It’s fair to argue that for the past few years, artificial intelligence has largely been in an experimental phase. But the focus for 2026 will likely be less about novelty and more about real-world application.
Ask anyone who’s interested in AI (or, indeed, AI itself) what the trends for this year will be, and you’re likely to get a lot of predictions.
Here are four that make the most sense to us at Flow Communications.
Trend 1: integrated and multimodal generative AI
This trend’s a mouthful, but it basically refers to embedding AI into core business workflows – think Google Workspace, Microsoft 365, or Adobe Creative Cloud – rather than treating it as a standalone tool, as has been the case with AIs such as ChatGPT or Claude.
It’s also about the seamless fusion of tasks, with a single AI platform executing all the elements of a project. For example, for a marketing campaign, it will put together the strategy, produce email and social media content, create imagery and draft a video script.
Interestingly, this trend includes the rise of the small language model (SLM): where AIs such as ChatGPT or Gemini are large language models, giant generative AIs used for everything (and not always with optimal results), we’ll now see companies adopt industry-specific, specialist SLMs.
For instance, a law firm will use a law-specific SLM, or a doctor will use a medical diagnostics SLM, presumably leading to more accurate and relevant results.
Trend 2: autonomous AI agents
PAs and interns, beware: this is the next evolution of AI assistants, and it is starkly different from simple workflow automation. An AI agent will be able to receive a complex instruction, break it down into steps and carry them out.
For example, you could instruct the AI agent: “Find the top three suppliers for this widget in South Africa, contact them for quotes and then schedule a meeting with the most promising one.”
This upgraded AI agent promises a new layer of productivity. Proponents argue that it equates to a team of ultra-efficient junior analysts or executive assistants, available on demand, which frees up human workers for more strategic tasks.
Trend 3: from pixels to physics
This is where AI meets the Internet of Things, with AI moving out of the cloud and into physical devices. It’s no longer about data analysis, but rather real-time control.
AI in self-driving cars is the most obvious example, but here are a few more: smart factories where AI predicts machine failures before they happen; AI systems that manage a city’s power grid, or traffic flow, for maximum efficiency; AI mapping out a farmer’s irrigation plan down to square metres; or warehouse robots that no longer just follow lines on the floor, but make intelligent decisions to navigate complex spaces.
Trend 4: trust and safety
This one’s arguably the most important for both AI users and their customers. Regulation is finally starting to respond to the explosion of AI and some major economies are implementing AI laws (such as the European Union’s AI Act), frameworks, guidelines or rules.
AI compliance is thus moving from a theoretical, ethical discussion to a practical, legal and financial issue, in which non-compliance could result in heavy fines and devastating reputational damage.
From a liability perspective, users will increasingly need to deploy explainable AI systems: ones that can clearly articulate why they make certain life-changing decisions. Imagine, for example, that doctor using AI for diagnostic purposes, or a bank that employs AI to approve or deny loans.
The most trusted brands will be those that can prove their AI is compliant: secure, fair (in other words, unbiased) and transparent about its data provenance. Building this trust with customers will surely unlock a major competitive advantage for brands.
Of course, AI remains a nascent technology, and we will continue to see it grow and be harnessed in ways that we can barely conceive of now. But the days of playing around with AI and marvelling at how well it can do things (or not) are numbered.
The year 2026 may well be a watershed year for how we practically use AI – and how we govern that use.
Willem Steenkamp is a senior writer and editor at Flow Communications (www.flowsa.com ), one of South Africa’s leading marketing and communications agencies. Founded in 2005 in a small spare bedroom, Flow now has a permanent team of 60 professional staff, with more than 1000 years of collective experience in communications.
