Originally aired January 7, 2026 on Extropy On AIr.
In 2026, the organizations that win with AI won’t be the ones that buy it. They’ll be the ones that actually use it well. In the first episode of the year, Ravit Gutman and Mike Jasper look back at what companies learned from AI adoption in 2025, how to measure AI value, where AI is already making the biggest difference, and how better, more intentional prompting turns AI from a search box into a real thinking partner.
Listen: Extropy On AIr podcast page | Spotify
Key takeaways
- Companies that didn’t see AI value in 2025 aren’t giving up. They are rethinking their strategy.
- Measuring AI value is hard. Quantify the steps you eliminate, and look at outcomes like forecast accuracy and speed to value.
- The biggest AI impact so far is in data-heavy work like contract management, customer service, order to cash and cash application.
- AI isn’t replacing people. It is reallocating where human value comes from.
- Better prompts come from being intentional: who, what, why, and several rounds of refinement.
What did companies learn from AI in 2025?
The previous episode on McKinsey research, which suggested only about 6% of organizations are seeing significant value from AI, got a big reaction. Mike heard from people who said they were in the other 94%. Some had adopted early without enough of a vision. But most weren’t giving up. They were restructuring their approach, because AI will be a competitive differentiator whether they use it or not.
Ravit sees the conversation moving from talking about AI strategy to defining what it actually means. Throughout 2025, the AI label was slapped on everything at conferences, even tools that were really machine learning. In 2026, the question is what a real AI strategy looks like for your people and how you get value from it.
How do you measure the value of AI?
Mike sees two kinds of benefit. The first is easy to quantify: if AI removes 15 manual steps, you can measure the time saved and multiply it by the burdened cost of that time. The second is harder. Much of AI’s value comes from adding new processes or reaching better, more validated answers faster. There, you measure the outcome, like improved forecast accuracy, rather than the AI process itself.
For Ravit, the best measure is speed: how much more you can accomplish in a day and how much faster you reach value. Extropy is also looking at how to recognize people who use AI to rethink how they work, because what will set organizations apart is not buying AI but using it well.
Where is AI making the biggest impact?
Mike points to contract management, where AI agents can compare the intent of your clauses with the intent of a supplier’s paper, opening up bandwidth in a process that is often a bottleneck. He also highlights customer service. In a global economy, customers almost always have another option, so the relationship becomes the differentiator.
Ravit sees the biggest impact in work that involves taking data, validating it and verifying it across multiple sources, like order to cash, collections and cash application. She has also seen healthcare gain a lot from automation. In 2025 much of the old RPA world shifted to AI, and she is excited about the next step: treating different AI models like employees with different skill sets and having them work together.
Why is trust in AI growing so quickly?
Mike remembers when automated phone systems meant pressing zero to reach a human. Today, AI service tools are so attuned to what customers need that people don’t mind using them. That everyday comfort carries over into business and makes organizations more open to new opportunities.
Ravit compares it to the early days of RPA, when people feared bots would take their jobs and it took years to show how humans and bots work together. With AI, that comfort is coming faster, from people in their 70s to students. Mike has watched schools go from treating AI use as plagiarism to teaching students how to use it well. His conclusion is that AI isn’t replacing humans. The human value now lies in asking the right questions, thinking alongside the tool and doing something meaningful with what it produces.
How should you choose which AI tools to use?
Mike’s advice is to try them all. Take a prompt you used in one tool and run it in another. Most offer free trials, and you will quickly find which ones you prefer and which work best for certain tasks. He has found that talking to AI with voice, rather than typing, makes him more creative when brainstorming because he is focused on the conversation instead of the input.
Ravit uses different tools for different needs. She calls Claude her favorite for speeding up day-to-day work as a business owner, and says Gemini 3 was a big step forward. She also notes that tools like Perplexity now let you use several models in one place, which makes it easier to brainstorm in one model and refine in another.
How do you write better AI prompts?
Ravit’s approach mirrors how Extropy works with clients: get clear on the who, the what and the how before you start. Tell the AI who you are, who you want it to act as, such as a CFO, CMO or legal advisor, and what problem you are solving. Then, instead of writing one long prompt, ask it what information it needs from you. That turns it into a conversation and surfaces questions you hadn’t thought of. Her rule is at least three rounds of refinement on anything important.
Mike uses the who, what, when, where and why he learned in seventh grade writing class, and he has found the why changes the output the most. His example: a search engine gives you a generic top 10 chicken recipes, but telling an AI you are cooking a special anniversary dinner and why gives you a completely different, far more useful list. Then you refine it, just like a conversation with a person. Research that used to take him eight hours now takes a 20 minute AI conversation.
Ravit sums it up with a simple test. If you had 30 minutes with someone brilliant, you would prepare carefully and think about what you really want to ask. Being that intentional with AI is how you get the best results.
What does this mean for finance teams and hiring?
Mike is direct: there is someone in every industry using AI, and even the best finance professional who refuses to use it will eventually fall behind someone who does. Ravit now asks candidates how they use AI in their work today, and expects AI to shape how companies hire, run projects and are evaluated.
In their words
It’s not the people that buy the AI, it’s actually the people that use AI and use it the right way within their organizations.
Ravit Gutman, Extropy Advisors
AI is not replacing humans. It’s reallocating the value.
Mike Jasper, Extropy Advisors
About the speakers
Ravit Gutman is Founder and CEO of Extropy Advisors, a women-owned finance transformation consultancy.
Mike Jasper is VP of Finance Transformation at Extropy Advisors, where he leads digital transformation and client success, working hands-on with clients to set success criteria, optimize their people, process and technology, and measure the results.
FAQ
Why did so many companies struggle to see value from AI in 2025?
Many adopted AI early without a clear vision, or treated “AI first” as a slogan rather than a strategy tied to people and outcomes. Measuring value was also hard. Time saved on eliminated steps is easy to quantify, but AI that adds new processes or produces better answers is much harder to put a number on.
How should leaders measure the value of AI?
Look at both sides. For automation, measure the manual steps eliminated, multiplied by the burdened cost of that time. For AI that improves decisions, measure the outcome, like forecast accuracy or how much faster the team gets to value. Ravit Gutman’s own measure as a CEO is how much faster she can get things done.
How do you write better AI prompts?
Start with who you are, who you want the AI to act as, what you are trying to solve and why you need the output. Then ask what information it needs from you, and refine over several rounds rather than dumping everything into one long prompt. Ravit’s rule is at least three iterations on anything important.
Keep exploring
- More episodes of Extropy On AIr
- Previous episode with show notes: 2025 in Review, 2026 in Focus: Leading with Intention
- Next episode with show notes: Unlocking Intelligent Invoice Lifecycle Management with Basware
- Six months later: AI in Finance at the Midpoint of 2026
- Related: Execution Beats Experimentation: Agentic AI in AP
- Why Extropy
Ready to move your AI strategy from hype to real results? Book a discovery call with our team.
