As 2025 winds down, it is a good time to reflect on the industry trends that actually happened, what we learned/ achieved, and where the future may be heading.
From Generative AI to Simply AI
Let us start with what was the buzz in early 2025. Whether it was the large software vendors or the analysts holding court, Agentic AI was the buzzword. Then, along came the MCP-Model Context Protocol to make it easier for LLMs/AI Agents to integrate with data sources and use tools/resources. Generative AI (GenAI) was the rage, and multi-modal models came along. Most of us as consumers adopted ChatGPT, and some went with Claude, and then there were those who embraced Gemini. We all started searching for information much more effectively/semantically, and Retrieval Augmented Generation (RAG) became the early use case of GenAI.
Fast forward, and AI agents are now being positioned as digital workers that will work alongside humans. Armies of AI agents that will use MCP and connect to each other via Agent to Agent (A2A) protocol to connect to enterprise systems and re-invent workflow & business processes. Microsoft then introduced Agent 365 as the governance control plane to manage agents securely. It was universally recognized that good data is the backbone of AI and data governance, and platforms like Microsoft Fabric and SAP Business Data Cloud gained ground. And sometime late in 2025, Generative AI started being simply called AI.
What we Learned and Built
At Unvired, at first, it felt like we were drinking from the AI firehose. Then we settled down and quickly learned, mainly by building AI agents, using AI for Coding, and implementing data platforms:
- AI Agents: We initially built some simple AI agents for Knowledge Search using RAG and experimented with various frameworks like LangChain. We then graduated to building more complex autonomous AI agents, like for HR Recruiting and Interview Assistant. We started to build on Microsoft Copilot Studio and prepare for SAP Joule Studio. We built an AI Agent for Maintenance and got it certified by SAP.
- Coding Agents: Our developers started using coding agents like Claude Code and GitHub Copilot, among others, for code review, development, and testing. This increased their productivity by 20% and enabled us to deliver applications faster to our customers.
- MCP Server: We built our very own MCP server for the Unvired Digital Enterprise Platform, making it easier to build AI agents.
- Data: Our customers wanted to lay the data foundation of the data lake so that they could use AI. This led us to implement Microsoft Fabric to simplify the implementation of data platforms. But, mere Power BI dashboards were not enough. We built Data Agents using Microsoft Fabric and used Copilot within Power BI to create reports using natural language.
The Future
I wish that I could predict the future, but can only hazard a guess. It seems that AI will cause a loss of jobs overall. The human workforce will need to be reskilled, and digital employees will augment humans. Physical AI will be the next big thing, and humanoid robots will dominate. And in Unvired’s immediate world of re-inventing business processes, Agentic AI or AI agents that plan, reason, take actions, self-evaluate, and course correct will come alive. That is what we are focused on so that we can continue to make our customers successful and help them sustain competitive advantage via differentiated/reinvented business processes.
We grew and achieved much in 2025 with the help of our customers, employees, and partners. We are excited about 2026 and the AI advancements yet to come.
