Agentic AI: A New Era of Intelligent App Development

The dawn of a new era of intelligent app development using AI Agents marks a significant milestone in the evolution of software engineering. This paradigm shift is transforming the way we design, develop, and deploy applications, making them more adaptive, responsive, and user-centric than ever before.

Tracing back the history of application development, it all started with mainframe systems in the 1960s and 1970s, where applications were monolithic and ran on large centralized computers. The advent of personal computers in the 1980s and 1990s brought about the era of desktop applications, which were standalone and operated on individual machines.

With the explosion of the internet in the late 1990s and early 2000s, web-based applications became the norm, enabling users to access applications from anywhere via a browser. The mobile revolution in the late 2000s and 2010s further changed the landscape, with a shift towards developing applications specifically designed for mobile devices

The cloud era that followed allowed for applications to be hosted on remote servers, providing scalability, flexibility, and cost-effectiveness. Today, we are witnessing the rise of Low-Code/No-Code (LCNC) and Language Model (LM) based application development using AI Agents, which simplifies the development process and democratizes application creation. This new era leverages AI to automate coding, testing, and deployment, thereby reducing the barrier to entry and enabling a wider range of individuals to create applications

In this new era, AI Agents are not just tools but collaborators that assist developers in creating more intelligent and sophisticated applications. This evolution signifies a promising future for application development, where AI plays a central role in shaping the applications of tomorrow.

AI Agents

In artificial intelligence, an agent is a computer program or system that is designed to perceive its environment, make decisions and take actions to achieve a specific goal or set of goals. The agent operates autonomously, meaning it may not be directly controlled by a human operator always.

GenAI Agents are a significant advancement in the field of Generative AI. They are autonomous entities that leverage Language Learning Models (LLMs) to manage unstructured data and navigate non-deterministic paths. Unlike stand-alone LLM-based applications that require human input, GenAI Agents can plan and execute tasks end-to-end, monitor the output, adapt, and use tools to accomplish goals. They are dynamic systems that can sense and act on their environment. GenAI Agents are part of the GenWorlds framework, which allows for the creation of multiple agents, each with a narrow mission, that can work together towards a common goal. They are expected to be transformative, automating entire workflows and bringing about significant efficiency gains.

Features of AI Agents

• Integration with Language Learning Models (LLMs): GenAI Agents leverage LLMs to manage unstructured data and navigate non-deterministic paths.
• Autonomous Operation: Unlike stand-alone LLM-based applications that require human input, GenAI Agents can plan and execute tasks end-to-end, monitor the output, adapt, and use tools to accomplish goals.
• Dynamic Systems: GenAI Agents are dynamic systems that can sense and act on their environment.
• Continuous Learning and Adaptation: They continuously learn, self-improve, and adapt to changing circumstances and environments.
• Collaboration: GenAI Agents are part of the GenWorlds framework, which allows for the creation of multiple agents, each with a narrow mission, that can work together towards a common goal.
• Efficiency Gains: They are expected to be transformative, automating entire workflows and bringing about significant efficiency gains.

Some Application of AI Agents

• Customer Service: GenAI Agents are used in customer service to auto-generate customer replies. For instance, Service GPT by Salesforce uses GenAI to understand the customer query and sift through various knowledge sources to automate a relevant customer response. Another example is Zendesk’s “expanding agent replies” solution, which allows agents to type the bare bones of their response and then fleshes it out for them.
• Insurance Claims Processing: In the insurance industry, GenAI Agents are used to automate the end-to-end claims process. For example, using an uploaded image, GenAI can automatically generate an instant settlement offer, relying on an archive of millions of vehicle damages photos and incident reports.
• Wireless Collective Intelligence: In the field of wireless collective intelligence, BabyAGI, a task-driven autonomous agent framework, can generate, execute, and prioritize tasks in real-time.

Application development frameworks

Semantic Kernel

Semantic Kernel is an open-source Software Development Kit (SDK) that allows developers to easily build agents that can call existing code. It integrates Large Language Models (LLMs) like OpenAI, Azure OpenAI, and Hugging Face with conventional programming languages like C#, Python, and Java.

Semantic Kernel is at the center of the agent stack, allowing developers to combine AI models and plugins together to create brand new experiences for users. It achieves this by allowing you to define plugins that can be chained together in just a few lines of code.

Follow this link for additional details: https://www.dhirubhai.net/pulse/build-copilots-using-semantic-kernel-tarun-sharma-1zghc/

LangChain

LangChain is an open-source framework designed to simplify the creation of applications using large language models (LLMs). It provides a standard interface for chains, lots of integrations with other tools, and end-to-end chains for common applications.

LangChain is essentially a library of abstractions for Python and Javascript, representing common steps and concepts necessary to work with language models. These modular components—like functions and object classes—serve as the building blocks of generative AI programs.

LangChain is a suite of products that helps developers build and deploy reliable GenAI apps faster. LangSmith is the enterprise DevOps platform for LangChain that provides evaluation, testing, and monitoring tools for LLMs.

Follow this link for additional details: https://www.dhirubhai.net/pulse/intelligent-ai-app-langchain-tarun-sharma-l1z3c/

AutoGen

AutoGen is an open-source programming framework developed by Microsoft that enables the development of Large Language Model (LLM) applications using multiple agents. These agents can converse with each other to solve tasks.

AutoGen agents are customizable, conversable, and seamlessly allow human participation. They can operate in various modes that employ combinations of LLMs, human inputs, and tools.

AutoGen simplifies the orchestration, automation, and optimization of a complex LLM workflow. It maximizes the performance of LLM models and overcomes their weaknesses. It supports diverse conversation patterns for complex workflows.

With customizable and conversable agents, developers can use AutoGen to build a wide range of conversation patterns concerning conversation autonomy, the number of agents, and agent conversation topology.

AutoGen provides a collection of working systems with different complexities. These systems span a wide range of applications from various domains and complexities.

Follow this link for additional details: https://www.dhirubhai.net/pulse/autogen-enabling-llm-applications-tarun-sharma-bmbqc/

Evolution of App development

In the past, traditional software, before Generative AI, was limited by its reliance on structured data and deterministic, rule-based algorithms, struggling with unstructured inputs and lacking the adaptability of human thought. These systems, bound by the programmer’s foresight, lacked the nuanced judgment and flexibility inherent in human reasoning.

Presently, in the era of Language Learning Models (LLMs) like OpenAI’s GPT series, we see a paradigm shift as these models dynamically generate outputs from unstructured inputs, seemingly reflecting human reasoning. However, as passive tools that only respond to inputs, LLMs alone won’t usher in the anticipated mass automation with Generative AI.

In the future, AI Agents, at the vanguard of software evolution, leverage Language Learning Models (LLMs) to manage unstructured data and navigate non-deterministic paths, embodying the concept of “Autonomous Software”. Far from being static, these agents continuously learn, self-improve, and adapt to changing circumstances and environments.

Generative AI architecture

1. The bottom layer is the model. We have more possibilities in model selection, including not only Azure OpenAI Service, but also open-source small models provided by Azure AI Studio or Huggingface.
2. We can use Microsoft Olive combined with Windows AI Studio to complete the local fine-tuning of the open-source small language model. Of course, in the stage where the parameters are relatively complex, we can also migrate the fine-tuning to Azure AI Studio.
3. We can use ONNX Runtime and Microsoft Olive to run the model at the reference layer, or directly reference and deploy the model through Windows AI Studio and Azure AI Studio.
4. Use Prompt flow to evaluate the effectiveness of prompt projects and models in enterprise application scenarios to improve generative artificial intelligence.
5. Combine different needs and frameworks to quickly build applications and deploy them to different terminals.

Summary

GenAI Agents, is a significant advancement in field of Artificial Intelligence and Agents takes the discourse to even higher level. These autonomous AI Agents leverage Language Learning Models (LLMs) to manage unstructured data and navigate non-deterministic paths. Unlike traditional LLM-based applications, GenAI Agents can plan and execute tasks end-to-end, adapt, and use tools to accomplish goals. They are part of the GenWorlds framework, allowing for the creation of multiple agents that can collaborate towards a common goal.

References

• What is Semantic Kernel? https://learn.microsoft.com/en-us/semantic-kernel/overview/
• Semantic Kernel: https://github.com/microsoft/semantic-kernel
• Applications that can reason: https://www.langchain.com/
• Understanding LangChain - A Framework for LLM Applications: https://www.projectpro.io/article/langchain/894
• Enable Next-Gen Large Language Model Applications: https://microsoft.github.io/autogen/
• Multi-agent Conversation Framework: https://microsoft.github.io/autogen/docs/Use-Cases/agent_chat
• Gen(erative) AI Agents: https://www.managen.ai/Understanding/agents/index.html
• Journey Series for Generative AI Application Architecture - Foundation: https://techcommunity.microsoft.com/t5/educator-developer-blog/journey-series-for-generative-ai-application-architecture/ba-p/4065564