Toloka welcomes new investors Bezos Expeditions and Mikhail Parakhin in strategic funding round

Toloka welcomes new investors Bezos Expeditions and Mikhail Parakhin in strategic funding round

Toloka welcomes new investors Bezos Expeditions and Mikhail Parakhin in strategic funding round

AI Agents in action: 20+ real-world business applications across industries

June 6, 2025

June 6, 2025

Essential ML Guide

Essential ML Guide

AI Agents in action: 20+ real-world business applications across industries

From backend automation to real-time decision support, intelligent agents are reshaping enterprise operations—handling complex tasks, coordinating workflows, and quietly transforming the future of work.

The real rise of AI agents

AI agents have moved beyond lab demos and speculative prototypes. Today, they operate inside production systems, power operational dashboards, and coordinate decisions across industrial networks. These autonomous systems, often powered by machine learning, are now embedded across the enterprise—dispatching service fleets, optimizing supply chains, managing planting schedules, flagging code compliance issues, and triaging HR tickets.

Siemens Industrial Copilot, developed with Microsoft and launched in 2023, is evolving into a multi-agent system for industrial automation.

Siemens Industrial Copilot, developed with Microsoft and launched in 2023, is evolving into a multi-agent system for industrial automation. The latest AI agents—unveiled at Automate 2025—go beyond query-based support, autonomously managing tasks across the production lifecycle. Source: Siemens

AI Agent market: from niche to necessary

The global AI agent market is growing fast. Grand View Research estimates its size at $5.40 billion in 2024 and $7.60 billion in 2025, growing at a CAGR of 45.8% through 2030.

Grand View Research predicts the global AI agents market to reach USD 50.31 billion by 2030.

Grand View Research predicts the global AI agents market to reach USD 50.31 billion by 2030. Source: Grand View Research

This surge reflects more than just technological enthusiasm. Enterprise deployments are moving from pilot programs to production systems, driven by the operational advantages of agentic automation, where AI autonomy delivers business value across functions like customer service, product ops, HR, logistics, manufacturing, and beyond.

Still, most adoption remains clustered in a few leading segments. As of late 2023, Customer Service and Virtual Assistants held over 34% of the market, thanks to their maturity and rapid ROI. However, other agents in supply chains, product management, legal ops, and cybersecurity are catching up fast, especially as orchestration layers and autonomy scales improve.

Satya Nadella at Build 2025, May 19.

Satya Nadella at Build 2025, May 19. Source: Microsoft

Microsoft’s latest Build software development conference declared the “era of AI agents” officially underway. This era involves embedding autonomous copilots directly into operating systems and productivity platforms. In addition to Office 365 integrations, Microsoft is rolling out AI agents within Windows itself, enabling real-time task management, system-level actions, and cross-application coordination.

Google’s Gemini 1.5 Pro, while not an agent on its own, is a foundational model for building them, offering long-context processing, function calling, and robust multimodal reasoning. At I/O 2025, Google introduced Agent Mode. This major leap transforms Gemini from a reactive assistant into a proactive agent capable of managing tasks, planning workflows, and operating across apps like Chrome, Gmail, and Search.

What makes a true AI agent?

Not every system labeled an “agent” meets the bar. An actual AI agent isn’t just a chatbot interface or a script behind a button—it’s a system capable of autonomous, goal-driven behavior. At a minimum, it must be able to:

  • Perceive its environment (through data, sensors, or inputs),

  • Make decisions based on internal goals or constraints,

  • Take action—not just respond to queries, but initiate workflows or operations,

  • And often, adapt over time through feedback, context shifts, or learning.

Many copilots based on large language models assist with completing tasks. Still, they are not autonomous agents unless they can act independently, chain steps, use tools, or orchestrate multiple systems without relying on human agents for every instruction. 

Likewise, scripted bots that respond to conditions with fixed rules may be automated, but they lack the ability to reason and plan like a modern artificial intelligence agent.

General workflow of an AI agent. Typically, AI agents consist of three components: perception, a brain using Large Language Models for reasoning and planning, and action.

General workflow of an AI agent. Typically, AI agents consist of three components: perception, a brain using Large Language Models for reasoning and planning, and action. Source: AI Agents Under Threat: A Survey of Key Security Challenges and Future Pathways

Some agents operate in physical environments (like robotic harvesters) while others perform digital tasks (like legal research assistants). What unites them is how AI agents work—performing tasks with intent, within parameters, but with minimal human oversight.

How AI agents work across industries

As of 2023, customer management systems and virtual assistants held the largest share of the global AI agent market—34.85%, valued at $1.277 billion. This growth is driven by demand for 24/7 support, faster resolution times, and reduced staffing costs. Other high-uptake segments included Sales and Marketing ($891M) and Human Resources ($434M).

AI agents applications, 2020-2024 (USD Billion).

AI agents applications, 2020-2024 (USD Billion). Source: Market.us

These numbers highlight sectors where agentic systems are already embedded into business operations, but adoption is rapidly expanding. Emerging deployments are now surfacing in Healthcare, Cybersecurity, Supply Chain, and Legal and Compliance, where agents support diagnostics, flag threats, analyze risks, and navigate regulatory frameworks. While their market share remains smaller, these areas are positioned for accelerated growth.

What follows is a cross-sector snapshot of real-world agent technology—from simple reflex agents to complex strategic planning systems—illustrating how different types of AI agents are evolving from task automation to autonomous decision-making.

1. AI models in customer engagement and support

Salesforce: conversational agents for enterprise service resolution

Unlike traditional chatbots, Einstein Service Agent independently manages complex tasks such as processing returns, scheduling deliveries, and providing personalized product recommendations, using natural language processing to understand and react to customer queries. 

Salesforce showcased the agent’s capabilities using simulated conversations with the “Pacifica AI Assistant” during its 2024 launch. Source: Salesforce

OpenTable now uses the system to manage reservations and loyalty programs across 60,000 restaurants. Educational publisher Wiley reported a 40% increase in case resolution during peak periods, significantly easing the load on human agents.

Einstein Service Agent is part of Salesforce’s broader Agentforce strategy—an initiative to embed autonomous AI agents across customer service, sales, and other enterprise workflows.

Trengo: Workflow agents for multichannel customer support

Trengo, a Dutch SaaS platform, introduced AI HelpMate as part of its customer service automation suite. Designed to manage routine customer inquiries across messaging channels autonomously, HelpMate has been deployed by Valk Digital to support 43 Van der Valk hotels in the Netherlands in 2024. 

During its beta rollout, over 80% of guest requests were handled within two minutes, using AI agents and without human intervention, freeing up staff for higher-value interactions and enabling 24/7 support. The agent processes bookings, answers FAQs, and assists with loyalty program queries.

2. AI Agent examples in sales and e-commerce: goal-based autonomy at scale

Capgemini: service-oriented agents for order and inventory management

Capgemini, a technology consultancy, is building AI agents in partnership with Google Cloud to automate complex enterprise workflows. In May 2023, the firms launched a Generative AI Center of Excellence (CoE) with the goal of delivering 500 industry-specific use cases in two years, targeting sectors like financial services, insurance, retail, and automotive.

Building on this, in April 2025, Capgemini introduced a strategic initiative explicitly focused on agentic AI to turn customer experience into a measurable business driver. These agent types are designed to autonomously manage customer interactions, order processing, and inventory control, taking care of a significant share of sales teams' well-defined tasks.

Amazon Alexa+: personalized agents for voice-driven retail

Amazon’s upgraded voice assistant, Alexa+, functions as an AI agent by autonomously managing retail-related tasks. It can reorder groceries from Whole Foods and Amazon Fresh without explicit prompts, suggest personalized deals based on user preferences and past behavior, and proactively notify users of upcoming sales through features like Advanced Deal Alerts.

With integrated generative AI and diverse capabilities to retain context from past interactions, Alexa+ increasingly handles complex, goal-driven tasks. With collected data processing, it manages subscriptions and booking services, without human intervention, positioning it as a persistent, context-aware retail agent.

3. Intelligent agents for complex tasks in manufacturing

Tesla: sensor-guided agents for assembly line automation

Tesla has long relied on robotics to streamline vehicle production, deploying AI-driven machines for welding, painting, and tasks as intricate as wire harness assembly. These robots adapt to real-time conditions on the factory floor, with multiple AI agents coordinating movements and adjustments to optimize efficiency—a core trait of agentic systems.

The company’s ambitions have expanded with Optimus, a humanoid robot first revealed in 2021 and significantly upgraded by 2023. Optimus now boasts improved mobility and finger-level dexterity—delicately handling fragile items like eggs and autonomously sorting battery cells, taking over repetitive tasks once performed by human workers. 

While early demos showed limited remote supervision, Tesla aims to evolve Optimus into a fully autonomous AI agent capable of making informed decisions and operating without human oversight

While early demos showed limited remote supervision, Tesla aims to evolve Optimus into a fully autonomous AI agent capable of making informed decisions and operating without human oversight. Source: Tesla

Foxconn: Simulation-Driven Agents for Smart Manufacturing

Foxconn, the world’s largest electronics manufacturer, is actively integrating specialized AI agents to enhance manufacturing efficiency, quality control, and operational decision-making. Central to this strategy is FoxBrain, an internal model launched in March 2025. Built on Meta’s Llama 3.1 architecture and trained on 120 Nvidia H100 GPUs, FoxBrain functions as an agent by interfacing with production systems to recommend and execute actions autonomously.

In parallel, Foxconn collaborates with NVIDIA to develop digital twins of its factories using the Omniverse platform. These digital replicas enable AI agents to simulate entire production lines, train autonomous robots, and perform quality assurance testing virtually before real-world deployment.

Foxconn’s Ingrasys subsidiary, at its Taoyuan NanChing factory, achieved a 73% increase in production efficiency and a 97% reduction in product defects by integrating AI-driven solutions across core production processes in 2021.

Foxconn’s Ingrasys subsidiary, at its Taoyuan NanChing factory, achieved a 73% increase in production efficiency and a 97% reduction in product defects by integrating AI-driven solutions across core production processes in 2021. Source: Ingrasys

4. Complex Workflows in Human Resources

Deloitte: Rule-Based Agents for HR Service Delivery

Deloitte has integrated ServiceNow’s HR Agent Workspace into its operations to streamline and partially automate key HR-related business processes. As part of its “Digital Briefcase” initiative in the UK, the firm developed an onboarding application that reduced the average onboarding time by three hours per hire and eliminated over 100,000 printed documents annually. 

The HR Agent Workspace routes leave requests, processes expense approvals, and responds to employee inquiries through a virtual agent capable of conversational handling. Integrated with platforms like Workday and built using ServiceNow’s App Engine, these AI agents continuously adapt to policy rules and context, automating repetitive tasks such as approvals or document routing. 

5. Hierarchical Agents in Healthcare

Stanford Health Care: Ambient Agents for Clinical Documentation

Stanford Health Care has deployed Nuance’s DAX Copilot enterprise-wide to automate clinical documentation using AI agents. These systems listen during doctor-patient conversations, extract medically relevant content, and generate structured summaries for physicians to review. 

Some of the DAX Copilot’s features.

Some of the DAX Copilot’s features. Source: Image Management

Rather than relying on form-based automation, DAX operates hierarchically: capturing unstructured audio, interpreting medical context, and drafting preliminary notes. In a 2024 peer-reviewed study, Stanford clinicians reported reduced administrative burden, lower cognitive load, and improved documentation quality during a three-month rollout.

Amazon One Medical: Multi-layered Agents for Provider Support

Amazon One Medical's AI agents offer assistance with note-taking and record management. These agents, such as HealthScribe, transcribe clinical encounters, summarize patient histories, and directly extract structured data (e.g., medications, symptoms) into EHR systems. 

The framework reflects a hierarchical model: AI agents handle transcription, classification agents identify relevant entities, and orchestration agents manage output formatting. A tiered agent architecture helps providers focus on care rather than paperwork and other routine tasks.

6. Model-Based Reflex Agents in Financial Services

J.P. Morgan: Conversational Agents for Intelligent Banking

J.P. Morgan employs Kasisto’s KAI platform to deploy AI agents that help clients manage personal finances, handle account queries, and flag suspicious activity. These context-aware agents serve as autonomous banking assistants, utilizing natural language processing to understand financial questions and adapt to evolving user behavior across various digital channels.

7. Autonomous AI Agents in Cybersecurity

Enterprise Security Teams: AI Agents for Real-Time Threat Response

Around 10,000 companies deploy Darktrace’s AI agents to monitor and autonomously respond to cyber threats. These agents isolate endpoints, disable compromised accounts, launch containment protocols, and search for interpretable insights without human intervention—acting as embedded security co-pilots or even adding full-time security analysts to their customers’ teams.

8. Goal-Based Agents in Product Management

ThriveAI: Analytical Agents for Product Team Augmentation

ThriveAI, a startup that recently raised $1.8 million in pre-seed funding, is developing AI tools designed to act as junior product managers. These AI agents integrated with external systems handle tasks like sprint planning, user research summarization, and roadmap recommendations, aiming to support product teams by automating routine workflows and providing data-driven insights. 

9. Utility-Driven Agents in Supply Chain and Logistics

LeewayHertz: Autonomous Agents for Inventory Flow

San Francisco-based startup LeewayHertz specializes in building AI agents that autonomously monitor warehouse stock, forecast demand, and trigger restocking decisions. Built on platforms like CrewAI and AutoGen Studio, these agents reduce manual oversight and accelerate supply chain responsiveness by dynamically adjusting procurement in real time.

Fetch.ai: AI Agents for Freight Contracting

UK-based Web3 lab Fetch.ai builds custom AI agents that autonomously represent buyers, carriers, and brokers within logistics networks. These agents negotiate freight rates, secure transport capacity, and coordinate real-time delivery schedules, executing decisions independently to streamline operations and cut brokerage overhead.

10. Model-Based AI Agents in Legal and Compliance

IBM: AI Agents for Regulatory Compliance in Financial Services

IBM has developed intelligent agents that assist financial institutions in navigating complex regulatory environments. Integrated within the IBM watsonx.governance platform, these AI agents analyze regulatory changes, ensuring that internal policies remain aligned with external mandates. 

11. Sensor-Guided Agents in Agriculture

John Deere: Autonomous Field Agents for Precision Farming

John Deere has developed fully autonomous tractors using AI agents that combine computer vision, GPS guidance, and sensors to perform essential farming tasks without an operator in the cab. 

Autonomous 9RX tractor presented by John Deere in 2025.

Autonomous 9RX tractor presented by John Deere in 2025. Source: John Deere

These AI agents enable the tractors to navigate fields, plant seeds, and harvest crops precisely, adjusting operations in real time based on soil data and environmental conditions. This innovation addresses labor shortages and enhances efficiency in large-scale agriculture with its highly specific tasks. 

Prospera Technologies: Vision-Based Agents for Crop Monitoring

Prospera Technologies employs AI agents that integrate data from high-resolution satellite and drone imagery to monitor crop health. These AI models analyze visual and sensor data to detect plant health issues, pest infestations, and diseases early, allowing farmers to take timely interventions. This technology provides real-time insights into field conditions, optimizing resource use and improving yields. 

IBM: Strategic AI Agents for Agricultural Decision-Making

IBM's Watson Decision Platform for Agriculture utilizes AI agents that process vast amounts of data, including weather forecasts, soil conditions, and crop stress indicators, to provide actionable insights for farmers. These goal-based agents assist in making informed decisions regarding irrigation, planting, fertilization, and harvesting, thereby enhancing productivity and sustainability in farming practices.

12. Artificial Intelligence in Goal-Based Navigation

Waymo: AI Agents Powering Fully Autonomous Ride-Hailing

Waymo, Alphabet's autonomous driving subsidiary, operates a fully autonomous ride-hailing service called Waymo One in cities including Phoenix, San Francisco, Los Angeles, and Austin. Their self-driving vehicles are equipped with a sophisticated AI system composed of layered AI agents, which integrate advanced machine learning models to perceive the environment, predict the behavior of other road users, and make real-time driving decisions. 

How Waymo’s driving system works. Source: Waymo

These AI agents enable vehicles to navigate complex urban environments, handle edge-case scenarios, and even deal with customer queries without human intervention. As of early 2025, Waymo has surpassed 10 million paid rides and is on track to double that by year-end, reflecting the growing acceptance and scalability of AI-driven mobility solutions. 

13. Utility-Based Agents in Dynamic Pricing Systems

Uber: AI Agents for Real-Time Fare Optimization

Uber’s dynamic pricing engine functions as a reactive AI agent, continuously analyzing supply, demand, traffic, weather, and local events to set ride fares autonomously. Designed to optimize real-time market equilibrium, these utility-based agents adjust prices without human input, balancing availability, driver incentives, and profitability at scale.

14. Generative AI Agents in Media

Synthesia: Agents for Scalable Video Content Creation

London-based Synthesia has developed AI agents capable of autonomously generating video content. These agents can create videos featuring digital avatars that deliver scripted content in multiple languages, streamlining the production of corporate training materials, marketing videos, and more.

Synthesia’s AI-generated avatar compared to its human prototype. Source: New York Post

By automating the video creation process, Synthesia's AI agents allow companies to produce personalized and localized content at scale, reducing the need for traditional filming and editing resources. 

Final Thoughts: The Quiet Takeover Has Begun

AI agents aren’t coming. They’re already here, not always with blinking lights or humanoid smiles. They run spreadsheets, drive tractors, flag compliance risks, and negotiate freight routes, automating complex tasks and reshaping existing business models. 

The agentic shift isn’t about a single app or sector—it’s about intelligence becoming infrastructure. We're entering a phase where enterprise software isn't just reactive, but proactive. Various types of AI agents offer more than assisting humans in performing tasks—they identify patterns, collaborate, decide, and execute.

Subscribe to Toloka News

Case studies, product news, and other articles straight to your inbox.

Subscribe to Toloka News

Case studies, product news, and other articles straight to your inbox.

Subscribe to Toloka News

Case studies, product news, and other articles straight to your inbox.

More about Toloka

What is Toloka’s mission?

Where is Toloka located?

What is Toloka’s key area of expertise?

How long has Toloka been in the AI market?

How does Toloka ensure the quality and accuracy of the data collected?

How does Toloka source and manage its experts and AI tutors?

What types of projects or tasks does Toloka typically handle?

What industries and use cases does Toloka focus on?

What is Toloka’s mission?

Where is Toloka located?

What is Toloka’s key area of expertise?

How long has Toloka been in the AI market?

How does Toloka ensure the quality and accuracy of the data collected?

How does Toloka source and manage its experts and AI tutors?

What types of projects or tasks does Toloka typically handle?

What industries and use cases does Toloka focus on?

What is Toloka’s mission?

Where is Toloka located?

What is Toloka’s key area of expertise?

How long has Toloka been in the AI market?

How does Toloka ensure the quality and accuracy of the data collected?

How does Toloka source and manage its experts and AI tutors?

What types of projects or tasks does Toloka typically handle?

What industries and use cases does Toloka focus on?