Five major predictions for 2026

By Geoff Livingston, The Buzz by Geoff Livingston, a reader-supported publication.

It’s that time of year, so I’m breaking out the crystal ball with my tech predictions for 2026. Just a warning, my mother was a famous astrologer, so I relish this just a little too much. Of course, many of these are AI-centric, but there is one in which AI is subtext, not the dominant theme.

So with no further ado, here are my top five predictions for 2026.

1) AI Skill Testing Arrives

Debate all you want, one thing that’s become clearer is that people who use AI are faster than those who do not. Employers are increasingly demanding that job applicants possess AI skills, perhaps their best way to bring their workforce up to speed with new productivity methods. And while many claim to have such skills, some just know how to ask basic questions of ChatGPT.

Moving forward, employers will increasingly ask candidates to take AI tests to ensure skills are present, and may hire the most proficient AI users. This was not just my thinking, but also Gartner, which made it their top prediction of the year.

Gartner anticipates 75% of employers will demand candidates take AI tests, but I do think that’s only true of medium and large enterprises that can purchase third-party testing, such as Codility, for developers.

Further, as time progresses, employers will demand continued testing as part of ongoing employment. After all, AI evolves quickly than any technology we’ve seen in recent memory. Yesterday’s prompt engineer may not adapt well to today’s increasingly nimble and agentic LLMs and AI-infused SaaS platforms. Expect employers to require ongoing skills development, including training and certifications.

2) Agentic AI Catches Up to the Hype

So many LLM-based AI workflows are triggered manually, disaggregated in flow, and require human connectivity. Wasn’t the promise of agents that they would reduce the rote nature of managing cojoining and linear tasks? We’ve seen a lot of hype and many failures. Yet big tech cos keep touting all the agentic work they have already automated (and the people they claim they no longer need).

So what gives?

Time has given developers the ability to work through data integrations via the MCP protocol and refine automations with LLMs to create better and more polished AI agents. The result is the rise of new user-driven, agentic-like flows in Google Workspace, ChatGPT, and Anthropic, new platforms that have taken on the risk of competing with larger platforms, and software-specific AI integrations that automate previously tedious tasks.

While many of these are automations, they nevertheless fall under the current umbrella of AI agents. Here are three examples I have experienced in the past week

• Google Gemini is automatically asking me to add appointments to my calendar and source relevant Google Drive documents
• The Top unifies your workspace, email, and calendar to provide a strategic Chief of Staff perspective of what needs to get done today. While at risk of being simply relegated to irrelevance by platform upgrades from Google and Microsoft, The Top is currently superior.
• MailChimp finally got its domain verification agent straightened out. I experienced a similar easy process with Beautiful.AI. Verifying domains is a painful, tedious task, but now platforms ask you to verify your identity and handle the rest through their agents. Thumbs up!

Expect to see an increasing proliferation of similar handy agents in our digital lives. Now if Apple could finally get agents straight with Siri, we’d be in a much better place!

3) Social Network Engagement Drops

Depending on how you measure it, AI engagement has largely been flat or decreasing per post. That’s because of a larger proliferation of posts in general, and the increasing arrival of slop (AI-generated content). In the end, value matters most to people, but do brands, influencers, and anyone else using slop, tricks, and memes get that? Or are they just slaves to social networking algorithms?

One of the major problems with social networks is fatigue from the content, but also from the algorithmic sourcing of content that generates most time on site. One concern is the lack of veracity and usefulness of such content.

While cheap entertainment, it may not help people find the information they need, which could, in turn, cause people to turn to other sources, such as LLMs and other media forms. Add the increasing toxicity of online political conversation, which rewards tabloid positions that provoke reactions, and you have the recipe for declining engagement rates.

Still, traffic is expected to stay flat or grow slightly. Call it a hunch, but I believe more and more people are simply going to opt out and stop giving Meta, TikTok, and other entertainment-centric social networks as much of their time. Maybe it’s better to read a book or exercise instead.

4) OpenAI Suffers a Major Crisis

I think 2026 is the year OpenAI suffers a major crisis that impacts market share. You could say this already happened with the botched launch of ChatGPT 5 and the rise of Gemini. But you could say the same thing about Anthropic’s rise on the B2B side. This new crisis will be different as it will be a self-inflicted wound.

Those who follow the AI space already know Sam Altman’s proclivity for bombastic statements that don’t come true, questionable ethics, embracing of circular deals, and failure to build a strong mote that would insulate the company from further erosion.

With so much weight on the company’s shoulders and heavy competition at the door, one major crisis could become catastrophic for the company. And the guardrails are gone. Leaders have been fleeing OpenAI for two years now; the nonprofit structure is gone, the company has fostered and created a questionable business without regulation, and has repeatedly violated its users’ rights.

Inevitably, arrogance creates problems. And I expect Sam Altman will have a significant moment to rise up to or fail to meet, whether that is a cybersecurity incident, a major financial collapse, or an ethics-inspired backlash with its user base. OpenAI is heading for dangerous waters in 2026.

5) Anthropic Acquires a Visual AI Algorithm Company

Both Google and OpenAI benefit from multimodal capabilities with superior image and video generation capabilities integrated into Gemini and ChatGPT, respectively. If Anthropic expects to keep Claude competitive, it needs to match those capabilities, as Claude’s visual multimodal capabilities are inferior.

To match Google and OpenAI, Anthropic will need to acquire a visual media generator. Midjourney, Runway, Stable Diffusion, Flux, Leonardo, and many others all have attractive algorithm offerings that would make great additions to Claude.

(This post originally appeared on The Buzz by Geoff Livingston, a reader-supported publication. To receive new posts, consider becoming a free or paid subscriber.)