In the dynamic business world or tech environment of 2025, it has become increasingly challenging to keep up with trends. In the world of AI, the business, political, and technology stacks continue to shift weekly. We’re now entering the agentic area of AI. Agentic research tools will become a huge time saver. Researchers and the tech curious can potentially explore big questions using tools like deep research. As a Google developer group leader, I will focus my comments on the innovations released from Google Gemini. Many of the comments that I’ll share during this exploration will extend to OpenAI and other foundational players.
What is agentic computing? How is it different from leveraging a foundational large language model tool? An AI agent connects the power of a large language model to tools. Let’s say we’re building an AI agent that specializes in creating family friendly travel plans with reasonable price constraints. This agent will need access to a search engine like Google, a travel estimation tool like Expedia, and other data sources. An AI agent will be given autonomy by the user to achieve a useful goal. This requires the agent to consider different threads or possibilities. The agent will have the capacity to plan and explore the different options, decide if the agent achieved the goal, and try again if it fails. Let’s think about the AI travel agent scenario. A human travel agent does many Google searches to help inform potential sub-plans and options. A human travel agent integrates these sub plans into a unified plan and evaluates if they have met the expectations of their client. ( financial or fun factor wise )
I took Google Gemini Deep Research for a test drive a few weeks ago. I have to confess that I was blown away. In the Google Gemini interface, the research enters their research prompt and turns on the “deep research” mode. Gemini decomposes the research topic into multiple topics. I appreciated how Gemini shows the proposed research plan to the user. At this point, you can refine the key questions explored by the agent. Once you confirm the research task, you can grab a coffee and watch the AI do its thing. The UX shows the status of the research and various insights the agent has gathered. In my quick trials, the research activity lasts five minutes. At the end of the session, the user can export the insight to a Google document.
To test drive Gemini Deep Research, I asked Gemini to explain the value of open source software to business stakeholders. I further asked it to explore the key business models for running an open source business. You can inspect the report here. The report felt valid and the system properly shows the sources for different claims in the paper.
I compared the Gemini work against a post written by a human author on a similar topic:
A Practical Guide to Open Source Business Models by Regina Nkenchor
I do wish the Gemini Deep Research had sharper details on the various open source business models. I still find the output from deep research pretty remarkable. The value of insight with source links was pretty amazing. In a work context, I have pushed the tool to build a survey paper around key AI strategy questions. I have found the Gemini Deep Research insight equally helpful.
Let’s add NotebookLM
Let’s push the experience one step further. Over the weekend, I was curious about the future of AI agents and the potential industries that would benefit from this kind of technology. I used Google Deep Research to explore the value from AI agent software using open source tools, cloud providers, and commercial tools. I took the deep research report and imported it into Notebook LM. From there, I created the following podcast summary.
Executive Summary:
Artificial intelligence agents are rapidly evolving from theoretical concepts to practical tools with the potential to revolutionize operations across a multitude of industries. These autonomous systems, capable of reasoning and interacting with their environment to achieve specific objectives, represent a significant leap forward in AI-driven automation. This report provides a comprehensive analysis of the major tools and platforms that are fueling the creation of these intelligent agents. It explores the landscape of cloud-based offerings from leading providers, the dynamic ecosystem of open-source frameworks, and the specialized solutions offered by commercial platforms. Furthermore, the analysis identifies the key industries that are poised to experience profound transformations through the adoption and implementation of AI agents, highlighting the potential for increased efficiency, enhanced productivity, and innovative solutions to complex challenges. The report concludes with an outlook on the future trajectory of AI agent development and adoption, underscoring the trends and considerations that will shape this evolving technological landscape.
Leave a Reply