Articles about AI dominated media protection in 2025, as the power to carry out extra than simply rudimentary duties grew into making software program improvement life cycle processes autonomous. The widespread piloting and implementation of generative AI, brokers and MCP servers have been among the many matters that produced the highest trending articles on sdtimes.com this 12 months. However AI wasn’t the one factor on the minds of builders and their managers. Right here’s a take a look at the highest trending tales on sdtimes.com for 2025.
Probably the most-viewed article this 12 months requested the query: “Is Agile Useless within the Age of AI?” Creator Adam Sandman, CEO of Inflectra, posited that Agile processes are in truth not useless, however are evolving. “The way forward for software program improvement isn’t Agile vs. AI, it’s Agile with AI. Strategic alignment, mentorship, and good governance guarantee that AI’s energy enhances security, maintainability, and long-term product worth.,” he wrote.
No. 2: 88% of firms are considering leaving Oracle Java
SD Occasions editor Jenna Barron dug right into a January report by Java platform supplier Azul Methods that confirmed Oracle modified its licensing mannequin to base the fee on the variety of staff a company has, as a substitute of charging by the Java occasion in use. “Since then,” Barron wrote, “organizations have been considering options, and 72% of respondents have been already eager about it when surveyed in 2023.”
The report reveals that the highest 5 causes for wanting to maneuver on from Oracle embrace price (42%), desire for open-source (40%), Oracle’s gross sales ways (37%), uncertainty round altering licenses and pricing (36%), and restrictive Oracle insurance policies (33%).
“AI adoption is saving builders a major period of time, however a number of friction factors nonetheless exist all through the software program improvement life cycle that builders are nonetheless dropping time to,” Barron wrote from August research of developer expertise and engineering administration.
Atlassian’s 2025 State of Developer Expertise report confirmed, “So we’re proper again the place we began, with builders saving 10 hours per week utilizing AI and dropping 10 hours per week to inefficiencies. Enhancing the developer expertise requires a scientific method to understanding and resolving developer friction factors.”
No. 4: AI gained’t change builders, however it would depart some behind
Dan Faulkner, CEO of Good Bear, wrote: When execs ask, “What number of builders can we reduce if we embrace AI?”— They’re asking the fallacious query.
The suitable query is: How can we evolve our whole software program lifecycle to match the speed AI makes potential with out breaking belief or burning down high quality?
“The longer term doesn’t belong to groups that write probably the most code. It belongs to those that ship probably the most resilient, reliable, and scalable software program,” he wrote. “That future wants improvement groups. But it surely wants a distinct mindset and a distinct type of management.”
No. 5: Why AI brokers want a protocol like MCP to succeed in their potential
MCP can do for AI brokers what USB does for computer systems, Lin Solar, senior director of open supply at cloud native connectivity firm Solo.io, defined.
As an example, a pc wants a method to connect with peripherals like a mouse, keyboard, or exterior storage, and USB is a typical that gives that connectivity. Equally, MCP permits AI brokers to connect with completely different instruments and knowledge sources, like Google Calendar. It offers “a typical method to declare the instruments so the instruments may be simply found and may be simply reused by completely different AI purposes,” she mentioned.
No. 6: Google goes all in on agent improvement, Gemini at Google Cloud Subsequent 25
At its Google Cloud Subsequent 25 occasion in April, Google introduced numerous new bulletins, primarily associated to enhancing Gemini choices and making it simpler to construct and undertake brokers.
Among the many bulletins have been the discharge of ADK, Google’s Agent Improvement Package; an Agentspace platform on which builders can construct brokers combining Gemini fashions, Google-quality search performance, and enterprise knowledge; and launched Google Workspace Flows, enabling customers to create agentic workflows to streamline processes and automate repetitive duties throughout Workspace.
No. 7: Containers in 2025: Bridging the hole between software program and {hardware}
Containers have lengthy been a preferred method of packaging up and delivering software program, however many builders have additionally begun to discover utilizing containers in additional methods than initially meant.
In a current episode of the SD Occasions podcast, What the Dev, Scott McCarty, senior principal product supervisor for Crimson Hat Enterprise Linux, sat down with us to debate the tendencies he’s been seeing and likewise make predictions for what’s to return.
No. 8: tRPC vs GraphQL vs REST: Selecting the best API design for contemporary net purposes
Software program engineer Kacper Mihalik contributed this information to SD Occasions evaluating these applied sciences from a sensible engineering standpoint, after a number of years of constructing manufacturing methods with React and TypeScript.
“I’ve shipped REST, GraphQL, and tRPC APIs,” he wrote. “Every choice presents distinct strengths, with real-world tradeoffs builders and engineering leaders ought to perceive. … specializing in structure, sort security, toolchains, and developer expertise.
ChatGPT captured the world’s creativeness, however it might have additionally trapped it. The chatbot interface—with its acquainted conversational format—made AI accessible to thousands and thousands, demonstrating the exceptional capabilities of huge language fashions (LLMs) in a bundle that felt pure and alluring. But this very success has created a false impression: that AI equals chatbots, and that each software wants a chat window to be AI-powered.
Peter Day, a normal accomplice at tremendous{set} – a startup studio the place nice knowledge+AI concepts turn into transformational software program firms – wrote: “The way forward for AI isn’t about chat home windows. It’s about invisible intelligence woven seamlessly into how individuals work, making complicated duties less complicated and tedious work disappear. That future requires rethinking consumer expertise from the bottom up, not retrofitting chatbots onto current merchandise. The winners might be those that acknowledge this distinction and design accordingly.”
As SD Occasions editor Barron reported in November, this 12 months’s up to date record options lots of the identical issues from the 2021 variations, with a number of notable modifications, reminiscent of Server-Facet Request Forgery, which was in final place in 2021, being rolled into the Damaged Entry Management class.
“Moreover, a brand new class, Software program Provide Chain Failures, was added and contains Susceptible and Outdated Parts (#6 in 2021), and Mishandling of Distinctive Circumstances made the record for the primary time, containing CWEs associated to improper error dealing with, logical errors, failing open, and different associated situations,” Barron wrote.