AI Gives Answers. Businesses Need Better Questions.
Watch on YouTube, or listen on Spotify and Apple Podcasts.
Mash met Benjamin Baer at the Plug and Play co-working space in downtown San Jose, where Benjamin and a co-founder are building DecideWise, a community for the people who buy, implement and use “decision intelligence” technology — the tools companies use to put their analytics and AI to work inside real business processes. Benjamin has spent about 30 years in B2B product marketing — at companies like VMware and Sun Microsystems, at startups, and for roughly the last decade in the FICO platform world. The conversation covers why the corporate years didn’t make leaving feel like a risk, how a free community plans to make money, why AI without guardrails is a problem for regulated industries, and why answers matter less than the questions.
This conversation has been edited and condensed.
“Operationalizing their AI”
Tell us a little about DecideWise — who you are, what you do, and what this community is all about.
DecideWise is a community we’ve just created for IT professionals: buyers, implementers and users of what’s called decision intelligence technologies. Decision intelligence is about how companies go about operationalizing their AI and their analytics. What is the process companies go through to use data, use analytics and ultimately use AI to automate and scale their business processes? Decision intelligence is a relatively new concept, but its basis is in technologies that have been around for 25 or 30 years.
When a company makes a bad decision, what are the consequences?
Depending on the industry, they can be catastrophic. Some of the early adopters of these technologies have been financial services and insurance companies. How do I originate a loan or a credit card? “Originations” is the financial services term for new customer onboarding. You can make a bad decision about a loan or a credit card and end up seeing hundreds of thousands or even millions of dollars in losses.
That’s a decision we’ve all faced — applying for a car loan, or to rent or buy a house.
And you want to operationalize it so it’s consistent — treating everybody without bias, remaining compliant with the law, which requires you to audit all the approvals — and scale it. Banks are looking at thousands, sometimes hundreds of thousands, of credit card applications every day. They can’t have a human review every single one.
Let’s walk back. How did you become one of the experts in this space?
I’ve been in technology marketing, specifically B2B product marketing, for about 30 years. I’ve worked for really big companies like VMware and Sun Microsystems, and for startups. Around 2012 I took a bit of a risk and pivoted to Fair Isaac Corporation, also known as FICO. People might know them from the FICO score. But the vast majority of FICO’s business is helping banks, insurance companies and other companies figure out how to use their analytics in their business process.
The technologies run in a continuum from data management to streaming analytics, predictive analytics, optimization and business rules. A lot of really esoteric technology. But connect it all together and it’s instantiated within companies of all sizes and shapes: how do I take my data, learn from it in real time, and use it to inform my decisioning process?
So why wouldn’t a company just hire FICO? Why does it need a community of people going through the same challenges?
Within DecideWise we created a vendor database, freely accessible at decidewise.net, of over 380 companies that touch this continuum. FICO is one of the platform vendors, offering a one-stop solution. But in a lot of cases they’re not necessarily best of breed, and there’s a lot of legacy analytics and CRM tooling already inside the enterprise. So companies are looking for interoperability, integration, and people with experience implementing these.
The use cases are extraordinarily diverse. What works for financial services won’t necessarily work for telecom or retail. So we’re looking for people with experience across these tools and industries to share what they’ve learned: “We had this use case, we looked at these five vendors, we chose this one for these reasons, and here’s our experience with them.”
Today there is no other place to do that. You can go to Reddit or Discord, or to the vendors themselves, who are usually just talking about their products. We see an opportunity right in between. And frankly, AI changes the whole story.
“Neither of us see this as a risk”
You’ve had 30 years in the industry at a bunch of great companies. When did you know there was an opportunity here worth risking it all for?
It’s been in the back of my mind for the last 10 years, working in the FICO platform world. The breadth and depth of competitors in the space just gets bigger and bigger, so I started to think a lot about interoperability and integration.
A year ago I found myself having a conversation with an old co-worker from many years before. She came from the world of building community. Among her many jobs as a marketer, she helped build out the TechXchange at IBM, which now has 450,000 community members discussing IBM components and how to work in the IBM ecosystem. She’s also spent 30 years in this business. We came to the conclusion that not only is there an opportunity here, but she has a very complementary set of experiences in building community. And frankly, we just wanted to build our own thing.
The big corporate environment is really good and comes with lots of benefits. But somebody once gave me the metaphor that you are nothing more than a handful of water out of a bucket. Take a handful of water out of the bucket, look into the bucket, and it’s like you were never there. We liked our time in these big companies, but we felt it was time to carve our own path.
The interesting thing is you used the words “risk it all.” Neither of us see this as a risk. We both see it as a great opportunity. If it doesn’t work out and make us the money or the return, we’re no worse for it. Our CVs still stand on their own, and this is still seen as a positive.
I teach entrepreneurship at San Jose State, and lots of my students want to start companies straight out of college. I see value in working for a while first — but being naive with very little to lose is also a great time to start. What’s your take?
I don’t think there’s a right answer, and I’ll tell you why. If I looked back on my career and gave myself advice, it would have been: do the startup thing. Start your own business right out of the gate. Your monthly nut is a lot less. I didn’t have a mortgage, kids or a wife. You can live on ramen. When you get older and you’ve got to deal with blood pressure, you can’t live on ramen.
On the flip side, there are a lot of things I learned in the corporate world — voice of customer in particular — that I didn’t know. Coming out of college, frankly, you’re going to be a little naive. You’re going to think you have a lot of the answers. Maybe you do, maybe you get lucky, but maybe you don’t. Going into a problem statement not knowing all the answers, and being very comfortable with that, is a much more mature, experienced process. So it’s a trade-off, and I don’t know that there’s a wrong answer.
“You can’t have that unless you have a lot of people”
I’m intrigued by your decision to build a community. I don’t understand the revenue. How do you make money with a community?
We get that question a lot. We’ve come up with eight or 12 monetization models, depending on the last count, but all of them are predicated on having people in the community. This is the challenge with community. People join because they see like-minded people and really interesting content, and they want to be part of the conversation. But you can’t have that unless you have a lot of people. So we’re struggling today to grow the community and the footprint.
Students and entrepreneurs are part of our target market. Decision intelligence is having a dramatic impact on data science and the way it’s viewed, and AI is getting rid of a lot of it. If you’re studying SQL, you’re not going to use that. You’re just creating a generative agent to do it.
But once we get to — I don’t know if it’s 1,000, 5,000 or 10,000 users — there are so many ways to monetize it. We’ve already started looking at vendors to provide sponsorships. We have three sponsors, a single paying sponsor today, and we’ll grow that over time. We’ll do webinars, podcasts, content syndication. We may rent out the database for lead generation. We think decision intelligence is in its nascent stages and will become a really big market over the next five to 10 years, so we can do virtual and physical events too.
Is it a free community or a paid one?
Today it’s entirely free. We really want to encourage people to come. But it’s focused on the buyer, implementer, user and vendor community. We don’t want conversations about your favorite car.
“You should drink Pepsi”
Let’s talk about AI. How do you see it affecting your community, and how are you helping smooth out that transition?
It’s a double-edged sword. People with a long history in predictive analytics and optimization — optimization solver technology was the original AI — see AI really speeding up their business. But it doesn’t solve the fundamental question of how I use it in a business process.
Think about business process at its simplest. If I go to an agentic AI without any guardrails or rules and ask, “What soft drink am I going to like, based on who I am?” — if you’re Coca-Cola, you’re running a big risk that the agentic AI is going to say, “You should drink Pepsi.”
So underlying technologies, primarily business rules, provide those constraints within an agentic AI framework. Business rules have been around for 40 years, and now they’re coming to the forefront yet again. That lends itself to a community of people asking, “What’s the best practice? Has anybody ever done this before?” Simply deploying AI by itself leads to a whole bunch of problems, and we’re trying to help people understand from experience how to avoid them.
Have you heard any horror stories of decision makers using AI and making really bad decisions?
There are lots of them. I was actually writing a blog for another company in the guardrail space, and all I did was write horror stories of generative AI — creating events, marketing campaigns. It runs the gamut. I come from the world of financial services in my FICO experience, and there are so many compliance issues. The one thing you know about generative AI today: if not constrained, it’s incredibly biased, based on the data that’s in the model. And as a financial institution, you cannot afford to be biased in any of the decision-making you do.
“We’re interested in the question”
A lot of my students just want to be prepared for a workforce with AI everywhere. The data analysts and data scientists will come face to face with these new decision tools. What tips do you have for them?
I’ve had this conversation a lot recently — just yesterday, as a matter of fact. I think jobs will fit into one of three buckets. You don’t want to be in the first one, which is that the job just goes away. So it’s really important to understand the other two.
The second is jobs that fundamentally change in terms of their scale. One of the early use cases for generative AI was in medical radiography, to detect cancers, lesions or other illnesses. I can build a model that detects brain cancer better than any human can. The immediate thought was that the people who run radiography labs and the professionals in medical imaging were all going to lose their jobs.
Exactly the opposite happened. Today medical radiography is, I think, the second-highest-paid medical profession, and there’s huge demand. The models are only tuned toward very specific diagnoses, so you’ve got to generate hundreds of models depending on what you’re looking for. All of a sudden there was great value in those medical images, and demand for medical imaging went through the roof — which meant huge demand for people to run the imaging equipment, and to go through the individual models and do the diagnosis.
We’ve got to feed the models.
Got to feed the models. It fundamentally changed what a medical imaging specialist does. So think about what that means for somebody in a financial institution, an energy company, retail.
It could even open the door wider. A lot of people who want to be financial analysts couldn’t get through the narrow door that used to be there. If AI can supplement them and help them make better decisions, more of them might get in coming out of school.
But people are going to need a much more sophisticated, maybe longer, educational cycle to give them the experience they need to walk into a market that already has a lot of experienced people. I’m not sure how that rolls out. That’s an open-ended question.
Let me give you the third category, the one that’s near and dear to my heart: data science. Fifteen years ago there was a sea change called Hadoop and MapReduce — big data. Collect as much data as you can, run some descriptive analytic on it, and figure out what insight is in there. That market’s gone. Today generative AI can tell you exactly what’s in that data repository. You don’t need people to do that.
What’s really new in decision intelligence, different from data science, is that we’re not interested in the outcome anymore. We’re interested in the question. Generative AI isn’t going to ask the right questions. It’s going to give you the answer to the question.
“We’re not interested in the outcome anymore. We’re interested in the question. Generative AI isn’t going to ask the right questions.”
— Benjamin Baer · Inspiring Founders Podcast, Ep. 06
Increasingly, companies don’t want to know what the data tells them. They want to know what they’ve got to do, and they want that answer now, in real time. It isn’t good enough to tell me how much merchandise I sold last quarter, which is what a CRM does. They want to know how much to stock today, and what kind of discount to give the guy at the retail kiosk, in real time, while he’s buying something. And if they ask the wrong question, they get great answers that aren’t relevant.
Have you ever read The Hitchhiker’s Guide to the Galaxy? The big catch was that the answer was 42, and then they had to spend billions of years trying to figure out what question that was the answer to. They didn’t have the question, but they had the answer. I think that’s fundamental to what businesses face today. They get a lot of answers, but they don’t know what the question is.
When I write in my journal, something that comes up often is that it took me a long time to learn how to make a good decision. I wish there had been a way to learn. You’re saying there are ways now, and the technology is accessible to everybody.
I want to make the distinction between the personal or consumer decision and the enterprise decision. Enterprises have lots of people at the table, lots of rules they need to follow, and things they value — a constitution for the company. We all know it in our gut: what’s the difference between buying a product from Apple versus Dell? It’s not that one is better, but they value different things, and those things can be codified and executed on a consistent, scalable basis. That codification of decision logic is where I think today’s data science can have a profound impact.
Can small businesses, solopreneurs and startup founders do the same thing?
Absolutely. There are a number of products available for free, open source, that they can use and should use. There are also industry standards that make this easy to understand. One is called DMN, Decision Model and Notation, from an organization called OMG, the Object Management Group. You can find it online and in Wikipedia. It codifies the decisions you need to make in the simplest of terms — who owns the decision, what data and analytics you use, what outside influences affect it — right up front.
Historically businesses have approached decision-making by asking, “What data do we have? What analytics do we have?” This turns that on its head. On a clean sheet of paper, what are the components that make the perfect decision? Then you apply the data, the analytics and the ownership afterwards.
There’s another component: what business outcome do you want to influence? What does the world look like after you’ve made these decisions? That’s often missed. The metrics don’t match the goal at all.
AI could make that worse, because people want to skip the hard work. And some people want to learn everything — “Tell me where the open source is” — while others just say, “Who do I hire?” Is there a place for that second group to come to your community and say, “I need help”?
That’s part of the challenge we’re looking to solve with DecideWise. Reddit, LinkedIn and broader agnostic communities have a lot of dialogue, but they’re not really places to raise your hand and look for assistance. We want to create a place where people can ask questions or look for that skill set.
I had an aha moment about this a few years back. A big bank in Latin America built a really cool streaming analytics, sense-and-respond system. The biggest bank in the region said, “We need to build that,” and hired that team right out of the other bank. Then a smaller regional credit supplier hired them away too. The same team went from bank to bank to bank, which is great for them, but it doesn’t grow the knowledge base or the number of people. Imagine if they’d been able to put their shingle out and take offers from retail, or from banks in Europe.
Is there a social component too — responsible AI, responsible analytics? What role do you play there?
It will always be important, particularly with neural networks. They’re completely unexplainable, and I don’t think we’ll ever get to a point where you can explain the logic of generative AI without a lot of tweaking. That will be core in financial services, healthcare, insurance — highly regulated industries that cannot be hindered by bias. They have to be able to explain why they did what they did.
But in DecideWise we’ve so far not structured the conversation. If people want a deep dive on responsible AI, we can build a space for that. If it becomes an AI community over time, or a rules community, or an optimization community, so be it.
“We want to iterate fast”
Last question. As a founder, what are the big challenges you’re facing right now — maybe things you didn’t expect when you hung up your shingle?
One example: about 20 years ago, a buddy of mine worked for a Web 2.0 startup, and he kept saying it was a warning flag — and I only mean this with love and respect — that the CEO was an aeronautical engineer by training. When you design an airplane, you’ve got to make it perfect the first time, because otherwise the airplane falls out of the sky and people die. He couldn’t grok iterative software development.
So that was the first thing we built into decidewise.net, building out the database and the community: we want to iterate fast. We launched the site in June or July of this year, and we’ve already iterated on it three times. It turns out that’s easy. That’s not difficult to do.
Our biggest challenge today is how we evangelize, promote and capture new members in a way that creates value in the interim — between it being a few hundred of us and it being five or 10 thousand of us. We know it will get to Metcalfe’s law. We need to get to a place where there’s a flywheel spinning, and we’re not there yet. So we’re encouraging everybody who can see themselves being impacted by AI — including the data science students figuring out how they’re going to differentiate themselves — to join our community at decidewise.net.
More conversations with founders on the Inspiring Founders Podcast.