Sometimes these blogs turn into a personal diary, I think this one is going that way. As a founder I am my own worst critic, I think many other founders can relate to this. Nothing is ever good enough, no effort can justify ending the day, and success is not a real thing - only the pursuit of it is. The times I judge myself hardest are when I’m working on the 20% of the 80/20 rule.

If you haven’t heard of the 80/20 rule, it’s this concept where 80% of the result comes from 20% of the effort, and the other 20% of the result takes 80% of the effort. Beginning a task is simple, getting most of the way there is manageable, but completing it is difficult. That’s how I am. I struggle with long-term effort because the results don’t appear quickly. You spend day after day on that last 20%, and nothing shows up to keep you motivated. Now the funny thing is, this is the entire definition of a startup. Most days are bleak.

You start a company really motivated. You work long hours with your small team, to see almost no progress. You send sales email to the abyss, obsess over UI/UX with no customers using it, and watch your funding dwindle while it feels like you’re making no incremental improvement forever. There’s a name for this in startups: the valley of despair. It’s not just me—Querio has experienced it too.

This comes from the Dunning-Kruger effect. You begin without understanding the difficulty, thinking it’s easy, even when you lack skills. Then you see how challenging it is and how unprepared you are, and you feel discouraged. Later, you gain abilities but think the task is harder than it is and doubt yourself. Querio has gone through these stages as a company and a product. I have too. Fundraising calls were simple to get at first, then closing a deal became extremely hard. Candidates agreed to interviews easily, but finding the right person was tough. Now, with good funding and the best team we’ve had, I face a new challenge: learning to manage a tech company as a first-time CEO while making Querio the top AI agent.

I have investors who trust us with their money, employees who depend on Querio for their jobs, and customers who rely on us to solve their issues. Sales began well again—growth is double digits every month, no worries, anon investor—but scaling it is getting complicated. My team knows more than I do in many areas, and it’s humbling (yes, team, I’m serious). Customers need an experience that delights them, which required Querio to push through its own valley.

This is where the AI agent valley of death starts. Creating an AI MVP is very easy. Honestly. These tools generate perfect SQL and Python without effort. Coding feels fast! You can build web apps in minutes. It seems like production-grade AI workflows are close. They aren’t. AI follows the 80/20 rule perfectly. Controlled demos are straightforward, but real-world databases are chaotic—tables don’t connect, fields are missing, data is outdated. The market doesn’t fully see this yet, but Querio’s team has worked hard to overcome it.

We began with a SQL-only AI tool. Now Querio handles SQL and Python, offers manual notebooks for technical users, co-pilot mode for detailed work, and helps non-technical people get answers quickly. It manages large datasets, stays accurate, follows governance rules, and remains enjoyable to use. That took a lot of effort. AI is simple for the first 80%, but the last 20% is incredibly hard. Customers value that we don’t exaggerate; we provide results. We’re one of the few AI agents, and even fewer AI-native BI tools, used in production by real companies daily.

Querio has evolved from a SQL only AI call, to a complex agent that supports SQL and Python, that has manual and co-pilot notebooks for your deepest work, and the first line of defense for the non-technical side of the business to get all their data needs met. The effort it has taken to build a product that can handle full workflows, huge databases, stay reliable and accurate, allow governance, and deliver an amazing experience has largely been an exercise of pure grit. AI is easy for the 80, but it is SO hard for the 20 it’s ridiculous. This has largely gained us respect with our customers, as we don’t overpromise and we over deliver. We can proudly say we are one of a handful of AI agents, let alone AI native BI tools, that are used in production by real companies every day.

Writing this makes me think. Querio and I have already moved past so many valleys of despair. Maybe I’m just doubting myself too much again. If your business is in its own valley, struggling with data or trying to use AI for real value, contact me. Book a demo on the website or email me at rami@querio.ai.

In fact, tell me you’re in the valley of despair and i’ll extend your pilot by 6 weeks so we can get you out of it before you commit.