There’s no way to talk about 2025 without AI being a central part of the conversation. In 2024, AI was still scary and exciting and new. Today, it’s no longer a question of whether or how quickly businesses will adopt AI. With more than half of companies with >5,000 employees leveraging AI capabilities and an additional 40% in the exploration phase, generative AI (GenAI) is here.
Growth always requires innovation. At Iron Horse, embracing and experimenting with transformative technologies is part of what we do. We help our B2B customers grow by optimizing tried-and-true tactics and discovering ways to leverage new technologies and services.
Although many in our company hold a healthy skepticism about how AI will and can be used, we went forward together with a spirit of exploration and I can honestly say that by the end of the year, every function is using AI in a way that enables them to work better and aligns with our business objectives. (You can hear more about what we did, and what worked and what didn’t in our January Coffee Break, Beyond the Hype: Building a Strategic Framework for AI Adoption.)
In 2025, B2B buyers and sellers will use GenAI to break free of the limitations of the tech we’ve been using for the last 35 years to be more strategic, targeted, and precise with dramatic effects on the market, and marketing, as a whole.
Here are 4 impacts of GenAI I’m anticipating in 2025—and what you can do to prepare.
The decline of search will have ripple effects on the entire B2B marketing and selling landscape.
The way buyers search is changing at an unbelievable rate. More than 250+ million people use ChatGPT weekly. Queries on Perplexity average ~10 words (compared to 2–3 for those using traditional search engines), and almost half of those searches lead to follow-up questions.
Buyers aren’t just using new tools for research, they are researching in a new way that is more in tune with how humans learn about and decide on solutions. While interacting with a computerized agent may feel counterintuitive or like too big of a learning curve to some, it’s actually closer to how we learned about and decided on solutions before we started Googling. We asked questions, digested answers, discussed use cases, etc.
I’m one of those users. I like that when I use an AI tool to research a solution, I don’t have to wade through a bunch of sponsored links presenting a single point of view. I can get both sides of the story, complete with links to sources where I can quickly validate the information. I’m making more educated decisions much faster.
What to watch for: With buyers shifting more of their engagement offsite, the ingredients of a good website will change. Pages and navigation that helps buyers quickly validate what they’ve learned may become the most important factor for human visitors, while AI will need a body of easily navigable information to respond to queries about features, integrations, and more. Companies that act quickly to optimize websites and content to be digested by AI rather than Google, and employ custom GPTs to help visitors engage in a more personalized manner will have a distinct advantage here.
What to watch out for: Traditional indicators of engagement—especially site traffic—will become much less meaningful. Companies that are still overindexed on these top of funnel KPIs will see a dramatic decline in the “success” of their marketing. Platform vendors that help marketers identify intent and buyer readiness are already exploring ways to adapt to this new buying motion and continue to provide marketers with signals that buyers are progressing on their journey, but a mindshift will be key to understanding and acting on the impact of their marketing and sales efforts.
AI agents will allow users to work in ways that are best for them, not the platforms.
Today’s martech platforms tend to be very rigid. That’s because their automation runs on a set of programmed rules—and those rules are designed to do a specific set of things, every time they run. When the data in these systems or human behavior doesn’t conform to the rules, the processes break. Most marketers barely use 20-40% of their platform capabilities and find it difficult to execute the kind of personalized approach they know today’s buyers want.
AI agents, on the other hand, are flexible, intelligent AI tools trained for a specialized task. Agents don’t require hard coded instructions and are designed to determine the best way to do the task—even learning and adapting from their mistakes. A series of single-focus agents can be strung together to perform a complex set of tasks. And, because they are AI, they’re able to synthesize data quickly, including data we couldn’t access before because it wasn’t normalized.
One of the most interesting ways this is playing out is in the ability to target buyers based on unstructured data. Imagine being able to glean insights from reading product reviews or Reddit threads and to create target lists or personalized outreach. This kind of research is too time-consuming for most SDRs to do really well, but companies like Clay and Common Room are already working on ways to use agents to mine unstructured data and convert it into the structured data that existing tools can use.
What to watch for: “The platform can’t do that” becomes a thing of the past with the right agentic system. No longer constrained by the functionality that vendors define, marketers will be able to mine more data, set up more intelligent automations, and finally translate the approaches in their brain into practice. Agentic systems will deliver on the promises martech has always made—but has been too rigid to deliver on.
What to watch out for: It’s not just the Davids that have an opportunity to break out and dominate because of Al. More ability to do things their way will enable marketers to be more scrappy and experimental, even in Goliath-sized enterprises. Organizations that encourage this kind of exploration and adoption stand to win the most.
Rise of voice and natural language will accelerate adoption and alter engagement.
One of the big draws of tools like ChatGPT has been that users can basically talk to it. I don’t need to filter my thoughts through a language that works for the platform. I can just type the way I think. For some users, that’s meant forming their prompts using sentences that include niceties like “please.”
Recently, AI tools have started talking back. NotebookLM can create a summary of sources you provide—from websites to white papers—as an “audio overview” that sounds like a podcast, complete with hosts that engage in human-like banter. Many providers are experimenting with AI-powered dynamic video for B2B use cases from humanizing their web presence to prospecting.
What to watch for: Voice will eclipse chat as the modality for interacting with content, allowing users to iterate on requests and get the information they are looking for in a more dynamic and consumable way. As humans get more comfortable conversing with AIs, many will use their GPTs as literal sounding boards to help think through ideas. (At least one member of our exec team does this now and it has revolutionized the way they work.)
What to watch out for: As dynamic, conversational consumption becomes the norm, the ability for the AI to easily navigate your website and content becomes even more important. Surface-level product information will not cut it. Businesses with a body of accessible, audience-centric content that anticipates the questions buyers have will have a leg up in supporting conversational engagement.
The ability to mine previously inaccessible or too vast data sets will lead to an explosion of new opportunities.
LLMs have been built on large datasets. This year, we’ll see the focus shift from datasets controlled by the platforms to your own data, leveraging the processes built by those models to unlock approaches to your data, especially unstructured data, that haven’t previously been possible—giving companies new ways to understand buying signals, refine their ICP, and more, without needing a data scientist to filter the signal from the noise.
What to watch for: Although GPTs aren’t great at data analysis yet, they are good at getting more from unstructured data and applying reasoning logic to it. Companies will create custom GPTs to tap into these capabilities and use those insights for everything from better, more precise targeting, leading to more effective and budget-efficient paid media, and better pipeline to creating more personalized automated customer engagements. (Think chatbots with the ability to intelligently adjust the conversation based on trends it recognizes in the customer interaction.)
What to watch out for: The big focus thus far has been on general purpose models like ChatGPT that use a broad data set to perform functions across multiple areas. But there are also many companies that are developing vertical SaaS AI solutions using very specific industry data sets to solve previously unsolvable industry problems. Organizations in these verticals that are quick to adopt these technologies will have an enormous competitive advantage over those that aren’t.
The Iron Horse insight.
You can’t stop change by looking away. AI will continue to wreak havoc on how we do things—but it’s full of opportunities for those willing to take them. Success in this new landscape requires keeping a close eye on how AI is changing the marketplace and leaning in to using AI in novel ways to intercept the new buyer with the information they need when they need it, wherever they are.
