Dive into the innovative world of agentic automation with UiPath – We need your insights!

[loredana_ifrim]

Dear UiPath Community,

As we journey into AI’s potential, we see RPA evolve and anticipate new AI agents for better work. Our platform helps you invest in automation while making it easier to use AI. This leads us to agentic automation, opening up a new era of efficiency and smart operations.

What does this mean for you? We want you, our community, to talk, share, and give feedback on this new AI area. Share your experiences:

1. How has the combination of UiPath and AI revolutionized your work?
2. How do you think agentic automation will improve efficiency in the UiPath framework?
3. Are there unique success stories or breakthroughs that you’ve achieved using agentic automation and UiPath?

This is more than just a thread. It’s our collective journey towards a smarter and more efficient future. So, grab your favorite drinks, warm up your typing fingers and brace yourself for this agentic automation adventure. Remember, every contributor garners learning, every insight unfolds a possibility, and every shared experience enriches our community. Are you ready to set forth on this agentic automation journey with UiPath?

Looking forward to hearing your stories,

UiPath Community

Here you can listen a conversation between @zell12 one of our MVPs and Sandeep Panda, Director, Technology Alliances at UiPath — on our UiPath AI infused capabilities.

[zell12 (Russel Alfeche)]

Thanks @loredana_ifrim ! Great initiative!!

Here’s my 2 cents:

Software automation has come a long way from being just a glorified screen scraping solution, to a full-fledged Business Applied AI, that is core in automating end to end critical business processes.

Recent advancements mean we can now automate tasks that are far more complex, with more efficiency, than we could just two years ago—or even six months ago.
Take UiPath DocPath for example - a fine-tuned large language model for information extraction from documents. This enables highly accurate extraction for unstructured documents without having the hassle to train a specialized model, to start. On the other hand, it makes model training 10x faster, with Active learning capability within Modern DU experience.

Case in point: Here’s a very material and recent success story, that was made possible with UiPath DocPath and other GenAI capabilities within UiPath: Pricing for Deluxe is Simplified with GenAI | UiPath

The integration of Generative AI across entire UiPath product suites exemplifies this progress. Whether you’re a developer using UiPath Studio or Apps, a tester working with Test Suite, or a business user relying on an assistant, GenAI powered Autopilot is embedded throughout, enhancing capabilities across the board.

But what does the term ‘agentic’ truly mean in this context? Is it merely a buzzword that convolutes discussions about the promise of agentic automation? I would say, much of the current conversation—such as claims like “RPA is dead, agentic workflows is the future”— is partial, because, it is more plausible that the promise of the latter will cease to exist without the former.

I’d go as far to say that the real meaning of agentic automation is already within our grasp!
Solutions like UiPath Autopilot™ for Assistant and Agent Productivity Kit - an AI powered productivity tool designed for contact center agents, already exists, ones that which you can try now.

All this said, it’s not about whether agents can replace bots, but it’s more about how agents can supplement existing automations, making it more intelligent in more ways than one.

In simplest terms, agents - when given a goal/purpose and a set of tools - can plan/execute/iterate actions until the primary goal is reached

Below are some example and frameworks:

TLDR;

• Goal 1: Compare stock performance of google and micrsoft
• Framework: Chain of Thought - helps automate complex processes by structuring the AI's reasoning in a way that is logical and transparent, making it easier to identify areas for improvement or optimization. 
• Tools: google search, scrape website, stocks api

• Goal 2: Process new loan requests from mailbox, reconcile the books
• Framework: ReAct framework - essential for handling unpredictable or evolving tasks, as it allows the automation to adapt in real time, making decisions that balance immediate actions with ongoing analysis.
• Tools: microautomations interacting with multiple business systems.

Chain of Thought Framework:
The Chain of Thought framework is a cognitive approach that mirrors human reasoning. It breaks down the decision-making process into a sequence of logical steps, each building upon the previous one. This framework is particularly effective in situations that require careful consideration and layered thinking.

Steps Involved:

• Question: Identify the primary objective or problem to solve.
• Observation: Gather data and relevant information about the problem.
• Action: Take a specific action based on the gathered information.
• Thought: Reflect on the outcome of the action, which may lead to new questions or further actions.

Application Example:

• Goal: Compare the stock performance of Google and Microsoft.
• Process:
  Question: How have Google and Microsoft stocks performed in the past year?
  Observation: Use tools like Google Search and stock APIs to collect data.
  Action: Scrape relevant financial websites for historical stock prices.
  Thought: Analyze the data to determine trends, leading to conclusions or further questions (e.g., what external factors influenced these trends?).

ReAct Framework
The ReAct framework, short for Reason + Act, emphasizes a more dynamic interplay between reasoning and action. It is particularly well-suited for tasks that require real-time decision-making and adaptation, allowing for more agile responses to changing circumstances.

Steps Involved:

• Reason: Analyze the current situation or problem to determine the best course of action.
• Act: Execute the chosen action based on the reasoning.
• Iterate: Continuously assess the outcomes and refine the reasoning and actions as necessary.

Application Example:

• Goal: Process new loan requests from a mailbox and/or reconcile the books.
• Process:
  Reason: Determine which loan requests are new and identify the data required for processing.
  Act: Use micro-automations to extract relevant data, validate it, and update the financial records.
  Iterate: As new loan requests arrive or discrepancies in the books are detected, the system adjusts its actions and refines the process.

If you’d like to see more of, all things agents and automation, see below prototype I recently recorded showcasing a multi-agent system. Creating a Multi-Agent Workflow that Builds Automations!

Do share your thoughts as well. Will be happy to discuss and engage with you all!

[chaitanya.kulkarni]

Integrating UiPath with Agentic AI has truly transformed how we approach automation. As an Automation Architect deeply involved in AI, I’ve seen the power of combining these technologies to create more intelligent, adaptive systems that go beyond traditional automation. With Agentic AI, we’re not just automating tasks; we’re creating bots that can learn, adapt, and make decisions on their own.

One of the most exciting projects I worked on involved implementing Agentic AI in a customer support process. Initially, the automation was designed to handle routine tasks like ticket classification and basic responses. However, by integrating Agentic AI, we enabled the system to learn from previous interactions, predict customer needs, and even suggest personalized solutions without human intervention. This not only improved response times but also enhanced customer satisfaction by providing more accurate and timely support.

What’s fascinating about Agentic AI is its ability to operate autonomously while continuously improving. This is where I see the future of automation heading—bots that aren’t just following instructions but are evolving to handle increasingly complex tasks. It’s a thrilling time to be working with this technology, and I’m excited about the potential it holds for reshaping our workflows.

[SUDARSHAN_SHARMA]

As someone deeply involved in the RPA domain with hands-on experience in implementing intelligent automation solutions, the integration of UiPath and AI has been transformative in my journey.

The synergy of UiPath and AI has allowed me to tackle complex automation challenges that previously required significant manual intervention. For instance, in a project involving Registration for Golden VISA holders, AI capabilities like document classification and NLP seamlessly enhanced verification accuracy and reduced process cycle time.

Agentic automation, with its focus on autonomous and adaptive agents, is the next frontier. I foresee these agents driving greater contextual decision-making within workflows. For example, in projects involving telecom infrastructure, these agents could independently monitor and adjust workflows, ensuring seamless operation even during unexpected changes.

Agentic automation holds the potential to push the boundaries further, fostering a future where automation is not just task-oriented but truly intelligent and self-reliant. I’d love to hear others’ thoughts and experiences in this exciting area!