AI Predictions for 2025

Published January 3-12, 2025

2024

As we enter a new year, it’s the perfect time to look ahead and make predictions about what lies on the horizon. In this essay, I’ve gathered a range of forecasts on how AI may evolve by 2025, drawing from multiple experts and reliable sources. Though these predictions are speculative, they aim to prompt us to reflect on the future and consider how it might shape the way we think and work.

But before we get to 2025, lets consider the key developments in AI in the past year.

I asked Perplexity for the top 10 AI developments in 2024:

OpenAI’s o1 Model: Advanced reasoning capabilities in mathematics and coding.
Agentic AI: A groundbreaking development offering unprecedented autonomy in decision-making, capable of analyzing complex situations and executing multi-step tasks with minimal human intervention.
AI Search Technologies: Large language models transforming information retrieval, potentially disrupting traditional search engine models.
Google DeepMind’s GenCast: Enhanced weather prediction technology to optimize agriculture and disaster preparedness
Microsoft’s Copilot Vision: AI system designed to enhance productivity by integrating with digital environments.
Advanced AI Video Creation Tools: Innovative platforms like Google’s Veo and OpenAI’s Sora revolutionizing content creation.
Anthropic’s Claude Chatbot: Focusing on AI safety and reliability for critical applications like disaster response.
Conversational AI Advancements: More natural, human-like interactions with significant improvements in understanding complex queries.
Quantum AI: Utilizing quantum computing to enhance AI algorithms and solve complex computational problems.
Generative AI for Hyper-Personalization: AI systems capable of creating highly personalized content, products, and experiences across various sectors.

Forbes:

The continuing evolution and advancement of Generative AI chatbots
Apple enters the Generative AI arena with Apple Intelligence
Head of Google AI wins the Nobel Prize for Chemistry
The EU AI act – AI legislation begins to take shape
Optimus breaks new ground for humanoid robots

VentureBeat:

OpenAI expanded far and wide beyond ChatGPT
Open-source AI took off
Google’s Gemini series became a serious contender for the best available
Agentic AI swept the enterprise

Mint on AI innovations in 2024:

OpenAI’s Sora Turbo enhances AI video generation
Google updates Veo and Imagen models
Microsoft introduces autonomous AI agents
Claude AI adds custom writing styles
xAI unveils Grok-2 models

Ashu Garg: “For me, the story of 2024 in technology can be summed up in a single number: 1000x. That’s the factor by which the cost of machine intelligence has fallen in just three years – from $60 per million tokens with GPT-3 in 2021 to $0.06 with Meta’s Llama 3.2. To my knowledge, this represents the most rapid democratization of any technological capability in human history. Intelligence, once humanity’s most precious and scarce resource, is becoming ubiquitous, abundant, and essentially free.”

Myra Roldan: “2024 was the year AI got a little glow-up. Multimodal AI stole the spotlight, combining text, audio, and visuals into single models that can do it all, maybe not well all the time but good enough. Then we had the quiet rise of Small Language Models (SLMs) — an attempt at launching little powerhouses that can run on smaller devices with stripped features. And finally, Customizable Generative AI became the must-have trend, as businesses attempted to get it together and started to lean into tailored AI systems.”

A list from appinventiv:

Conversational AI
Predictive Analytics
AI Democratization
Ethical and Explainable AI
Multi-Modal AI
Digital Twins
CoBots
Cybersecurity
Generative AI
Shadow AI
Agentic AI
Retrieval-Augmented Generation (RAG)
Sentimental AI
Quantum AI

Top 10 AI tools:

ChatGPT – 50%
Bing AI – 15%
Google Gemini – 12%
Claude – 8%
MidJourney – 5%
Hugging Face – 4%
Jasper – 2%
Janitor AI – 1.5%
Civitai – 1.3%
Perplexity AI – 1.2%

Simon Willison writes about LLMs in 2024:

The GPT-4 barrier was comprehensively broken
LLM prices crashed, thanks to competition and increased efficiency
Multimodal vision is common, audio and video are starting to emerge
Voice and live camera mode are science fiction come to life
Prompt driven app generation is a commodity already
“Agents” still haven’t really happened yet
The rise of inference-scaling “reasoning” models

General – 1

Ben Evans in his presentation:

2025 predictions via Perplexity:

AI-native applications will revolutionize technology, with breakthrough companies reaching $50M in annual recurring revenue and transforming how businesses leverage intelligent systems.
Autonomous AI agents will become dramatically more sophisticated, capable of executing complex multi-step tasks with minimal human intervention and potentially beginning to design and improve other AI systems.
The global economic impact of AI will be staggering, with projections indicating contributions of $200-$340 billion annually in sectors like banking, finance, and technology.
Major AI model providers like OpenAI, Anthropic, and Meta will continue aggressive investments, with more restrictive licensing and a shift towards building practical, industry-specific applications.
Quantum AI and advanced computational technologies will push the boundaries of machine learning, enabling unprecedented problem-solving capabilities across scientific and industrial domains.
Content generation will see explosive growth, with AI potentially producing entire research papers, generating high-quality video content, and creating increasingly nuanced creative works.
Cybersecurity will become both more vulnerable and more resilient, with AI simultaneously presenting significant threats and offering advanced protective technologies.
The regulatory landscape for AI will remain complex, with governments struggling to create comprehensive frameworks that balance innovation and potential risks.
Workforce dynamics will dramatically transform, with AI augmenting human capabilities and creating new job categories while rendering some traditional roles obsolete.
Decentralized AI systems will emerge, focusing on enhanced authentication, predictive analytics, and more personalized technological interactions across multiple platforms.

FT asks (and answers) 4 questions:

Will AI development hit a wall? The most promising developments look like those coming from models that carry out a series of steps before returning an answer, allowing them to query and refine their first responses to deliver more “reasoned” results. It is debatable whether this is really comparable to human reasoning, but systems such as OpenAI’s o3 still look like the most interesting advance since the emergence of AI chatbots.
Will AI’s ‘killer app’ emerge? [2025] is likely to bring the first demonstrations of apps that can intervene more directly: absorbing all your digital information and learning from your actions so that they can act as virtual memory banks or take over entire aspects of your life…Instead of true killer apps for AI, this means we will be left in the “AI in everything” world that technology users have already become accustomed to
Will Nvidia’s GPUs still rule the tech world? Even as its market share starts to erode, though, Nvidia’s software still represents a considerable moat for its business, and by the end of the year it should be on the verge of another important new product cycle.
Will the stock market’s AI boom continue?With Big Tech in the midst of an AI race that its leaders believe will determine the future shape of their industry, one of the main forces behind the AI capital spending boom will remain in place.

Observer:

Agentic A.I. will be “the next giant breakthrough”
Test-time compute could be a solution to A.I.’s training data crisis
Synthetic data is another promising solution
“Large world models” will create 3D A.I. worlds
I. search engines will reshape online search

IBM:

Agentic AI
Inference Time Compute
Very Large Models
Very Small Models
More Advanced Use Cases
Near Infinite Memory
Human-in-the-Loop Augmentation

General – 2

Ashu Garg:

The limits of pretraining will drive new AI breakthroughs. The assumed progress of scaled pretraining has hit three walls: data, compute/energy, and model architectures. But, in 2025, these walls won’t limit AI’s advance – they’ll redirect it toward new frontiers. One of the most promising frontiers is “reasoning” – where models don’t just recall patterns from training but actively work through problems during inference. Take OpenAI’s o3 model: rather than producing instant answers, it generates detailed reasoning paths tailored to each task, much like a mathematician methodically working through a proof
AI will rewrite the rules of software economics. 2025 will see AI companies break free from traditional software budgets as they target the vastly larger services market – a roughly 10x expansion in TAM. They’ll succeed by selling actual work completion rather than just workflow enablement.
AI interfaces will evolve beyond the chatbox. By the end of 2025, the simple chatbox that defined early AI products will feel as dated as the command line. We’ll see specialized UIs emerge for different kinds of work: interactive dashboards for monitoring AI processes, visual tools that make AI reasoning transparent and debuggable, and intuitive interfaces for creative collaboration. These new interfaces will acknowledge that AI isn’t just a back-and-forth question-answering system, but a complex tool we need to guide, monitor, and collaborate with in more advanced ways.

Fello AI summarises predictions by AI leaders:

AI Agents: In 2025, expect them to handle anything from booking your vacations to handling financial reports, all on their own. Companies like IBM and OpenAI believe we’ll soon see AI agents so advanced, they’ll quickly become essential in fields like marketing, design, programming, and more.
AI Models That “Think” First: New models will spend more time reasoning—like a person pausing to think something through—before giving you a response…The reason this matters is simple: you won’t need to keep retraining a model just to get better answers. Instead, the AI can use extra “thinking time” during inference (when it’s generating responses) to refine and improve its logic on the fly. This might make your chatbots feel a lot closer to genuinely intelligent helpers.
Near-Infinite Context: More and more AI systems will have context windows in the millions of tokens—some experts even joke about “infinite memory.” Imagine a customer support chatbot that remembers every conversation you’ve ever had. Next time you ask a question, it knows your entire history, saving you from endlessly repeating yourself. That’s the power of near-infinite context… The result? More personalized recommendations, faster solutions, and fewer mistakes.
Physical AI Robots: Picture factories full of robots that can make decisions on the fly. They won’t just pick up boxes; they’ll also figure out the best route to move them and adjust if something changes… Over time, expect to see these AI-driven bots in warehouses, healthcare facilities, and maybe even your local supermarket. They could speed up supply chains, assist in delicate surgeries, or handle any job that’s repetitive or risky.

Rafe Brena:

AI Agents will be everywhere
Pure scaling is done
Higher reliability brings trust
Products and industry-specific AI will win over general chatbots
Self-driving robotaxis become mainstream

Forbes (with inputs from founders):

AGI arrives early
AI agents take charge
Voice becomes default
Video goes mainstream
AI gets embedded everywhere
Wearables and retail go smart
One person, billion dollar company

VCs

Sapphire Ventures:

AI-native apps will see the strongest funding momentum
Many more AI-native companies will reach $50M in ARR
AI exits will increase, but M&As will trump IPOs
Models will improve across multiple dimensions
Agents will begin to deliver on hype, though impact will be uneven
The DoD will double down on AI
AI-generated content to surge with video becoming a rising star
As consumption grows, outcome-based AI pricing models will be slow to ramp
AI: A major security threat and antidote
AI regulation will move slowly absent a major calamity

Radical Ventures:

Meta will begin charging for use of its Llama models.
Scaling laws will be discovered and exploited in areas beyond text—in particular, in robotics and biology.
Donald Trump and Elon Musk will have a messy falling-out. This will have meaningful consequences for the world of AI.
Web agents will go mainstream, becoming the next major killer application in consumer AI.
Multiple serious efforts to put AI data centers in space will take shape.
An AI system will pass the “Turing test for speech.”
Major progress will be made on building AI systems that can themselves autonomously build better AI systems.
OpenAI, Anthropic and other frontier labs will begin “moving up the stack,” increasingly shifting their strategic focus to building applications.
As Klarna prepares for a 2025 IPO, the company’s claims about its use of AI will come under scrutiny and prove to be wildly overstated.
The first real AI safety incident will occur.

Tomas Tunguz’s AI-related predictions:

Google continues its surge in AI: They lept from no placement to top 1 or 2 on the OpenRouter rankings. They further advance their market share. Grok benefits from Elon’s position in government to become a viable contender with OpenAI & Anthropic.
Voice becomes a dominant interface with AI as speech models are pushed on device & the accuracy/latency astounds. Voice produces text, image, & video. Why type? It’s the start of a generation of people who will never learn to type on a keyboard.
The first $100M ARR company with ≤30 employees is created. An AI native product coupled to an AI native team produces incredible market cap creation efficiency.
Data center spending by hyperscalers eclipses $125b for the year as the AI race fuels demand for GPUs. Broadcom is the hottest semiconductor stock of the year.

Race Capital:

The Birth of AI Embedded Application Services & Finally the Formation of a Proper GenAI Development Model for Applications: In 2025, we will see the beginning of relying on AI Agents to assist our current application development, deployment and run time models, following by a modern day fully AI embedded Application Services for the development of “formless” applications – allowing users to define processes by natural language, fine tune and self improving of processes, AI driven decision making and fully autonomous actions. Microsoft will likely be the forerunner of such development and follow up multiple large and small independent players. We will then see security, RAG, small model AI, data lake technologies to repurposed into the new development and deployment model.
Enterprise Search Reimagined: Similar to what happened to consumer search when LLM emerges – who doesn’t like an answer vs looking for an answer from a page ranked laundry list? We will see the emergence of a whole new enterprise specific (data sovereignty, security and governance proofed) enterprise search engine emerge. These search engines will work across platforms, disparate applications and data sources. Just like consumer search, Enterprise AI search will become the most commonly used tool in the work environment and most application functions will be demarked from these search results.
The AI Arms Race: Open Source Revolution Reshapes Global Power Dynamics – Open source AI is rapidly catching up to closed source giants, with models like Mistral AI’s M-o-E, Meta’s Llama 3 and Alibaba’s Qwen challenging OpenAI and Anthropic’s dominance. This shift is democratizing AI and reshaping the international competitive landscape, with China emerging as the unexpected frontrunner as they have leaped ahead in the open-source AI race.
Say Hello to Long Term Memory for LLM Architectures: The current prevalent architecture of LLM is RAG + context tokens. While RAG is the ability to access relevant data and context tokens are short term memory for the model, there is currently a lack of long term memory in models. Long term memory could come in many forms including updating model parameters on the fly to learn from incoming data much like how a brain would learn or other approaches. The focus will shift to creating smaller, more efficient models that can run on edge devices without requiring massive computational resources. Multimodal systems (e.g., combining text, images, and video) will see widespread use in applications like e-commerce and entertainment.

Techcrunch:

Paul Drews, managing partner, Salesforce Ventures: Essentially all enterprise workflows can be optimized with AI — especially agentic AI. We’re seeing real demand for AI and ML tools that can make underlying models 50% more efficient while delivering improved results. AI is experiencing froth, but from a larger market perspective (not just Silicon Valley), AI is still new and everyone is trying to figure out how to use it, price it, and purchase it.
Mark Rostick, vice president and senior managing director, Intel Capital: For the moment, it is clearly easier to adopt AI through application vendors than trying to build your own platform given that the market for enterprise platform tools is still very, very fragmented. I do think there is pent-up demand for some sort of platform solution, so I believe we’ll see many founders trying to address that problem this coming year.
Raviraj Jain, partner, Lightspeed Venture Partners: It’s a consensus view but AI adoption will continue to accelerate in 2025 as (1) model capabilities improve, (2) enabling infrastructure is built out, and (3) stronger AI-first products come to market.

More – 1

Economist: “Start at the cutting edge of innovation. Several constraints are slowing the pace at which the technological frontier is being pushed out. Training big models needs huge amounts of energy. The electricity used to train GPT-4, the large language model underpinning ChatGPT, could have powered 5,000 American homes for a year; the equivalent figure for GPT-3, its predecessor, was 100. Developing ever larger and whizzier models thus requires ever deeper pockets. By some estimates, the next generation of models could cost $1bn to train; and the larger they become, the more the cost of querying them (known as “inference”) will mount. Meanwhile, there is a looming shortage of training data. By one estimate, the stock of high-quality textual data on the internet will have run out by 2028. Companies around the world are rushing to come up with clever fixes to these problems, from more efficient and specialised chips to more specialised and smaller models that need less power. Others are dreaming up ways of tapping new high-quality data sources such as textbooks, or generating synthetic data, for use in training. Whether this will lead to incremental improvements in the technology, or make the next big leap forward affordable and feasible, is still unclear. Investors have poured money into superstar firms like OpenAI. But in practice there is not much difference in performance and capabilities between the flagship models offered by OpenAI, Anthropic and Google. And other firms including Meta, Mistral and xAI are close behind.”

More: “The use of agents to advance from “chatting to doing” could be one of the big tech breakthroughs of 2025, says Alex Wang of Scale, an AI data company… Several factors, however, make it harder to create agents than chatbots. One is data. Unlike chatbots, which scrape information from the web to answer questions, agents require data on the way tasks are performed, including the sequencing of actions and the reasoning behind them…A second problem is trust. Checking whether a chatbot has given a right or wrong answer is usually easy. Determining whether an AI agent has booked the best restaurant or holiday it could within your budget may be more difficult…A final problem is cost. In order to reason, plan and solve problems on behalf of users, AI agents need access to models that can handle complex tasks.”

Myra Roldan: “2025 will be the year AI goes from experimental to essential. It’s no longer about whether you’ll use AI but how you’ll integrate it seamlessly into every aspect of work and life.”

Edge AI LLMs in Your Pocket: Large language models (LLMs) will be compressed to make them super small so it can run on smaller devices like phones (these are edge devices — devices that can run without always having to connect to the internet). The trade off will be accuracy, speed, and efficiency.
Autonomous AI Agents Take Over: These agents won’t just assist; they might manage, execute, and even troubleshoot entire projects. From coordinating supply chains to running customer service operations, they’re about to become ubiquitous.
AI Courses in Every Discipline: By 2025, AI education will infiltrate traditional academic programs across disciplines.

Decagon cofounder and CEO Jesse Zhang on a coming shift in software pricing: “Our view on this is that, in the past, software is based per seat because it’s roughly scaled based on the number of people that can take advantage of the software.“With most AI agents, the value . . . doesn’t really scale in terms of the number of people that are maintaining it; it’s just the amount of work output. . . . The pricing that you want to provide has to be a model where the more work you do, the more that gets paid. “So for us, there’s two obvious ways to do that: you can pay per conversation, or you can pay per resolution. One fun learning for us has been that most people have opted into the per-conversation model . . . It just creates a lot more simplicity and predictability.”

Sean Rankin:

Models shift from Pre-Training to Post-Training
Test Time-Compute Becomes the primary Paradigm
Reasoning and Inference (Chips) Take Centre Stage
Agents are Finally Unleashed
Consolidation of early GenAI Companies and New Biz Models Emerge

Google:

Multimodal AI will deliver more context
AI agents will simplify complex tasks
Enterprise search will give people the knowledge they need
AI-powered customer experiences will get even better: “AI-powered solutions are changing the customer experience, in part by anticipating customer needs and helping businesses stay connected to their customers. This translates to increased revenue, efficiency and brand loyalty. Imagine personalized recommendations and AI-enhanced search that understands customer intent. In retail, AI will be able to create personalized shopping experiences and customer support across all shopping channels — online, in store or mobile. Manufacturers will use AI to improve production and customer service.”
AI will enhance security systems

MuleSoft writes about the rise of LAMs (Large Action Models): “Move over LLMs – there’s a new acronym in town. While large language models (LLMs) excel at understanding and generating human language, large action models (LAMs) take it further by translating that understanding into action. Think of it this way: LLMs are like the brains of the operation, capable of processing information and generating insights. LAMs are the hands, putting those insights into action. LAMs are poised to be the driving force behind the success of agentic AI, which focuses on creating AI agents that can autonomously perform tasks and achieve goals. By integrating with APIs across the enterprise, LAMs can interact with various systems and applications, enabling them to carry out complex actions. For example, an LAM-powered AI agent could automate the onboarding process for new employees. It could gather the necessary information, create accounts, grant access to relevant systems, and even schedule introductory meetings.”

More – 2

Russ Altman (Stanford University): “General Contractor’ LLMs: We will start seeing complex systems for problem solving that are made out of a bunch of AI systems that talk to each other. For example, picture a series of large language models with specific expertise (fine-tuning) combining to solve problems. In some cases, they may be negotiating with one another; in other cases, they will hand off tasks to “expert LLMs” which will return answers. And so there will be a kind of “general contractor” LLM that deals with the human customers and it will subcontract some of its problem solving to other agents that have expertise. Where these arise first may include complex simulations, health decision-making, financial arrangements, or educational programs.”

James Zou (Stanford University): “AI Agents Work Together: In 2025, we will see a significant shift from relying on individual AI models to using systems where multiple AI agents of diverse expertise work together. As an example, we recently introduced the Virtual Lab, where a professor AI agent leads a team of AI scientist agents (e.g., AI chemist, AI biologist) to tackle challenging, open-ended research, with a human researcher providing high-level feedback. By leveraging the multidisciplinary expertise of different agents, the Virtual Lab successfully designed new nanobodies that we validated as effective binders to recent SARS-CoV-2 variants. Looking ahead, I predict that many high-impact applications will use such teams of AI agents, which are more reliable and effective than a single model. I’m particularly excited about the potential of hybrid collaborative teams where a human leads a group of diverse AI agents.”

Will Bryk:

The o3 class models are reeeaally good at optimizing for anything you can define a reward function for. Math and coding are pretty easy to design a reward function for. Fiction writing is harder. So that means in the short term (1 year), we’re going to get spiky models. They’re going to be basically AGI-level at math and coding and general reasoning but write generic fiction.
Agents really are coming in 2025. There’s no way o3-like models won’t be able to navigate the browser/apps and take actions. That stuff is easy to design reward models for. It’s also a huge market — automating computer work — so there’s big incentives for the labs that need to justify their big spend. I’d guess by December 2025 you’ll be able to tell your computer to do any sort of workflow that involves navigating webpages/apps and moving data around.
What about us software engineers? In the short-term it’s going to be heaven. Every SWE just got a promotion to tech lead, nicely done. For those who fully adopt LLMs, coding by end of 2025 will feel more like orchestrating a bunch of small jobs that little agents go and perform. Any PR that has very clear specification should be doable by an o4 system with an error rate that’s small enough to be acceptable.

Sandy Carter: “Rise of AI-Native Businesses: A new breed of company will emerge, built from the ground up around AI capabilities. These organizations will use AI for everything from strategic planning to daily operations, creating new standards for business efficiency and adaptability. Key characteristics will include: automated decision-making systems that handle 80% of routine business operations, AI-driven talent acquisition and development programs that predict employee success with 90% accuracy, real-time market analysis and strategy adjustment capabilities, [and] personalized customer experiences that adapt instantaneously to behavioral changes. The disruption won’t come from legacy businesses retrofitting AI into existing processes but from startups that fully leverage AI at their conception. These AI-native businesses will rethink industries by building their entire structure around AI’s transformative potential.”

Platfomer:

AI will show continuous incremental improvement, but no exponential leaps. The bad news may be that the low-hanging fruit of the scaling laws is gone; the good news is that the platforms have barely even begun to explore how to turn LLMs into top-notch products. If 2024 was the year of the research team, 2025 will be the year of the product team.
The first year of “the agentic era” mostly disappoints. The first AI agents began to come into view at the end of this year. All are in the early experimental stages, and none are the perfect AI assistant of our dreams. The best agents will do some impressive stuff, but slowly and inconsistently. By the end of 2025, most people will not be regularly using them.
The big AI companies remain competitive with each other. With talent seemingly more or less distributed evenly among the biggest players, and everyone trying more or less the same techniques to improve their models, no one runs away with the game in 2025.

Barr Moses: “The unstructured data stack will emerge. The idea of leveraging unstructured data in production isn’t new by any means — but in the age of AI, unstructured data has taken on a whole new role. According to a report by IDC only about half of an organization’s unstructured data is currently being analyzed. All that is about to change. When it comes to generative AI, enterprise success depends largely on the panoply of unstructured data that’s used to train, fine-tune, and augment it. As more organizations look to operationalize AI for enterprise use cases, enthusiasm for unstructured data — and the burgeoning “unstructured data stack” — will continue to grow as well.”

More – 3

CIO has 12 AI predictions for 2025. Among them:

Multi Agent Systems: AI agents are interesting. But things are going to get really interesting when agents start talking to each other, says Babak Hodjat, CTO of AI at Cognizant. It won’t happen overnight, of course, and companies will need to be careful that these agentic systems don’t go off the rails….“If you have one agent and tell it to do everything in the sales department, it’ll fail a lot. But if you have lots of agents, and give them smaller responsibilities, you’ll see more work being automated.”
Mass customization of enterprise software: ““Right now, people are all using the same version of Teams or Slack or what have you,” says Ernst & Young’s Malhotra. “Microsoft can’t make a custom version just for me.” But once AI begins to accelerate the speed of software development while reducing costs, it starts to become much more feasible. “Imagine an agent watching you work for a couple of weeks and designing a custom desktop just for you,” he says. “Companies build custom software all the time, but now AI is making this accessible to everyone. We’re going to start seeing it. Having the ability to get custom software made for me without having to hire someone to do it is awesome.””

FT: “Nvidia is betting on robotics as its next big driver of growth, as the world’s most valuable semiconductor company faces increasing competition in its core artificial intelligence chipmaking business…“The ChatGPT moment for physical AI and robotics is around the corner,” Deepu Talla, Nvidia’s vice-president of robotics, told the Financial Times, adding that he believes the market has reached a “tipping point”… Talla said a shift in the robotics market is being driven by two technological breakthroughs: the explosion of generative AI models and the ability to train robots on these foundational models using simulated environments. The latter has been a particularly significant development as it helps solve what roboticists call the “Sim-to-Real gap”, ensuring robots trained in virtual environments can operate effectively in the real world, he said.”

Forbes : “AI Integration With Robots: Bornet and Banafa both mentioned that AI will likely breakout from desktop and mobile devices — moving into mobile robots that we’ll be interacting with much more frequently in the real world. “Robots will no longer feel futuristic but everyday; wearables and smart devices will merge seamlessly with our environments, making AI a tactile and visible presence. These advancements will tackle ever more complex challenges, from global logistics to personal accessibility, laying the groundwork for a better, more equitable world,” wrote Bornet. “AI systems capable of processing and integrating diverse data types — text, images, audio and video — will achieve a significantly deeper level of contextual understanding. This will enable more human-like interactions and revolutionize applications in fields such as education, healthcare and entertainment. This progress will be fueled by advancements in computational power and the growing availability of large, multimodal datasets,” expounded Banafa.”

Jason Tamara Widjaj on AI Agents: “With the promise of agents, it is important to remember this is hardly the first time the world has attempted to create value from intelligent agents. The fundamentals of the technology has advanced, but it brings new issues for AI security and AI safety to solve, which takes time. Essentially, we have upped capability but traded one failure mode for another. We have gone from brittle, narrow, handcrafted workflows and tightly defined knowledge, to broader, probabilistic workflows based on orchestrating, stacking and chaining failure prone reasoning and classification. And for good measure giving them memory and tools. Like parenting a five year old learning to navigate the physical world, we could talk about how smart they will be in the future, but the key consideration today is working out how we communicate to them, what they are allowed to do, and what tools to keep out of their reach until they are sufficiently mature. And until then teaching people around them not to take them too seriously.” And this: “The most important models we can train are mental models, and the most important models to deploy are business models.”

Jeremy Kahn (Fortune): “If “agentic” was the AI buzzword of 2024, “reasoning” will be the term everyone is talking about in 2025. Reasoning models work differently than the previous generations of large language models (LLMs). They provide better answers if given more time to “think” about a prompt. But this upends the way both tech companies and their customers have traditionally thought about software. Software used to have a marginal cost that approached zero. Once programmed, you could make endless copies, at almost no cost, and they all performed identically. This is what gave software companies such great profit margins. It is also why corporations could buy a license for software and then know what the cost of that software would be for the year. With these new reasoning models, that’s not true anymore. Now the same AI model will provide different responses depending on how much money a company is willing to spend on a given response from the model (because money equals thinking time.) That will fundamentally alter the way tech companies need to think about selling their software and the way businesses budget for it.”

Marketing – 1

Perplexity:

Hyper-Personalization: Marketing will shift towards predictive and deeply personalized experiences powered by AI. Key aspects include:

Anticipating customer needs before they arise
Creating tailored recommendations based on comprehensive data analysis
Delivering real-time, dynamically adjusted content that feels intuitive

Marketing leaders predict AI will enable one-to-one interactions at scale, where every customer touchpoint feels uniquely crafted. As Roberto Gennaro from redtag.ca notes, interactions will become “more intuitive and tailored” through advanced AI platforms.

Operational Efficiency and Automation: AI will dramatically streamline marketing processes and internal workflows:

Optimize entire marketing campaigns in real-time
Automate precision marketing strategies
Create specialized AI agents to enhance team productivity

Marketers are positioning AI as a critical tool to “optimize efficiencies” and focus more on innovation and strategic brand storytelling. The goal is making AI work “smarter” by simplifying workflows while driving better outcomes.

Ciklum makes a point about enhanced personalisation across industries: “AI will continue to revolutionize customer experiences, as user behavior and preferences can be used to personalize services and products to a greater extent than ever before, driving forward emerging AI personalization trends. In conjunction with predictive analytics, businesses will be able to understand and anticipate customer needs in real-time better than ever before, and develop content, offers and strategies accordingly. Generative AI has its own part to play in this by automating responses and touchpoints throughout the customer journey. However, with only 41% of consumers saying they’re comfortable with AI being used for personalization, work will continue through 2025 to address these concerns and help customers acclimatize to the new technology. Building transparency into data usage policies, supported by clear and detailed consent options, will help build that trust and ensure users feel informed and confident about how their data is being used.”

Steve Rotter (DeepL): “AI will accelerate hyper-personalized, more consistent marketing. We live in a hyper-personalized world—custom coffee, made-to-order clothing, and on-demand news feeds. Brands are even now tailoring their marketing messages and language to every customer in their preferred language, style, and tone. However, consistency of language across all of these personalized streams is also central to successful marketing. Research shows that it boosts revenue by 20% or more. Achieving this consistency across borders and languages is tough. It requires not only linguistic translation but also cultural adaptation to ensure that messages resonate the right way. If advertisers and marketers don’t get this right, they’ll open themselves up to misunderstandings, wasted resources, and missed growth opportunities. 2025 will be an exciting year for the marketing world as we start seeing better understanding of how AI can help strengthen customer relationships and businesses’ bottom lines.”

FreshEgg: “AI-driven automation is helping marketers streamline campaign management like never before, saving time and supercharging campaign performance. From generating ad creatives to optimising in-flight campaigns, marketers have access to tools that save countless hours and improve ROI in real time. In the coming year, new AI tools will appear that monitor campaigns 24/7 and adjust bids, messaging, and targeting dynamically, all based on performance data. The innovation will allow marketing teams to focus on strategy, creativity, and client relationships. These tools will enable agencies to manage more extensive campaign portfolios with greater precision and less manual effort, delivering better client results. AI and ML are unlocking personalisation at an entirely new level. Predictive models analyse consumer behaviour to anticipate needs, delivering content that feels custom-made for each audience member. 2025 will see further innovation as we build and refine our suite of analytics tools to help us forecast what content resonates best. This level of hyper-personalisation will allow us to help our clients serve dynamic, highly targeted campaigns that drive conversions.”

Google Cloud: “In the near term, retailers are zeroing in on areas like customer service, marketing, and digital commerce, using gen AI search and agents to enhance existing human capability and skills. Customer service centers are deploying tools that are AI first; can automate call transcription, generate smart replies, and respond to common customer questions. In marketing functions, teams are integrating gen AI to help write briefs, brainstorm campaign concepts, and produce personalized brand content at greater scale. We also see these capabilities coming together in AI-powered customer experiences, powering personal shopping advisors, generating new product content, and creating engaging, human-like conversational interfaces to improve online shopping experiences.”

Martech.org: “Integrating causal AI: In the new year, AI won’t just be used to scale up analytics and content creation. AI will also be used to analyze scenarios and help make decisions. This is the category of AI tools referred to as “causal AI.” “Without question, in 2025 and beyond, much energy will be expended on integrating causal AI with generative AI and large language models,” said Mridula Rahmsdorf, CRO at IKASI, a causal AI company with tools for retailers and financial services. “Current machine learning models are still extremely useful across multiple disciplines and are scheduled for an upgrade in the coming year. Causal AI — integrated with these forms of AI — will greatly improve accuracy and enhance decision-making, particularly when decision-making involves multiple and seemingly conflicting indicators based on correlations rather than causal relationships.” Rahmsdorf adds: “Integration of causal AI will also boost generative AI’s reliability by giving it a deeper and broader grasp of how different factors interact and affect one another. As a result, look to generative AI to be more adept at presenting scenarios that reflect realistic outcomes, leading to more coherent and relevant results.”

Marketing – 2

Sitecore:

Composable architecture will accelerate AI adoption and fuel intelligent commerce. As Ola Linder of SQLI highlights, the role of AI is in making sense of the data generated through composable architectures. Linder believes 2025 will see AI expanding beyond content creation to unlock business insights. He shares, “Next year, AI will make significant strides, especially in everyday digital tools. While content creation has been the focus, the real potential lies in AI’s ability to analyze data, streamline operations, and generate business insights. Composable architectures will make this easier, unlocking endless possibilities for predictive analytics and smarter decision-making.” In a composable environment, data from various tools and systems becomes more accessible, and AI can help generate valuable insights. This will allow businesses to predict stock levels, optimize supply chains, and better understand customer trends, creating efficiencies that drive success across the entire organization. It will also enable brands to create a foundation for unlocking additional innovations like voice-enabled commerce, AR/VR, personalization and more.
Conversational commerce will redefine customer engagement and decision-making. Avanade’s Roy Capon predicts that the biggest impact in 2025 will be the integration of AI across multiple facets of digital commerce. Capon envisions a world where conversational commerce thrives, with AI-powered chatbots and virtual assistants taking on roles ranging from customer service to complete shopping experiences. As he explains, “Conversational commerce will continue to grow with AI-powered chatbots and virtual assistants handling everything from customer service to complete shopping experiences. Voice-enabled shopping will become more mainstream, making shopping more seamless, while traditional interfaces will become adaptive, changing in real-time based on interactions.” AI-driven visual search and augmented reality (AR) will further transform product discovery. Customers will be able to search for products using images and even try them in real-time using AR, accelerating the purchasing decision process and creating a more immersive shopping experience. Capon believes this will enhance personalization and engagement, leading to a more intelligent commerce ecosystem.

Pål Machulla:

The Search Revolution: From Queries to Conversations. The way people discover information is fundamentally changing. Traditional search engines, once the gateway to the internet, are facing their “Yellow Pages moment” — a transformation as significant as when paper directories gave way to digital search. Tools like Perplexity AI and ChatGPT are turning search from a query-response mechanism into an intelligent dialogue. Instead of users sifting through pages of results, they’re engaging in conversations that surface precisely what they need. This shift has profound implications for marketing. SEO strategies built around keywords and backlinks will give way to contextual relevance and conversational value. For brands, this means rethinking how they become discoverable. Content will need to be structured not just for ranking, but for conversation. The winners will be those who can engage naturally in this new dialogic ecosystem, providing value through interaction rather than just information.”
AI Decisioning Takes Center Stage. Marketing orchestration will increasingly be driven by AI decisioning systems. These systems will move beyond simple automation to true orchestration, making complex decisions about customer journeys, content delivery, and resource allocation. The key shift here is from rule-based to outcome-based marketing. Instead of defining specific customer journeys, marketers will define desired outcomes and let AI systems determine the optimal path for each customer.

Artificial Intelligence +: “Personalized AI-Driven Customer Experiences. Customer experience has become a critical differentiator for businesses, and AI is at the heart of personalization efforts. By 2025, companies will use AI to provide hyper-personalized experiences, leveraging data on consumer behavior, preferences, and purchase history to deliver relevant and timely interactions. For instance, e-commerce platforms will use AI to suggest products tailored to each shopper’s unique tastes, while streaming services will recommend shows based on viewing habits and emotional engagement. This level of personalization enhances customer satisfaction, fosters loyalty, and drives repeat business. In customer support, AI-driven chatbots and virtual assistants will handle routine inquiries with ease, offering instant, accurate responses. AI will also enable proactive engagement, where companies can anticipate customer needs before they arise, such as reminding customers about recurring purchases or offering timely discounts. This shift toward personalized interactions will redefine the customer experience, making each interaction meaningful and reinforcing brand loyalty.”

Pascal Malotti (Valtech): “Moving beyond basic personalization, AI will enable retailers to create truly immersive, hyper-tailored experiences that deepen customer connections. Today’s consumers seek more than just efficiency; they crave experiences that resonate emotionally. Retailers who can deliver these meaningful, personalized moments will earn lasting loyalty. This new year, retailers will lean on AI to craft hyper-personalized shopping journeys, drawing on real-time customer data to shape each interaction. This will go beyond product recommendations to create experiences that feel uniquely tailored to the individual — for instance, offering curated outfit suggestions based on past purchases and browsing behavior or sending a notification with a special discount on a frequently bought item as the customer enters the store. These thoughtful touches foster loyalty through connection rather than convenience.”

Summary and My Take

I asked Claude and ChatGPT to summarise the key 2025 trends and predictions.

AI Agents Become Mainstream Workhorses

Evolution from chatbots to autonomous task executors
Handling complex multi-step processes without human intervention
Applications ranging from personal assistance to enterprise operations
Focus on practical, measurable task completion rather than just conversation

New Model Architectures & Processing Paradigms

Shift from pure scaling to “test-time compute” and reasoning
Development of both more powerful LLMs and efficient smaller models
Emergence of “Large Action Models” (LAMs) focused on concrete actions
Near-infinite context windows enabling better long-term understanding

Multi-Agent Collaboration Systems

Teams of specialized AI agents working together on complex tasks
“General contractor” LLMs coordinating with expert subsystems
Enhanced problem-solving through diverse AI agent expertise
Applications in research, business, and creative domains

Voice & Multimodal Interfaces Dominate

Shift from text-based to voice-first interactions
Integration of text, images, and video in unified interfaces
On-device inference enabling faster, more natural interactions
More human-like and intuitive AI experiences

AI-Native Business Transformation

Rise of companies built fundamentally around AI capabilities
New pricing models based on work output rather than seats
Potential for “one-person, billion-dollar companies”
Dramatic increases in revenue per employee

Ubiquitous AI Integration

AI capabilities embedded in nearly every software platform
Transformation from add-on feature to core functionality
Real-time insights and automation across all business processes
Seamless integration into everyday tools and workflows

Hyper-Personalization at Scale

True one-to-one experiences across industries
Real-time adaptation to individual preferences
Predictive analytics anticipating user needs
Balance between personalization and brand consistency

My Take (focused on Marketing)

2025 will mark the convergence of three transformative AI technologies that fundamentally reshape how brands connect with customers. The combination of Agentic AI (through AI Co-Marketers), AI Twins, and true N=1 personalisation will finally solve marketing’s persistent “Not for Me” problem – where generic messaging fails to resonate with individual customers.

AI Twins will evolve beyond basic segmentation to create personal digital companions for each customer. Starting with Adtech Twins built from public data, progressing to Madtech Twins incorporating marketing insights, and culminating in individual Singular Twins (MyTwins), these AI replicas will enable unprecedented understanding of customer needs and preferences.

The AI Co-Marketer will serve as the orchestration layer, using Agentic AI to coordinate across these Twins and create what I call “Generative Journeys” – dynamic customer paths that adapt in real-time like Google Maps recalculating routes. This combination will enable true N=1 personalisation at scale, where every interaction feels personally crafted for each customer.

Most importantly, this trinity of AI innovations will transform marketing from mass communication to individual conversation. Rather than bombarding customers with generic messages, brands will engage in meaningful dialogue through AI-powered personal companions. The result? Higher engagement, better retention, and dramatically lower customer acquisition costs as brands shift from endless acquisition to building lasting, profitable relationships.

This AI-powered transformation of customer engagement lies at the heart of what I call “NeoMarketing” – a revolutionary paradigm that moves beyond mass messaging and repeated acquisition to create genuine one-to-one relationships for lifelong retention at scale. Stay tuned as I explore how this convergence of AI innovations will reshape marketing’s future.

Share this:

Like this: