farfetch and its dynamic pricing: in an attempt to optimize sales in 2024-2025, the platform began experimenting with algorithmic discounts to compete with mass-market retailers. The algorithm automatically lowered prices for luxury brands (such as Chanel or Hermès) to stimulate turnover.
Brands began leaving the platform, and VIP clients felt their purchases’ exclusivity was devalued. All this led to the company’s collapse, so it was urgently acquired by the South Korean giant Coupang [9].
HR and recruitment
When automation affects humans, errors turn into legal and ethical lawsuits. This is illustrated by the case of Amazon’s AI Recruitment Tool, where automation reinforced human errors.
Trained on 10 years of resumes dominated by male applicants, the AI began automatically disqualifying women from technical positions. The project was shut down when it became clear that the algorithm was penalizing the inclusion of the word “women” in achievement descriptions (for example, captain of the women’s national team). This resulted in significant damage to the employers’ image and years of litigation [10].
Customer service
AI that lies or misleads customers costs a company its future, as demonstrated by the case of Air Canada, whose chatbot hallucinated in 2024 and promised a passenger a discount that didn’t exist in the official rules. The court ruled that the company was obligated to pay the promised discount, as the chatbot was an official representative of the company.
Ultimately, this precedent catalyzed widespread restrictions, and now, no major bank or airline can deploy AI agents without strict legal oversight [11].
Human Judgment vs Automated Systems
In recent years, a new management standard has emerged: human-in-the-loop 2.0, where there’s no need for human oversight. Instead, humans act as architects of meaning and final arbiters in areas of uncertainty. Below is an operating balance table to help you determine where automation should stop to avoid damaging your company’s image and profit.
|
Operational area
|
What automation does
|
What a human does
|
Why a human is irreplaceable
|
|---|---|---|---|
|
Ethics and risk management
|
Identifies anomalies and compares transactions with a database of sanctions and rules.
|
Makes decisions in areas where rules conflict with morality or long-term reputation.
|
AI has no conscience and can easily block a charity's account due to a formal error, causing a PR disaster.
|
|
Strategic development
|
Processes petabytes of data, creates trend charts, and models hundreds of market development scenarios.
|
Chooses a direction based on intuition and personal relationships with partners.
|
Experience shows that disruptive innovations often contradict historical data.
|
|
HR and corporate culture
|
Sorts resumes by keywords and analyzes productivity through biometrics and log files.
|
Evaluates engagement in work processes, as well as demonstrates empathy and leadership potential.
|
A machine sees a resource, while a leader sees a living person, something that cannot be digitized.
|
|
Crisis management
|
Instantly isolates a problem node (for example, shuts down a server during an attack or halts trading).
|
Conducts negotiations, interacts with shareholders, and takes public responsibility for errors.
|
In a crisis, people need a trusted leader, not a message from a bot.
|
Long-Term Operational Implications
By 2026, three fundamental approaches practiced by leading companies have emerged.
Hierarchical flattening by eliminating middle management
Traditional corporate hierarchies were built to transmit information from the bottom up (through reports) and from the top down (through orders), with middle management serving as a filter.
However, thanks to modern AI agents aggregating data in real time, this layer is becoming obsolete, as top management can now see the company’s situation without the distortions that humans inevitably bring. Ultimately, such decision automation has a positive impact on the speed of operational cycles.
Payroll transition to APIs
This year’s business model looks different on the balance sheet – the majority of OpEx is now shifting from “Salaries” to “Tech stack”. Companies are spending less and less on offices and insurance for thousands of clerks, instead investing millions in subscriptions to LLM models, API gateways, and computing power.
This enables cost-effective scalability, as now, to increase transaction volume tenfold, you don’t need to hire 1,000 accountants – you just have to scale server capacity.
Operational agility
Operational agility today implies a company’s ability to change its fundamental processes, supply chains, or even its business model in response to market signals. While previously, changing strategy required months of restructuring at all levels of the hierarchy, today it’s a matter of adjusting parameters in Agentic AI-powered workflows.
In general, before the advent of Agentic AI (approximately until 2024), even leading companies could only respond to crises within 2-4 weeks of their onset. However, with the widespread use of such solutions, anomaly detection has become possible within milliseconds. Ultimately, this means that even small players that have deployed this operational efficiency technology in time have gained the opportunity to win market share from corporations that are bogged down in bureaucratic approvals and still looking for productivity trade-offs.
Conclusion: Automation as Infrastructure, Not Innovation
Automation in business of 2026 is as basic an infrastructure as the internet, meaning if your company doesn’t use it, you simply won’t be able to keep up with market realities.
Start with a workflow audit to identify bottlenecks, and then implement automation layer by layer, from data to processes and autonomous decision-making.
And, of course, don’t forget to invest in staff training.
Sources:
- [1] The Fearless Future: 2025 Global AI Jobs Barometer
- [2] Data Quality: Best Practices for Accurate Insights
- [3] Minimizing variability in cargo transit time
- [4] Klarna AI assistant handles two-thirds of customer service chats in its first month
- [5] How JPMorgan Uses AI to Save 360,000 Legal Hours a Year
- [6] How Walmart is using A.I. to make shopping better for its millions of customers
- [7] Predictive maintenance for any industry
- [8] Zillow, facing big losses, quits flipping houses and will lay off a quarter of its staff
- [9] Farfetch Woes Worsened by Brands’ Distaste for Discounts
- [10] Insight – Amazon scraps secret AI recruiting tool that showed bias against women
- [11] Air Canada Has to Honor a Refund Policy Its Chatbot Made Up
VireonPress Editorial
VireonPress Editorial is the publication’s collective voice. We cover business, technology, culture, and beauty with a focus on trends, systems, and the ideas quietly shaping everyday life.
Discover with Vireon Press
Middle East Conflict Pushes Maersk to Reroute Again
Lockheed Martin and Raytheon Gain as Defense Spending Climbs
