Revolution When: The Waiting Game

Following years of hype, Generative AI has been added to pretty much everything. Even if it's little more than a checkbox feature. AI assistants have been embedded into every major app . Many marketing technology platforms have introduced some form of content generation workflow. Meanwhile, major tech companies are looking to redesign their products as glorified chatbots (Microsoft 365 is one such example). The trouble is inventing that killer use case, which justifies such a radical redesign.

AI In Production

Everyone knows that AI is the future. Copywriters and translators can attest to the impact of generative AI on their industries. Tech firms are betting on AI powering the next wave of digital transformation. Investors see AI as the path to untold riches. Corporate leaders are scared of being left behind, in case it disrupts established markets. Yet, many firms are still no closer to production deployment of GenAI automation than they were twelve months ago. Executive mandates and AI task forces aren't leading to production deployments.

Generative AI does have a role in the modern enterprise, even its current state. Knowledge workers are already using it for meeting notes and web searches. These are both highly effective uses of the technology, but they're not the type of activity that can justify the stratospheric valuations seen by many AI startups. That requires agents to become a corporate reality, but we're still a long way from that happening. Even if they're already being used in many customer service scenarios.

Customer Hesitation

It's no wonder we're seeing a high degree of pushback against AI. Much of the rhetoric emanating from AI leaders is scarcely believable. Increasingly shrill claims about AI driven job losses are being laughed at on social media. Not helped by tech firms using AI to justify offshoring, R&D cuts and pre-existing reductions in graduate recruitment. Meanwhile, even big Salesforce fans could not defend Marc Benioff's recent claim that AI automates 30%-50% of his companies' internal processes. Supposedly, Benioff was only referring to increases in developer productivity.

In the real world, Salesforce's Agentforce platform is struggling to overcome customer hesitation. Understandable perhaps when the platform is still effectively in beta. After all, Salesforce still haven't settled on a definitive pricing model for the platform. Major new features are being announced monthly. And the company's own benchmarks show that agents routinely make errors. The same situation applies to other agentic CRM platforms - HubSpot's agents are still mostly in private beta.

New Technology

Given the current market reality, it's tempting to dismiss Agentic AI entirely. There are very real issues with Generative AI that will affect its eventual usage for automation. Even Gartner are advising CIOs to hold off on agent deployments until next year and beyond. However, the market for agents is still very new. Salesforce's Agentforce product was only launched in September 2024 - less than 12 months ago. The current boom sector is development tools - yet, the term Vibe Coding was only coined in early February 2025. The practice is older than that, but it's still a very new technology in the very early stages of its hype cycle.

Senior executives are understandably worried about being left behind, but they needn't be. Generative AI is still very immature. Scientists are still figuring out the limitations of their models, and how to work around them. Few people genuinely know what the technology is truly capable of. It's rare to see a new technology with revolutionary potential developed in the global spotlight. Yet, that is exactly what is happening with AI. LLMs are undergoing an active R&D process that typically only plays out in a lab. It's an exciting prospect, but at a level of maturity that rarely gets deployed to production.

Pretty much everyone is still working out the best use cases for Large Language Models. We've seen some promising ones achieve mainstream usage over the past twelve months, with transformational impacts on SEO and on marketing more broadly. A plethora of startups have emerged with other new ideas. The best ones will be copied by established players and embedded into their existing product portfolios. Businesses should be actively reviewing their processes for potential uses of the technology. Ongoing experimentation is a good idea in any field, but don't get hung up on production readiness yet. Keeping a few promising pilots on the back burner is all that is required at this stage.