Many of our client conversations lately start the same way. “What if we automated that?” “Can we build an agent for this?” “We should just spin something up.”
We get it. We’re excited too. We’re deep into tools like Manus, Claude Code, and Base44 , figuring out what’s possible now that the barrier to building has basically collapsed. These tools genuinely change what teams like ours—and the clients we work with—can do.
A client came to us wanting to automate a spreadsheet they use to track project assets. Great, we said. Now, walk us through what happens when an asset moves between two teams that name things differently. Silence. They’d pictured the automation before they’d pictured any of the mess it needed to sort out.
This is the pattern we keep running into. Clients know we can build something, but what they don’t see yet is what goes into building something stable, secure, and worth the effort. Our job isn’t just to build the workflow; we need to make sure they understand what they’re getting into before we start. What real effort is involved in covering all use cases? What decisions will still need a human? Where will things inevitably break later?
Here are five things everyone should understand before building that app or agent.
It’s a starting point at best. We’ve had clients come in excited, ready to move fast, but without clarity on how the app should function or what success looks like. When that foundation is missing, everything downstream gets murky quickly. As the build stalls and the scope creeps, the team loses confidence in what they’re making.
The questions that need answering upfront are not glamorous. What should the user experience feel like? Does this need to connect to existing tools or platforms—and if so, which ones? How can we make sure security is baked in? What happens when it breaks, or when the input isn’t what you expected?
These questions are the foundation. If they’re not answered before the build starts, no amount of good tooling or talented execution is going to save the outcome. You can’t expect something great to come out the other end if you haven’t done the work going in.
What changes with vibe-coding tools is the depth of knowledge required. Non-engineering teams can now prototype, test, and iterate apps and agents without writing lines of code. That’s a real game changer.
But while you avoid the tedium of coding, you’re still mapping flows, writing logic, testing edge cases, and reworking things (over and over) when they don’t behave the way you wanted or expected. The skill requirement now is less about programming language fluency and more about how clearly you can think through a system. That takes time—often more time than you budget for.
It’s too early to build out a clean playbook or repeatable process for this. We can’t say: “This is how it’s always done.” We’re trying something, watching it break or kind of work, tweaking it, and repeating. And then tweaking again when the AI model is updated. This loop is just what the work looks like right now.
If you’re expecting a clean, linear, predictable build process, prepare to be frustrated. If you can stay comfortable experimenting your way forward, then this ambiguity can become an advantage. Because there’s no fixed blueprint, time is your biggest input. Every “quick tweak” is another round of testing, another decision point, another refinement cycle. When you’re working with an external agency, those billable hours are real costs—not because anything is inefficient, but because the work itself is iterative by nature right now.
The opening question shouldn’t be, “How fast can we build this?” It should be, “How much are we willing to invest in getting this right, and what does “right” look like?” Those are very different conversations.
Thirty percent of no-code builders are non-technical individuals. But while anyone can open a no-code platform, not everyone can build something useful with it. The difference is in the clarity of the idea, the understanding of who it’s for, and the discipline to refine and test rather than just ship.
The tools are genuinely great and getting better every day. But whether you’re building an app or agent yourself or bringing in a team to do it, the thinking has to come first: the strategy, the investment you’re actually willing to make, and an honest assessment of what you’re getting into. Once you’ve got that established, the possibilities are endless.
Lisa Larson-Kelley is president and CEO of Quantious.