I nearly lost 3 weeks building the wrong thing. The annoying part is that the code was good and the screenshots beautiful.
Monti started (and it still is) as a system to track decisions: what you decided, what you expected, what actually happened, so that the next time you take a decision, you do not start from scratch.
But decisions rarely happen in isolation.
A call with a potential client.
A pricing conversation.
A partnership you decided to pursue.
All of that needs context.
So I started building lightweight contact and relationship tracking around the decision system.
It was useful.
But somewhere around week two, I noticed I was spending more time on the relationship infrastructure than on the decision loop itself.
The screens looked polished. Pipeline stages. Email threading. Relationship notes.
You could open the app and point at things.
“This works. That works. Look, it syncs.”
The problem surfaced in the morning briefings that Monti itself was providing me.
Every day, Monti generates a starting recommendation based on open decisions, unfinished loops, stale assumptions, tasks, and recent activity.
And despite all the relationship tracking work, the briefing never cared about any of it. Maybe once when there was a bug, but other than that, nada.
It kept pulling me somewhere else instead.
Toward decisions I had made weeks ago and never reviewed.
People I said I would follow up with but hadn’t.
Ideas that sounded right when I logged them but quietly produced nothing afterward.
At first I ignored it. I was already deep in. "Sunk cost bias" doing its thing.
But after about the third week, it became difficult to ignore what the product itself was showing me. It started calling it out!
The contact context mattered less than the memory around decisions.
Somewhat in this direction:
What happened the last time I trusted this instinct?
What did I expect?
What actually happened?
That changed the roadmap.
Monti still has relationship tracking because decisions need context.
But I stopped deepening it and shifted focus back toward the decision side of the product.
I think that was the first time I felt the product disagree with me in a useful way, and that was so much more helpful than the generic "You're doing great" ChatGPT stuff.
Curious if anyone building alone has had this happen when you had that moment of realization.
"The first time I felt the product disagree with me in a useful way."
That's the line.
Building alone, the hardest part isn't the code. It's that there's no one to tell you you're going in the wrong direction. You can spend weeks on something that looks good and works correctly but isn't the thing.
I had a similar moment building a workflow system for vibe coders. Kept adding features — startup scripts, settings guides, escalation protocols. Then looked at what I actually used every day: a single MD file. Session end, three fields. That's it.
The product told me what mattered by what I actually reached for. Everything else was polished scope creep.
"What happened the last time I trusted this instinct?" is a question most solo builders never ask because there's no system to store the answer. That's probably the most underrated part of what you're building.
My failed Product Hunt launch with only eight upvotes taught me the exact same lesson about ignoring the core value. It is easy to get lost building polished infrastructure while the real signal is pointing somewhere else entirely.
Letting the product's own logic kill a feature is the ultimate win for long-term focus and clarity. That shift from being a data dashboard to a decision engine is where the real magic happens.
What was the hardest feature to stop developing once you realized it did not move the needle?
It was this one - contact and relationship tracking - because while it provides an obvious gain for a business in general, but only a very limited context around a CRM is truly useful for decision making.
But Monti itself pointed out the bias in this approach taking into account the other context around the goals established.
Funny enough, Monti was originally built for another product I’m working on called ARAMA. A few weeks ago it stopped us from repeating a decision pattern around a customer group we had already seen fail almost a year earlier.
That was probably the first moment where I thought: “okay, this thing actually remembers better than I do.”
What about you? What was the hardest feature to stop developing in Business GPT?
It’s honestly amazing to hear that Business GPT was actually useful for you. You have a knack for surprising me in the best ways! I’d love to hear more about your experience specifically, which part of Business GPT did you find most applicable to your work?
Business GPT follows a 4-step process, and I like to think of it as connecting separate dots into a single, cohesive line from the second a user inputs an idea. While every single step is vital, our team definitely hits walls when it comes to time and manpower we can’t pour 100% of our energy into everything at once.
That said, the goal of making a PRD so simple and intuitive that users can just breeze through it... well, we still have a long way to go. But it’s one of those features that’s just too important to give up on. Does your project have a 'challenging but can't-quit' part like that?
Yeah, I think every product ends up with one “looks useful vs actually useful” battle underneath it.
For Monti, the hard part is avoiding organizational theatre.
A founder can spend hours updating tasks, notes, contacts, pipelines, dashboards... and still avoid the one uncomfortable decision that actually matters.
What surprised me is that once Monti started tracking decisions over time, it became very good at spotting those avoidance patterns.
The challenge now is less about collecting more information and more about keeping the signal clear as the context grows.
Ironically, the more context the system has, the easier it becomes to hide from the real signal.
I have to give you a round of applause for that incredible insight. I’m constantly guilty of the same thing spending way too much time on 'maintenance' like updating tasks, notes, and dashboards, which essentially means tidying up without adding any new value. Meanwhile, I find myself putting off the high-stakes creative decisions, like locking in a new design concept. It happens way more often than I’d like to admit!
I wonder, could your service actually help me identify these avoidance patterns of mine?
That’s actually one of the main things Monti ended up becoming useful for.
Not in a “you are procrastinating” productivity-app way but rather “you keep revisiting low-risk organisational work whenever a higher-uncertainty decision appears.”
The interesting part is that those patterns usually only become visible over time.
Individually, task cleanup, research, notes, restructuring plans, all look reasonable.
But repeated together before every difficult decision, the pattern becomes hard to ignore and very difficult to not address.
I completely agree. Too often, when you get bogged down in low-priority organizational tasks, you lose the motivation to tackle truly important and productive work.
One thing I am curious about is how Monti recognizes avoidance patterns that the user hasn't explicitly mentioned. Is there anything the user needs to pre-configure or handle manually?
Interestingly, a service called Bunzee, launched by our team, is designed to handle exactly those 'high-priority and productive tasks' on your behalf. While Monti and Bunzee have different core purposes, they share an interesting commonality. When it comes to the grueling work that developers and makers like us want to avoid, Monti kindly points out what we are putting off, and Bunzee just steps in and does it for us!
I’d like to know more about Monti. Could you share your experience with Bunzee.ai as well? I think we will have a good relationship.