My 2025 AI Wrapped + 2026 Predictions

Hey everyone. Welcome back to the Prompt. Every January we see this wave of predictions for the new year. But this year I wanted to do something a little different. I built my own AI and used it to reflect on where we got things right, where we were early, and what to build next. But first, let's see how my 2025 predictions played out. The first prediction I made was that multimodal AI would take over. That prediction held up well. The global multimodal AI market is projected to hit nearly $11 billion by 2030, growing at 36.8% annually. Some forecasts even predict 27 billion by 2034. Apps now expect models that can see, listen and read at the same time. My prediction about personalized AI assistants becoming agents, that's happening now. 2025 AI trends emphasize agentic multimodal assistance as the main shift. We're talking autonomous or semi autonomous AI agents that plan and execute multi step tasks across systems. Major cloud platforms call them AI co workers now. They're orchestrating workflows end to end, not just answering questions on regulatory maturity. I was partially right. The EU AI act is now in force with detailed requirements for high risk and medical AI systems. The US has executive orders and agency guidance on safety and transparency, but we still don't have one unified global framework. It's a rapidly maturing but fragmented landscape with multiple regional and sector specific regimes. The one prediction that hasn't fully materialized yet Autonomous medical diagnosis approvals. Hundreds of AI medical devices have been cleared by regulators like the fda, especially in radiology, but most still function as clinical decision support under physician oversight. Fully autonomous diagnostic AI without a clinician in the loop, that's still on watch lists rather than widely deployed. So that brings me to my AI for 2025. I coded my own themes for the year and here are the top three that shaped my year and are steering the ship into 2026. Firstly, I used AI as core business infrastructure. I stopped treating AI as an occasional chat box and started embedding it into my operating system. For me, AI as infrastructure means I design for data flows and systems, not for single prompts. Instead of I'll ask ChatGpt when I'm stuck. I build persistent workflows, lead qualifying agents, research pipelines, auto tagging and summarization of calls, content engines and more thinking how future me will use them. This shift has been huge in my own work because it compounds every new process I work on becomes reusable intellectual infrastructure. The more I build, the faster everything else gets. The second theme, agents as the new workers agentic AI systems became my digital team members that can plan, decide and act across tools and data, not just respond in a chat. Agents let me collapse entire workflows into a single intent. They read, research, call APIs, generate assets, and update systems. That moves me from faster humans to fewer humans per unit output, which is the real economic shock I care about. Thinking in terms of agents forces me to define roles and outcomes, not prompts. Instead of help me write X, I design an agent whose job is to monitor a pipeline, pull in data, draft content, and trigger human review at the right points. Most leaders I support will never use a model directly. They'll use an agent that owns a slice of their business process. That's why I keep saying agents are the new workers. It reframes AI as headcount leverage, not just text assistants. The the third theme was around owning distribution beyond classic SEO. What I've seen is that the shift in organic traffic isn't about SEO is dead. It's about the middleman is now an AI. Organic search and social are being rewritten by AI Overviews and feed dynamics. My only safe strategy is to own more of my distribution AI overviews dramatically reduce click through on a lot of queries. Relying on old search patterns is dangerous now. That's why when I talk about future proofing, I push towards systems that capture attention, convert it to subscribers or members, and then nurture through channels I control. AI is a force multiplier here. I use it to summarize, repackage and personalize content across email, podcasts, blogs, and social without burning out. Rather than chasing whatever new SEO hack is trending, I'm focused on building robust content machines that keep delivering even as interfaces change. The new mindset you have to accept is both defensive, protecting the downside and offensive, creating more consistent growth. Through my own projects, I've seen how far this can go. But tools alone don't create outcomes. Most people stall on implementation, integration and strategy. That's why I keep building and emphasizing community. I want a place where we actually wire this into our stacks together. For me, that's the real leverage story. Not just what the tools can do on paper, but what a small, committed operator can do when surrounded by the right patterns, support, and peers. Which brings me to my predictions for 2026. I expect us to see agents become teammates. Most teams run a small agent stack wired into their ops. One for research and monitoring, one for inbox and customer comms, one for docs and contracts, one for ops and light purchasing, and one for Data qa. They coordinate through automated workflows and handoff context, not just outputs think agent to agent briefings, memory of past decisions and SOP aware execution. They'll sit inside the tools you already use, not separate dashboards. You'll measure them like staff SLAs accuracy, cost per task, exception rates and percent hands off. Expect 50 to 70% of repetitive tasks to move to agents with human supervising exceptions and judgment calls. The big change is adaptation. Agents improve after each run, compressing cycle time week over week. Search will shift from information to action. It won't just compile information, it will complete the task. AI overviews sit on top, tie to real time data and trigger actions across apps. Inside browsers and work apps, you'll see an AI mode that reads the page, gathers sources, drafts the artifact and executes next steps without tab hopping. For teams, that means living briefs, auto updating market maps and inbox first insights. For consumers, it's plan book pay in a single thread. Expect fewer clicks, more inline completions and stronger citations with source controls. This will have a big impact on content strategy. You optimize for being quotable, structured and verifiable, so AI overviews pick you up. Deep research will become an everyday muscle. Autonomous research runs in hours instead of weeks. You'll kick off a prompt like map the top 30 players, pricing moats and risks and wake up to a sourced, deduplicated report with pull quotes, charts and a point of view. Ongoing monitors keep it fresh. Daily strategy cycles compress from quarterly to bi weekly. In practice, founders and operators run three tracks. The first is market intel who moved, who raised, who shipped. Second will be customer truth. So things like live reviews, calls and summarizing support logs. The last track will be internal performance. We're talking cross checked sources, hallucination detection and clear confidence ratings. We can expect focused robotics in logistics, retail and healthcare think inventory scanning, shelf checks, restocking lanes, basic pick pack, specimen transport and scheduling optimization. It won't be humanoid hype. It's fit for purpose. Systems tied to software agents that coordinate jobs, verify completion and escalate anomalies. In a world now where production of content is becoming faster, cheaper and more comprehensive than ever before, I expect to see budgets shift from creation to promotion. Think of how quick and cheap it is now to create a digital ad. Since companies are saving so much on the front end, they they'll redeploy that to the promotion of it. I see this affecting movie studios the most. Production budgets will shift to marketing budgets and we'll see bigger and more wild creative outputs to capture attention in ways we can't yet dream of. These predictions aren't just about technology. They're about how we work, how we build, and how we create value in a world where AI is everywhere. The companies and individuals who win won't be the ones with the best models. They'll be the ones who wire AI into their operations, measure what matters, and build resilient distribution channels. If you want to go deeper on any of this and gain a Competitive edge in 2026, come join me in my Slack community Control Alt build. It's where I share everything I'm building, testing and learning in real time. We're a community of builders actually implementing this stuff, not just talking about it. I'll drop the link in the show notes. I'd love to see you in there. Thanks for listening. Here's to making 2026 the year we stop talking about AI's potential and and start capturing its value. I'll see you in the next episode.