40+ Agentic AI Use Cases with Real-life Examples in 2025

Autonomous generative AI agents execute complex tasks with little or no human supervision. Agentic AI differs from chatbots and co-pilots.

Unlike traditional AI, particularly generative AI, which often requires human intervention in complex workflows, agentic AI can autonomously navigate and optimize processes thanks to its decision-making capabilities and goal-directed behavior. AI agents can serve as:

AI agents as developers

1. Building applications

AI code editors like Cursor AI Editor, Windsurf Editor, and Replit can build and deploy apps (e.g. To-Do list app) by:

• Selecting the right tools for the job, like Flask for APIs or React for frontends.
• Generating code in the chosen language (e.g., Python, JavaScript) based on simple prompts.
• Automating workflows with integrations like GitHub Actions for testing and deployment.

Real-life example:

A developer used OpenAI’s Operator and Replit’s AI Agent to build an entire app in 90 minutes. Two agents autonomously exchanged credentials, and ran tests.

Real-life example:

Cursor’s agent mode Composer can generate a complete Tic Tac Toe game from a single prompt:“Generate an HTML, CSS, and JavaScript Tic Tac Toe game for 2 players.”

Cursor can code across multiple files, execute commands, and automatically determine what context it needs (no need to add files).

2. Build your own API with AI coding 

AI code editors automate API creation by transforming specifications into functional code. Here’s how the process typically works:

• Integrating the output into the developer’s environment for testing, version control, and deployment.
• Ingesting API specifications (e.g., OpenAPI/Swagger files).
• Generating backend code based on the documented endpoints, methods, and schemas.

Real-life example:

No-Code API workflows for AI Agents with n8n.

Here is a high-level abstraction of an API workflow:

3. Natural language code editing

Developers can select code snippets and issue plain English commands—such as:

“Double the size of the board. Make it green – like an Apple 2e.” (see real-life example below).

Real-life example:

Coding agents like Cursor identify the intent, modifies the relevant code across files, and applies the changes.

4. Building websites

AI website creators like v0 by Vercel, Bolt, Lovable, and CerebrasCoder can create complex platforms like e-learning websites, generating key pages such as:

• Homepages
• Course listing pages
• Personalized student dashboards

Similarly, AI code editors like Replit can build websites and leverage APIs.⁵

Real-life example:

How to build a website AI agent:

Here is the automation workflow:

5. Generating a CRM dashboard

Agents can generate the front-end interface, configure back-end logic, and set up database interactions.

Real-life example:

Roo Code uses DeepSeek model to autonomously build complete CRM dashboards.

6. Recursive coding and legacy code modernization

While basic coding is typically a task for LLMs, recursive coding workflows—where an agent iteratively improves, refactors, or extends code across multiple layers—are inherently agentic. These AI agents can autonomously rewrite large code blocks, apply configuration changes, and test outcomes in cycles until a goal is met.

Real-life examples:

GT Edge AI converts legacy COBOL code into modern Java.⁸

Persistent provides a multi-agent framework used to autonomously migrate COBOL code to Java, it works by:⁹

7. Code refactoring

By using recursive coding AI agents can continuously improve the design of the code without changing how it works, making it easier to understand and maintain.

Real-life example:

Tech startup developers created an agent refactoring your code in 25+ programming languages.¹⁰

8. Generating code suggestions in your IDE

Agents like GitHub Copilot provide real-time code suggestions and auto-completions and reduce the likelihood of syntax errors.

9. CI/CD pipeline monitoring and optimization

AI agents can autonomously manage infrastructure in cloud-native environments like Kubernetes. These DevOps agents can:

• Identify running ingress controllers or other workloads
• Identify pods running
• Interpret high-level commands such as “shut down the NGINX pod”

Real-life example:

With Claude acting as a DevOps agent, a user can simply ask: “Check if I have any pods running.”.

AI agents as SecOps assistants

10. Threat intelligence

Agents gather and correlates threat actor TTPs (tactics, techniques, procedures) from open-source and proprietary feeds, and integrates findings into detection workflows.

Real-life example:

Microsoft’s Security Copilot includes a specialized Threat Intelligence Briefing Agent that dynamically gathers, filters, and summarizes threat intelligence.

11. Detection & triage

These actions occur at the initial signal ingestion stage to reduce noise and organize alert data before deeper analysis.

• Alert deduplication: Removes identical or near-identical events from multiple sources to reduce noise.
• False positive suppression: Suppresses recurring benign alerts using past resolution patterns.
• Alert grouping / clustering: Clusters related alerts (same user, host, or tactic) into a single incident.

Real-life example:

Charlotte AI performs autonomous detection and triage by:

• Identifying malicious behavior based on known TTPs.
• Cross-referencing execution patterns and file paths.
• Providing a human-readable verdict explanation.

12. Contextual enrichment & threat attribution

Performed after initial triage, this step adds depth and context to alerts.

Automated attacker attribution systems ingest CTI feeds, extract behavioral and temporal features, and compute similarity scores across incidents. Clustering algorithms then map intrusions to known threat actors like (e.g., APT41, Mozi or Lazarus) based on pattern overlap.

Real-life example:

Google Chronicle + Mandiant + Gemini AI agents autonomously ingest telemetry and CTI feeds, enrich alerts with IOC context (e.g., IP reputation, malware hashes), and cross-reference behavioral patterns with known threat actor tactics from the MITRE ATT&CK framework.

In this agentic setup:

• Detection: An AI agent detects lateral movement using service accounts.
• Enrichment: It automatically enriches the alert with data from Chronicle logs, asset inventory, and Mandiant threat intelligence.
• Attribution: A second agent clusters this activity with historical intrusion chains attributed to a known group (e.g., APT41).

13. Escalation & handoff

• Alert translation / simplification: Converts technical alerts into analyst-friendly summaries.
• Auto-ticketing and analyst assignment: Creates tickets and routes incidents to the appropriate tier or team.

14. Proactive response actions

In SecOps, agents can isolate endpoints, disable accounts, or kill malicious processes., these systems can:

• Query logs automatically using natural language or contextual triggers
• Run automated playbooks for investigations and escalation
• Adapt workflows in real time based on new evidence or pivoted logic
• Generate infrastructure-as-code (e.g., Terraform, Pulumi) for remediation
• Drafts and modifies detection rules based on investigation insights

Real-life example:

Google developed the SOC Manager agent, which leverages multiple sub-agents to execute a structured Incident Response Plan for malware detection.¹⁷

Agentic project structure:

adk_runbooks/
├── multi-agent/              # Multi-agent system implementation
│   ├── manager/              # Manager agent and sub-agents
│   └── requirements.txt      # Python dependencies
├── rules-bank/               # Security runbooks and procedures
│   ├── personas/             # Agent behavior definitions
│   ├── run_books/            # Operational procedures
│   └── irps/                 # Incident Response Plans
├── source/                   # Sphinx documentation source
└── docs/                     # Generated documentation (GitHub Pages)

In the final step of the incident response plan (Step 5: Completion), the IOCs (Indicators of Compromise) were proactively blocked by the SOC Manager agent executing an automated containment runbook (see below).

Source:
Mandiant & Google Cloud Security¹⁸

15. Threat hunting

Agents in threat hunting can:

• Scan for anomalies continuously across identity, network, and cloud logs
• Automate repetitive hunts such as IOC lookups or pattern detection
• Flag unknown threats by comparing behavior against historical baselines and peer activity

Real-life example:

Researchers developed a MITRE ATT&CK Driven Threat Hunting Automated by Local LLM system, where AI agents collaborate to generate Sigma rules for threat detection.

In this example A user inputs a request (e.g., “Please generate a Sigma rule for hunting Kerberoasting”) through a web UI.

Agent 1 retrieves relevant detection methods from MITRE ATT&CK, while Agent 2 uses this input—along with Sigma documentation and environmental data—to generate context-aware Sigma rules using a language model.¹⁹

16. Automated software and application testing

AI testing agents can create and execute unit, integration, vulnerability, and performance tests without extensive manual intervention. However, building these AI systems is resource-intensive, as they require significant computational power.

Real-life example:

Pcloudy’s Copilot provides selenium test scripts, and finds available browsers to test on and execute the test cases.²⁰

Read more: Enterprise AI agents, AI agent builders, open-source AI agents.

AI agents as human-like gaming characters

AI agents can improve NPCs and other agentic processes in the game world by performing NPC behaviors, game playing & adaptability, and procedural content generation.

17. Agentic non-player characters (NPCs)

Fully autonomous AI agents in gaming provide human-like behavior and gameplay for non-player characters (NPCs).

Real-life example:

Researchers created a small virtual town populated with AI by building a sandbox setting similar to The Sims with 25 agents called “Stanford AI Village”.

In this village, users can observe and interact with agents as they share news, build relationships, and arrange group activities. ²¹

Here’s an overview of the key components and ideas behind these concepts:

• Behavioral scripts: Stanford AI Village agents use predefined scripts for actions and reactions. This can include basic behaviors like engaging positively with other NPCs. 
• Dynamic behavior: NPCs use AI to adjust their behavior based on the player’s actions, making the game world more responsive.
• Pathfinding: NPCs use algorithms to navigate the game world effectively, finding paths around obstacles and pursuing the player.

18 Game playing

AI agents can autonomously play video games or assist human players in achieving specific goals by leveraging:

• Pathfinding and navigation: Game agents explore game environments, using trial and error to find optimal strategies. For example, they learn navigation paths in complex mazes by continuously refining their decision-making process.
• Input emulation: Game agents can emulate inputs, such as keyboard presses, mouse movements, or gamepad controls. This allows them to interact with the game just as a human player would.
• Gameplay adaptability: Agents can analyze a player’s gameplay style and adapt the game world, interactions, and mission design accordingly. For example, a stealth player may encounter fewer direct combat challenges.

Real-life example:

Google DeepMind’s Scalable Instructable Multi-Agent (SIMA) can navigate and interact with gaming situations. SIMA can play games such as No Man’s Sky and Goat Simulator.

Source: Google²³

19. Procedural content generation

AI agents are highly capable of generating vast amounts of game content algorithmically, such as:

• Terrain and landscapes
• Combat scenarios
• Character designs
• Soundscapes and music

Real-life example:

No Man’s Sky, an adventure game, uses procedural generation to create entire planets with:

• Varied topographies (e.g., mountains, valleys, canyons).
• Unique biomes (e.g., deserts, forests, oceans).
• Dynamic weather patterns (e.g., storms, rainfall).²⁴

AI agents as content creators

AI agents can automate content creation, editing, and publishing. These AI agents can assist human writers, and generate content independently. Some applications of AI writing assistants include:

20. Automated narrative writing

AI agents can write a narrative, by outlining chapters, drafting content, and polishing prose chapters, drafting content, and polishing prose.

Real-life example:

In a GitHub AI agent project, 10 specialized AI agents worked autonomously to write a novel of 100,000 words (~300 pages) with zero human writing. Some examples from those 10 agents include:

• Specifications agent: Analyzes story requirements and maintains narrative consistency.
• Production agent: Generates content and implements creative changes.
• Management agent: Coordinates between agents and tracks creative flow.
• Duplication agent: Prevents content redundancy and duplications.

Here is a Livestream showing how agents create the novel:

21. Technical report or proposal writing

AI agents can autonomously draft technical reports including:

• Engineering reports
• Project proposals
• Research papers
• Product specifications

Real-life example:

ParagraphAI, an AI writing assistant, can write technical engineering reports by outlining the timeline, the budget, and the resources and personnel required.²⁶

22. Knowledge-based article generation

Agents can pull information from knowledge databases like Wikipedia, product manuals, or academic journals to create a comprehensive overview of a specific topic.

Real-life example:

Perplexity Pages turns gen AI search results into structured Wikipedia pages.²⁷

23. Dashboard and diagram generation

Agents can automatically generate UI/UX components, system diagrams, and flowcharts based on text prompts, streamlining the design process.

Real-life example:

FigJam AI uses text prompts to generate:

• Dashboards for meetings and team exercises
• Diagrams and flow charts
• Visual timelines and Gantt charts²⁸

AI agents as insurance assistants

Agents automate claims review, approval, and fraud detection, streamlining the entire claims processing lifecycle. For example, a large-scale insurer automates ~90% of individual automobile claims by integrating custom AI agents into their claims workflow. ²⁹

24. End-to-end claims processing

Once a claim is submitted, an agentic AI system can automatically extract relevant data from the submitted forms, verify the details against existing databases, and flag any inconsistencies or potential fraud signals.

Real-life example:

Microsoft’s Power Platform – automates an insurance claim form:

Source: YouTube³⁰

Here, Microsoft’s Power Platform:

• Captured claim form documents such as checkbox selections and texts.
• Uploaded claim form to SharePoint content management app.
• Extracted data from claim form such as claim ID, and checkbox values (e.g., claim type, policy number).
• Integrated with Dataverse for data storage to save the extracted data (including checkbox values) into Microsoft Dataverse.
• Processed workflows in forms, for example, used Power App agents to approve or deny claims.

25. Underwriting

Agentic AI can automate underwriting with specialized agents, including a risk evaluation agent for claim likelihood assessment, and a pricing agent for dynamic premium adjustments, etc.

Real-life example:

Akira AI‘s agents automate insurance underwriting and risk assessment through the multi-agent system, each specializing in a critical aspect of underwriting:

• Data collection agent: Gathers both structured and unstructured data.
• Risk evaluation agent: Analyzes historical data and applicant details to predict the likelihood of claims.
• Fraud detection agent: Detects anomalies or suspicious patterns in claims and applications.
• Pricing agent: Suggests a pricing based factors like market conditions, competitor pricing, and customer behavior.³¹

26. Personalized policy resolution

Agents can communicate the claim status and next steps to the policyholder, including any additional documentation required, approval/denial updates, or payout details.

AI agents as human resources (HR) assistants

HR operations often involve numerous repetitive and time-consuming tasks that AI agents can efficiently manage. Here are key examples of agentic AI in HR operations:

27. Resume screening

Agentic workflows automate the screening process, filter relevant skills, and automatically assign scores based on your pre-defined criteria.

Real-life example:

PepsiCo uses AI tools to rank candidates depending on how well they meet job requirements.³²

28. Interview scheduling

Agents can handle the scheduling of interviews, and coordinating between candidates and hiring managers to find optimal times.

Real-life example:

LinkedIn HR Assistant performs day-to-day tasks like synthesizing job descriptions, searching for candidates, and doing basic screening calls.³³

29. Payroll automation

Agents in payroll processing calculate salaries, process deductions, and handle tax withholdings. They integrate with human resources information system (HRIS) systems and accounting software to ensure accuracy and compliance with payroll standards.

Real-life example:

Akira AI’s multi-agent payroll system automates every aspect of the payroll cycle. The system uses several agents, including:

Source: Akira AI³⁴

Here, Akira AI’s multi-agent payroll system uses several agents, including:

• Data ingestion & preprocessing agent: Uses data extraction methods to clean data from HR, and finance systems.
• Payroll calculation agent: Automates payroll computations, including pay, deductions, and bonuses.
• Compliance reporting agent: Detects compliance issues using machine learning.
• Query resolution agent: Resolves payroll issues using RAG.

AI agents as customer service assistants

Traditional chatbots can answer basic questions—but they often hit a wall when it comes to actually helping the customer. Agentic customer service tools changes that by:

• Interpreting nuanced inquiries, even when questions are vague or complex
• Seamlessly operating across channels—chat, voice, email—while retrieving data from customer profiles, order history, and knowledge bases
• Retaining context from prior interactions, enabling more informed and personalized responses
• Engaging proactively, such as sending reminders for checkups or follow-ups before the customer initiates contact

30. Answering customer calls:

When a customer calls about an inquiry, AI agents process the call with natural language.

Real-life example:

Ada AI Agent answers customer calls:

Source: Ada³⁵

31. Responding to customer tickets

AI agents deliver context-specific responses or direct customers to the appropriate resources for further assistance.

32. Sending follow-up and informative SMS

After an interaction, agents can send SMS messages to follow up with customers.

33. FAQ generation and optimization

Agents:

• Auto-generates and updates FAQs from live customer interactions and ticket history.
• Integrate with crm, chat, and knowledge base tools to identify recurring queries.

AI agents as research assistants

AI agents as research assistants are used in various fields to assist with data analysis, literature review, hypothesis generation, and experimental design.

Real-life examples:

1. OpenAI’s Deep Research uses reasoning to synthesize large amounts of online information and complete multi-step research at a Ph.D level when running large searches using the o3 & DeepSeek large language models (LLMs).^{36 37}

In an experiment where researchers asked Deep Research to conduct a real-life project with a detailed prompt, Deep Research:

• Asked follow-up questions to clarify the scope and details of the research project before beginning.
• Synthesized findings from 22 sources based on academic and industry reports.³⁸

2. ChemicalQDevice’s clinical decision support (CDS) system was asked to execute agentic workflow for drug discovery. In this example ChemicalQDevice’s system:

• Analyzed vast amounts of clinical literature from datasets.
• Executed automated coding with tools like GitHub Actions.
• Used generative AI (e.g., GPT-4 or customized models), to generate hypotheses, suggest experimental designs, and write research reports.³⁹

3. End-to-end agentic workflow system, otto-SR, can leverage LLMs to conduct literature searches, applies inclusion/exclusion criteria, extracts structured data, and performs meta-analyses.⁴⁰

OpenAI’s Deep Research ChemicalQDevice’s system and otto-SR can be used in several agentic use cases given below:

34. Literature review and knowledge discovery

Autonomously searching academic databases, journals, and online research repositories (e.g., Google Scholar, PubMed) to gather relevant studies, papers, and articles related to specific research topics or hypotheses.

35. Hypothesis generation and testing

Proactively generating analysis hypotheses based on patterns in the data and testing them (work that analysts and business users typically do).

36. Data mining and analysis

Manipulating structured and unstructured data from several sources like research databases, social media, patents, or clinical trial results, providing insights into emerging trends.

37. Data visualization and presentation

Generating insightful visual representations of complex datasets.

AI agents as computer users

“Computer Use” aims to enable AI to interact with a computer like a person would. This gives the flexibility to perform digital tasks without using OS- or web-specific APIs. 

There are two approaches for AI agents to perform tasks like humans:

1. Performing OS graphical user interface (GUI) actions (Anthropic’s Claude, Microsoft’s OmniParser, etc).
2. Web browsing (OpenAI Operator, Kura AI, Runner H, HyperWriteAI, MultiOn Agent Q, etc).

Tool examples:

1. OpenAI’s new Operator tool allows A.I. agents to complete tasks across various websites, such as buying pet food.⁴¹
2. Kura AI and Runner H can complete tasks by using web browsers.^{42 43}
3. HyperWriteAI can fill out text forms, click buttons, and select menu selections to place an online order.⁴⁴
4. Microsoft’s OmniParser enhances agent understanding of visual interfaces for GUI automation..⁴⁵

38. Form filling & web automation

Agents navigate webpages, click fields, and fill out forms based on user prompts or structured data.

Real-life example:

39. Document generation and editing

Agents open files, make edits, rename, organize, and save documents across local or cloud environments.

Real-life example:

Anthropic’s Claude is asked to “Generate 25 rows of sample expenses, save them into a spreadsheet, and then open the spreadsheet”.

In this example, Claude:

• Opened the CSV file in the LibreOffice Calc app.
• Created a CSV file with 25 rows of sample expense data.
• Installed LibreOffice Calc spreadsheet.⁴⁷

40. Web research & structured data collection

Unlike basic automation scripts, deep web research agents interpret unstructured information across multiple pages and return insights in a structured format.

Real-life example:

OpenAI’s Deep Research, a new agentic capability within ChatGPT designed for multi-step, high-context web research plans, navigates, and synthesizes information across multiple sources to answer complex queries.

41. CLI operations & script execution

CLI-based coding agents like Aider that are designed for terminal-based development workflows run shell commands, install software, launch scripts, and interpret outputs in terminal interfaces.

Real-life example:

42. Multimodal app navigation & coordination

Unified GUI agents (e.g.,OpenAI Operator prototypes) can switch between applications—such as moving from web browsing to email—copy and paste data, detect changes in the interface, and respond dynamically.

Real-life example:

To test an order delivery use case, I provided a simple shopping request from Open Operator: Help me buy a boho-style throw pillow cover under $30.

Autonomous workflow agents (Long-horizon tasking)

Planning agents with memory + tool use (e.g., Auto-GPT, Agent Q with goal-setting) can execute multi-step goals across varied tools (e.g., gather input, take actions, revise plan), making decisions in real time.

Real-life example:

In thus multi-step financial report analysis the AI planning agent is given the task: “Analyze last quarter’s financial performance and prepare a summary for the finance team.”

The agent is asked to get:

• Recent company news (top 3 stories)
• Current stock price of the company (AAPL)
• Company information (e.g., profile, market cap, sector)
• Historical stock prices (1-day interval over the past month)
• Analyst recommendations (e.g., buy/hold/sell ratings)

Here is the financial report output:

Real-life example:

MultiOn Agent Q booking a flight.

Source: MultiOn⁵²

Building AI agents

The shift from basic agent applications, such as natural language weather searches using tools like LangChain, to more complex, autonomous use cases like app development (e.g., generating a Tic Tac Toe game with the Cursor AI editor) resulted in the challenges:

• Managing complexity in task execution
  • Standardizing patterns that guide their decision-making process.
  • Incorporating multiple options and selecting the optimal action to execute a task. (e.g., “Should I display this screen or another based on user input?”).
• Interfacing with external systems and APIs

43. Generating custom AI agents:

AI agent-building frameworks help address these challenges by providing:

• Decision-making & reasoning frameworks (e.g., machine learning or probabilistic models) to standardize patterns & incorporate multiple options.
• Data access & integration tools (e.g., API connectors) to interface with external systems and APIs.

Real-life example:

Creating a custom Slovenia trip guide agent with Microsoft 365:

Source: Lisa Crosbie⁵³

External Links

• 1. Lamar on X: "this is wild 🤯🤯🤯 just paired openai operator with replit agent to build an app watch as two ai agents team up, exchange credentials, and start testing ai agent 🤝 ai agent the future is here and it’s insane! https://t.co/jIZnqW4fJD" / X.
• 2. Cursor vs GitHub Copilot: Which AI Coding Assistant is better?.
• 3. APIs for AI Agents: The Only Beginner’s Guide You’ll Ever Need (n8n) - YouTube.
• 4. Replit’s AI agent built my entire website (I just watched) - YouTube.
• 5. Replit’s AI agent built my entire website (I just watched) - YouTube.
• 6. How to Build a Website AI Agent in 13 Min (Free N8N Template) - YouTube.
• 7. Deepseek-R1 + RooCode: BEST AI Coding Agent! Develop a Full-stack App Without Writing ANY Code! - YouTube.
• 8. Cobol to Java Conversion – gtedge.ai.
• 9. COBOL to Java Migration with Multi-Agent Frameworks. Persistent Systems
• 10. Refact.ai AI Coding Assistant | Fine-tuned, Secure, and Open-source - Refact.ai.
• 11. Getting code suggestions in your IDE with GitHub Copilot - GitHub Docs.
• 12. How to Connect Your AI Agent to Kubernetes: A Hands-On MCP Server Demo - YouTube.
• 13. Microsoft Security Copilot Threat Intelligence Briefing Agent | Microsoft Learn.
• 14. Using Agentic AI to Power Cloud Detection & Response | CrowdStrike.
• 15. IPAttributor: Cyber Attacker Attribution with Threat Intelligence-Enriched Intrusion Data.
• 16. Introducing AI-powered investigation in Chronicle Security Operations | Google Cloud Blog. Google Cloud
• 17. GitHub - dandye/adk_runbooks.
• 18. Agentic Incident Response - YouTube.
• 19. MITRE ATT&CK Driven Threat Hunting Automated by Local LLM - YouTube.
• 20. AI-Powered Test Automation: Ship Apps 3x Faster | Pcloudy. Pcloudy
• 21. Generative Agents: Interactive Simulacra of Human Behavior
• 22. Generative Agents: Interactive Simulacra of Human Behavior - YouTube.
• 23. A generalist AI agent for 3D virtual environments - Google DeepMind.
• 24. Procedural generation - No Man's Sky Wiki. Contributors to No Man's Sky Wiki
• 25. NLR 🌌 Nicolas Lester Reynolds on X: "3 unsupervised agents write a book - polishing Act III https://t.co/xxa9P3udNV" / X.
• 26. Engineering.
• 27. perplexitypages.com.
• 28. Introducing AI to FigJam | Figma Blog.
• 29. ”Dutch Insurance Provider”
• 30. Automating an insurance claim form with AI Builder - YouTube.
• 31. Transforming Insurance Underwriting with Agentic AI Risk Evaluation.
• 32. Revolutionizing Talent Acquisition in HR: PepsiCo's AI Success Story.
• 33. Introducing Hiring Assistant to help recruiters spend more time on their most impactful work . LinkedIn Talent Solutions
• 34. Transforming Insurance Underwriting with Agentic AI Risk Evaluation.
• 35. Ada Voice + Aircall - YouTube.
• 36. Introducing deep research | OpenAI.
• 37. Deep Research - Marginal REVOLUTION.
• 38. Deep Research: First Impressions. UX Tigers
• 39. ChemicalQDevice.
• 40. Automation of Systematic Reviews with Large Language Models | medRxiv.
• 41. Introducing Operator | OpenAI.
• 42. Launch YC: Kura: New State of the Art for Browser Agents [31% better than Computer Use] | Y Combinator.
• 43. H, the AI startup that raised $220M, launches its first product: Runner H for 'agentic' applications | TechCrunch. TechCrunch
• 44. Matt Shumer on X: "Today, we’re unveiling Personal Assistant - @HyperWriteAI's groundbreaking AI agent that can use a web browser like a human. One agent to rule them all. It’s time to reimagine the way we interact with the internet. https://t.co/csGjtIU0.
• 45. OmniParser.
• 46. Create an AI agent to fill forms from your private documents. IBMSkillsNetwork
• 47. Would You Let AI Control Your Computer? Testing Anthropic Claude's “Computer Use” Feature + Setup - YouTube.
• 48. Build a Deep Research Agent: $1 Alternative to $200 OpenAI's Tool. Analytics Vidhya
• 49. Introducing deep research | OpenAI.
• 50. Aider - AI Pair Programming in Your Terminal.
• 51. Creating an AI Agent for Financial Report Analysis - YouTube.
• 52. MultiOn AI Agent books a flight - YouTube.
• 53. Lisa Crosbie – Tech Tutorials | Microsoft Business Applications.