DoT Blog

  Logo courtesy of Databricks Over the last couple of years, the conversation around AI in enterprise software has moved quite quickly. Not long ago, the focus was on large language models (LLMs) and generative assistants; today, the discussion is shifting toward something more operational: agentic systems , AI agents capable of reasoning, planning, and acting across workflows. With the introduction of Genie Code , this past March, Databricks is making a clear statement about where it believes the future of data engineering and analytics is heading. The announcement positions Genie Code , the company’s tool designed for data teams for AI solution development, as a tool that allows developers and data professionals to build, orchestrate, and operate AI agents that interact directly with data environments. The interesting part is not just the tool itself. It’s, in my view, what it signals about the direction of the modern data platform.   From Data Pipelines to Agenti...

  Credit: Zoho Corporation In early March, Zoho announced the launch of its own Model Context Protocol ( Zoho MCP ). It is not often that a technology announcement signals something bigger than the feature itself, but the launch of Zoho MCP feels like one of those moments, not because it introduces yet another AI capability, but because it hints at how work platforms themselves may evolve in an era increasingly shaped by intelligent agents. Zoho describes its MCP as part of a framework designed to support the next generation of AI-driven work. The specifics revolve around enabling systems that can interact with applications, data, and workflows more autonomously. But beneath the terminology seems to lie a broader idea: the shift from application-centric software toward agent-centric systems of work. This shift is subtle today, but over time, it could become foundational.   From Applications to Coordinated Intelligence For decades, enterprise software has revolved around appli...

  Logo courtesy of  Amazon.com, Inc. For the past few years, the AI conversation has been dominated by large public models and cloud-scale services, but quietly, another trend has been gaining traction: the push toward private, autonomous AI agents running closer to where data and decisions live. With the recent introduction of OpenClaw on Amazon Lightsail , Amazon Web Services (AWS) is nudging that conversation forward. This new offering essentially allows developers and organizations to run autonomous AI agents in their own AWS-controlled environment using relatively simple cloud infrastructure. On the surface, this might look like just another developer-friendly deployment option, but if we step back for a moment, it hints at a deeper shift into how organizations might design AI systems in the near future.   From AI Assistants to Autonomous Agents So far, most of today’s AI implementations still operate basically as assistants. You prompt them, they respond, even in en...

  OpenText logo courtesy of Opentext Corporation When I read that OpenText has decided to sell Vertica in order to pay down debt, my first reaction wasn’t surprise, I admit, it was a bit of discomfort. Not because selling assets to reduce leverage is inherently wrong; for sure it isn’t. But because which asset you sell says a lot about how you see your future, and in this case, OpenText may be divesting one of the very pieces that could have mattered most in the next phase of enterprise software: high-performance analytics in an AI-driven world. In my view, this may turn out to be a sensible financial move, for the best, or it may end up being a strategic mistake that only becomes obvious later. Let’s unpack why.   Vertica Wasn’t Just Another Product Vertica is not a shiny, hype-driven analytics toy; it’s a battle-tested, high-performance analytical database designed for large-scale, complex workloads. Over the years, it earned a reputation for speed, efficiency...

  Teradata logo courtesy of Teradata, Inc. For the last two years, the enterprise AI market has been stuck in an awkward in-between state. Plenty of pilots, plenty of proofs of concept, but far fewer systems that run the business. So, with its newly announced Enterprise AgentStack, Teradata is clearly trying to address this gap, not by introducing yet another model, but by focusing on how AI agents are built, governed, and deployed at enterprise scale. What makes this announcement interesting, in my view, isn’t the buzzword density; it’s the direction of travel.   From “AI features” to agentic systems AgentStack is positioned as a framework for building and running AI agents that operate directly on governed enterprise data. The emphasis is not on experimentation but on operational AI: agents that can reason, retrieve, and act within defined business constraints. This matters because most enterprise AI failures don’t happen at the model layer; they happen at the integration ...

  Image by Freepik Two recent pieces, one from Wired and another from The Guardian , describe an experiment that feels equal parts absurd, fascinating, and unsettling: Moltbook , a social network designed almost entirely for AI agents talking to other AI agents. At first glance, Moltbook sounds like a gimmick: a bot-only social platform where AI personas post, reply, argue, form alliances, and generate content, without humans taking part directly. But once you look past the novelty, Moltbook becomes something more interesting and at times even concerning: a mirror held up to the direction social platforms, AI agents, and digital interaction may be drifting toward. So, let's address it.   A Social Network Without Humans (Mostly) According to the reporting, Moltbook allows users to deploy AI agents as social actors; these agents create profiles, generate posts, comment on each other’s content, and build reputations. Humans can see, tweak prompts, or set high-level goals, bu...

  Image by DC Studio on Freepik Dear friends, Yes, it’s true, organizations worldwide are racing to adopt artificial intelligence (AI), but many are tripping over a surprising obstacle: their data. But why? Well, because the very fuel AI relies on, the business data that should inform decisions, is often fragmented, inconsistent, or simply unreliable. This is not a niche issue; surveys consistently show that more than half of business and technology leaders cite poor data quality as a top barrier to AI adoption. The paradox seems clear: companies invest in sophisticated AI models without first ensuring their data is ready to support them. The result? AI projects stall, insights are misleading, and innovation slows. But fixing this doesn’t mean just collecting more data; it means rethinking how data is structured, connected, and understood. So, below are five takeaways that challenge conventional wisdom and reveal how organizations can, potentially, turn data from a liability into a...