Once again, everything in technology is changing. In 2011, Marc Andreessen wrote “Software is eating the world.” Now AI is eating software ¹, as technology adoption, usage, and enterprise buying behavior change rapidly, driven by startling new capabilities of, and as yet unanswered questions around, AI.

It has been 22 months since OpenAI launched ChatGPT and sparked the Generative AI wave. The field has already evolved in several discrete steps, from initially text answers to prompts, then multi-modal input / output, to more complete Agents, which make it possible to program a whole sequence of actions. Agents can process complex workflows and tasks, such as building a prototype of an entire web application in just a few minutes, although the jury is out on how usable or complete the application will be. Whatever the limitations, the next model will improve on the prior results in a shockingly short space of time. OpenAI’s most recent model, o1, released only 4 months after the prior model GPT-4o, evaluates and improves itself, demonstrating high scores on ever more complex thought processes.

This rapid evolution raises questions for leaders and employees at enterprises: 

What can I do with AI? What should I do with AI? What should I stop doing without AI?  

And as to each of these questions: how do we manage AI’s troubling lack of reliability? (i.e. how can I rely on its outputs?)

The pendulum swings.

As enterprises grapple with these and other questions, the Chief Technology Officer and Chief Information Officer suddenly find themselves at the center of technology strategy again. Over a decade, buying power decentralized out to the business lines as “vertical SaaS” companies, targeting specific enterprise functions, worked to erode the obstructionism of IT departments, and sold directly to line of business buyers unsatisfied with what internal IT delivered.  

Now, the pendulum is swinging back. The technology strategy is now an AI strategy which requires a unified approach, and increasingly comes with an AI governance committee that must approve any technology that includes AI.  

The software budget is increasing. Most companies expect to increase AI spending in 2025 by 22% on average, according to ICONIQ Growth’s GenAI survey, June 2024.  We understand budget is being found in reduced headcount across the enterprise, reduced spend on SaaS, especially per seat licensed SaaS, and reduced spend on Data.

What should we do?

Enterprises – see AI as a force-multiplier, not yet a fully fledged solution to automate broad business processes. Consider AI implementation a journey and an accelerant on that journey, not so much a destination.  

Pick technology partners who bring both domain expertise to solve your real business problems, and are applying AI in their solutions in a high-quality way. This ensures your organization is learning from cutting edge innovators, and that you are driving your AI strategy with innovation partners as part of it, not peripheral to it.

Startups – the window is finite. If you are not experimenting every day with ways in which new AI tools can increase quality and efficiency of software-driven processes and workflows, your competitors surely are. Speak to both Line of business and IT at your customers. Articulate how your solution is part of their AI future.  

Meanwhile, ask every single person in your company (from the janitor to the CTO) to experiment with AI tools. When ChatGPT launched, even OpenAI was stunned by the results. In this exciting new world, the next “a ha!” could truly come from anywhere!

 

Ilya Levtov

Founder and CEO, Craft

 

¹ Note: AI is mostly software, but not all software is AI.