Blog

Maximilian Kenfenheuer

Manager @ BearingPoint

Maximilian Kenfenheuer

Featured

Why Copilot Notebooks’ support for Markdown and text files matters more than it seems

Markdown may sound like a small file-format update. I do not think it is. Microsoft’s move to let Copilot Notebooks work with 𝐌𝐚𝐫𝐤𝐝𝐨𝐰𝐧, 𝐩𝐥𝐚𝐢𝐧-𝐭𝐞𝐱𝐭, 𝐚𝐧𝐝 𝐫𝐢𝐜𝐡-𝐭𝐞𝐱𝐭 𝐟𝐢𝐥𝐞𝐬 points to something more important for Microsoft AI solutions: enterprise AI is becoming more useful when it can reason over the 𝑎𝑐𝑡𝑢𝑎𝑙 𝑤𝑜𝑟𝑘𝑖𝑛𝑔 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑠 teams rely on every day. That includes READMEs, wikis, logs, transcripts, notes, and operational text that often sits outside polished documents and slide decks. In the article, I explore why that matters: • why unstructured working content is often where real context lives • how broader file support can make Copilot Notebooks more relevant for technical, operational, and cross-functional teams • why this improves continuity between knowledge capture and AI-assisted reasoning • and what organizations should think about as they expand the range of content they want AI to work with responsibly The next phase of enterprise AI may depend not only on better models, but on whether those models can work effectively with the formats people already use to run the business. Which overlooked content types do you think will create the most value once AI can use them more naturally?

Continue reading

Why Frontier Tuning Could Become a Strategic Advantage in Microsoft AI Solutions

Customizing AI is moving beyond prompts and policy settings. Microsoft’s new 𝐅𝐫𝐨𝐧𝐭𝐢𝐞𝐫 𝐓𝐮𝐧𝐢𝐧𝐠 approach stood out to me because it points to a more important shift in enterprise AI: organizations will increasingly want agents that do not just sound smart, but work in ways that reflect their own processes, terminology, controls, and standards. That is especially relevant for Microsoft AI solutions. If tuning can happen inside the organization’s compliance boundary, using real workflows, business knowledge, and evaluation signals, the conversation changes. It becomes less about generic AI capability and more about operational fit. In the article, I explore why this matters: • why enterprise AI value increasingly depends on adaptation, not just access • how Frontier Tuning could help agents align more closely with company-specific ways of working • why reinforcement learning, evaluation, and governance now need to be considered together • and what organizations should think about as they move from using AI tools to shaping AI behavior The next phase of enterprise AI may depend less on whether a model is powerful in general, and more on whether it can be taught to perform well in the specific context of the business. How important do you think organization-specific tuning will become as companies try to turn AI into a real operating advantage?

Why Connected Apps Inside Microsoft 365 Copilot Matter More Than Another AI Feature

One of the biggest AI adoption problems is still surprisingly simple: the answer shows up in one place, but the work still has to happen somewhere else. That is why Microsoft’s move to bring business apps directly into Microsoft 365 Copilot deserves more attention. When agents can surface experiences from tools like Dynamics 365, Adobe Express, Figma, Box, or monday.com inside the Copilot flow, the value is not just convenience. It is a shift in operating model. AI becomes more useful when it can stay connected to the systems where teams actually create, update, approve, and execute work. In the article, I explore why this matters for Microsoft AI solutions: • why reducing context switching is becoming a strategic design goal • how in-chat app experiences can narrow the gap between insight and execution • why connectors, agents, and governance now matter together • and what organizations should think about as Copilot becomes a more connected work surface The next phase of enterprise AI may depend less on generating another good answer, and more on whether that answer can move work forward across the tools the business already depends on. How important do you think connected app experiences will be in turning AI from assistance into real execution?

Latest posts

Why Copilot Notebooks’ support for Markdown and text files matters more than it seems

#Microsoft365Copilot #MicrosoftAI #AI #Copilot #EnterpriseAI

Markdown may sound like a small file-format update. I do not think it is. Microsoft’s move to let Copilot Notebooks work with 𝐌𝐚𝐫𝐤𝐝𝐨𝐰𝐧, 𝐩𝐥𝐚𝐢𝐧-𝐭𝐞𝐱𝐭, 𝐚𝐧𝐝 𝐫𝐢𝐜𝐡-𝐭𝐞𝐱𝐭 𝐟𝐢𝐥𝐞𝐬 points to something more important for Microsoft AI solutions: enterprise AI is becoming more useful when it can reason over the 𝑎𝑐𝑡𝑢𝑎𝑙 𝑤𝑜𝑟𝑘𝑖𝑛𝑔 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑠 teams rely on every day. That includes READMEs, wikis, logs, transcripts, notes, and operational text that often sits outside polished documents and slide decks. In the article, I explore why that matters: • why unstructured working content is often where real context lives • how broader file support can make Copilot Notebooks more relevant for technical, operational, and cross-functional teams • why this improves continuity between knowledge capture and AI-assisted reasoning • and what organizations should think about as they expand the range of content they want AI to work with responsibly The next phase of enterprise AI may depend not only on better models, but on whether those models can work effectively with the formats people already use to run the business. Which overlooked content types do you think will create the most value once AI can use them more naturally?