I’ve quoted more of these projects this year than I’ve won, and I’ll admit it has been deflating. It would be easier for me to take if they had been bad ideas, but they weren’t. They were problems worth solving. Someone inside a company had found a real problem, built something with an AI tool that actually worked, and then came to us for help getting it ready for real use. Then it stopped, and the reason is some version of “that’s not in our budget.”
I should say up front that taking AI pilots live is work my agency now sells, so I have a stake in this and you should read the rest with that in mind. But I don’t think I’m the one who feels this the most. I think it’s the builders, the people on staff in sales or marketing or operations who used the tools their company gave them to make something useful, watched it work, and then watched it go nowhere. If that’s you, the easy conclusion is that your idea wasn’t good enough. I’ve come to think that’s usually the wrong conclusion, and I want to explain why, starting with something that happened to me long before any of this was called AI.
I’ve been on the other side of this
In 2002 I was a systems administrator at Philips, working out of the Chicago office while the rest of the IT department sat in Seattle. One of my first assignments was to evaluate new copiers, and the money for those copiers was coming out of the operations budget. While I was working on that, I learned that the order management team was photocopying hundred-page order packets every time a deal closed and shipping them overnight to the reps in the field, which was costing them a lot of time and a lot of money. I’ve written before about how I came across that problem, so here I’ll stick to what happened after I had a fix for it.
The new copiers could scan to PDF, and the order data was already being exported every night into an Access database, so the pieces of a fix were sitting right there. I took the idea to IT and asked for help, and they wouldn’t put a team behind it. Looking back I can see how that happened. The money for the copiers belonged to operations, the problem belonged to sales and order fulfillment, and the skills to fix it belonged to IT, and there was nothing connecting the three.
So I built it myself. It took Dreamweaver, a copier with a scanner and OCR, and that Access database, and when it was done the reps could pull up their packets online instead of waiting on an overnight envelope. It worked and it saved the company money. It also got me in trouble, because when IT found out they were upset that I had gone around them, and they had a point. There had been no change management and no approval from anyone.
I’m not telling that story because I was right. I’m telling it because of what was missing. There was no process and no budget for someone like me to bring an idea like that forward and have it looked at. The only choices I could see were to drop it or go around, and I went around.
What’s different now
What took me a copier, a database, and some web experience in 2002 is within reach of a lot more people today. A U.S. Bank survey of small business owners last year found that nearly 7 in 10 of the ones using generative AI spend less than $50 a month on it. That’s a small-business number and it doesn’t prove my point, but it matches what I see, which is that the tools are cheap and a lot of people have them.
I don’t think companies have caught up with what that means. They’ve handed people in every department a tool that can build things, and what it builds can be good. The cost of trying an idea has dropped to almost nothing, but the cost of being responsible for it hasn’t changed at all. Something that works on one person’s laptop is a long way from something a company can safely run, and that distance is where the cost is. Someone has to check the security, make it hold up under real use, connect it to the systems it depends on, and own it when something goes wrong. None of that comes with the subscription.
Three ways I’ve seen it go
I’ve watched this play out three times now, and it went a different way each time. Three cases isn’t research, so take this as what I’ve seen and nothing more.
The first way is that it stalls. An executive at a consumer app company, someone who isn’t an engineer, used Claude to build a customer rewards site for a marketing campaign. This wasn’t a mockup. Members could sign in, prove who they were, and claim a credit, and there was a dashboard behind it for the staff to run the campaign. It was connected to the company’s real membership data and it was live on the web in test mode. When my team reviewed it, most of it held up. What it needed was a couple of security fixes that had to happen before launch, and a rebuild of the parts that were sized for tens of thousands of people when the campaign could reach millions. We quoted that work. The answer was that it was outside of their budget and that they would talk to their own technical team. It may still happen, so I’d call it stalled and not dead. But it was most of the way there, and there was nothing set aside for the last stretch.
The second way is that somebody goes around. That was me in 2002. It worked out, and I still wouldn’t recommend it as a plan.
The third way is the one that went well, and it went well for a reason I don’t think most companies can count on. A senior leader at a company I work with built a prototype of an internal tool himself. Before anything was built for real, he sat down with another senior leader, someone he talks to every day, and I was in the room for it. Over the course of that conversation the question changed from which tool to use to whether the process they were trying to improve was worth having at all. I wrote about that conversation in The Tool is Ten Percent. The build was paused, and it wasn’t because of price. It also meant we didn’t get the work, and I still think it was the right outcome. But it happened because those two people already had an open line to each other. In most organizations I think the prototype would have just been dropped, or someone would have gone rogue.
That third case is worth staying on for a minute, because it connects to what I’ve been writing about in this series, which is that the belief has to come before the build. In that case a prototype got made before the problem was clearly defined, and I don’t think there was anything wrong with that. It cost a few hours of someone’s time and a $50 subscription, and it helped get the right question asked. The defining has to happen before the expensive build, and it doesn’t have to happen before the cheap one. But suppose they had decided to move forward. They would have had nowhere to pull the money it would have taken to make that project live, which we had estimated at around $30,000.
What baffled me
There was so much enthusiasm before we quoted the work. People had built something useful and they wanted it live. They brought us in because they knew it needed more work, they were eager to see a number, and then the number arrived and the project stopped. I was baffled by that, because I don’t think the enthusiasm was fake.
I’ve wondered whether the ease of building the first version changes what people expect the rest to cost. When you can make something that works for $50, the price of making it dependable can come as a shock, and if nobody has worked out what the finished thing would be worth to the business, that shock can end the conversation before it really starts.
I should also say the obvious thing, which is that sometimes “it’s not in the budget” is a polite way of saying the return doesn’t justify the cost. That’s a fair answer, and in the third case I think it would have been the right one. What bothers me is that in the projects I’ve watched stall, I don’t think anyone got far enough to decide that. In the first case, the person who built it, the person we sent the quote to, and the person who answered us were three different people. In the case that got a real hearing, the person who built it was also someone who could decide.
What I mean by budget
I want to be careful with the word budget, because I think it gets heard as “they don’t have the money,” and that isn’t what I mean. A budget is where a company decides, in advance, what it is going to put money, people, and attention behind. It’s where the trade-offs get made, and once it’s set it drives most of the decisions that follow.
So when a builder hears “that’s not in our budget,” it doesn’t necessarily mean the company can’t afford it. It means that when the plan was made, nobody decided to put any energy behind this kind of work, because nobody knew they would need to. The decision was made earlier, by leaving it out. I think that’s why it’s so deflating to be on the receiving end. You’re making your case to someone who, by that point, doesn’t have a real way to say yes. And I wonder what that does to a person’s willingness to try again the next time they see a problem they could fix.
Projects need funding. They need it in money, but they also need the time of the people who have to review them and the attention of someone with the authority to own them. When none of that has been set aside ahead of time, a good idea and a bad one get the same answer.
That points to the second thing I think is missing. Even a company that sets money aside still has to decide which projects get it, and in the cases I’ve seen there wasn’t a way to do that. A budget line gives the money a place to live, but it doesn’t tell you which ideas deserve it. In the cases that stalled or went around, nobody could have told the builder who evaluates something like this, what would make it worth pursuing, who would be responsible for it once it was live, or where the money and people would come from if the answer was yes. In the one case where those questions got answered, it was because two people talk every day.
Why I’m writing this now
We’ve made it easier for people to build things without making it any easier for a company to decide which of those things are worth building for real. Budget conversations for 2027 are starting, which makes this the time of year when that’s cheapest to deal with. From our side as an agency, our move is going to be to ask our clients what part of their budget is set aside to take AI pilots live, because that work costs a lot more than $50 a month. I think a line on the budget for innovation could be part of the answer. I don’t think it’s all of it, for the reasons I just gave.
I have a view, and I’ve told you what it is, but I don’t have this figured out and I would rather hear from the people living it. So if you’ve built something at work that actually worked, I’d like to know what happened next. Where did it go, who did you take it to, and what did they say? You don’t need to name your company or share anything you shouldn’t. I’m trying to find out whether what I’ve seen is a pattern or just three stories.



Hey Curtis, as always, I really enjoy reading your posts and thought pieces.
This one is super interesting to me because I have lived some of this before AI was everywhere. I have supported our ‘innovation’ department for the past 3ish years in different capacities. I work for an AEC firm and so most of the ideas we see come through are very technical and usually very thought out by the submitter (99% of the time they have an engineering background in the field they are proposing the solution). And what we have seen over those years is that the idea may be technically sound, but the ROI or commercialization is an afterthought, if even that. And the struggle has been that they (engineers) just don’t really have that ‘muscle’ as it comes to the monetization (path to market, marketing, branding, customer loops, etc.) of the solution or idea. And that’s ok, if anything it makes sense because they have not needed that lens for their work, usually. But that is where the path to market lives or dies is understanding how it pays back our investment in their idea and then ideally leads to additional revenue in some way. Unfortunately the AEC space has a bigger problem with the fundamental business models, but that’s another side topic, but very much a larger input to where our business goes in the future.
The way we are trying to tackle this problem is to allow the right users and teams access to the tools that allow them to test and come up with ideas (that in an of itself is also a challenge with IT). And then what I have been suggesting as of recent is that we need to find people in the company that are commercially minded and willing to take new business propositions to their clients and try to move the client down a funnel or ladder that helps both them and us in a new/different way of doing business (this is the part where our business model hinders us, but its controlled by the client, so we need to dangle a carrot to get them to try a step in a different direction). The other shift we have made is to move incrementally so that evidence unlocks funding, rather than just going all in on a build that seems to make sense, but is being stood up by someone’s gut instinct at the end of the day. A gut is good, but VC’s don’t always bite based on the gut of some founder (I try to tell our team that we are the internal VC and the SMEs pitching ideas are founders, maybe a bad analogy). And as we go through our phases of discovery and evidence building, pulling the eject button is always an option. I will say that one of the cons to this is that people feel like they have lots of homework to do and sometimes they don’t have that time (again our business model), so in some instances that is where the funding goes, to let this person flush out the idea more and come back with what could be next (this can also be dangerous because ppl don’t like killing their babies).
The future for us looks like making a few bets on different workstreams because the idea and market may be there, but sometimes adoption is the tricker part so that becomes the next bottleneck to gain traction. And then sometimes is the ppl internally that get in the way. As you and I both know, the fundamentals are good data, good structure and I do my best to preach that as well, but that is not the shiny object so it can be a challenge.
I hope I didn’t go off on to many tangents and will be curious to hear an update on what others repot back or how things shape out for you and your team.
LV