At this time last year, our startup was in freefall. We were convinced we’d built the perfect solution for a major industry pain point—an API load testing tool inspired by my experience as CTO of Kid Cudi’s live-streaming app, Encore. But after countless conversations and false starts, the market gave us a brutal reality check: load testing simply wasn’t the urgent priority we believed it to be. That discovery triggered a major pivot to Jolt, our AI coding assistant for large production codebases. The process was tough and left plenty of scars, but it shaped me into a stronger founder. In hindsight, the lessons seem obvious, yet they felt anything but when we were in the moment. Now, I want to share our story so anyone building new products, whether at a startup or an established company, can learn from our experience.
The idea for our original product took shape during my time at Encore. Whenever a major artist performed or tweeted about us, traffic surged, and we had to ensure we stayed online. Existing load-testing tools were clunky and failed to cover the protocols we needed. With that in mind, I spoke to more than 20 teams, collected a couple of letters of intent, and raised money for the new company. It all seemed so promising at first: we had clear evidence of a common pain point, and we charged ahead to get our MVP out the door.
Once our MVP was ready, I hit the ground running with customer outreach through my personal network and LinkedIn. Surprisingly, securing initial meetings wasn’t the problem. I was booking new meetings at about an 8% success rate, which felt promising. But curiosity didn’t translate into urgency. Prospective customers generally agreed load testing mattered “in theory,” but it simply wasn’t high enough on their priority lists to justify time or budget. Follow-up meetings fell off sharply, timelines kept slipping, and most critically, we didn’t have genuine design partners to develop the product with. We were building in a vacuum, hoping interest would magically convert into usage.
We were facing one of the most common and dangerous pitfalls: false starts and false hopes. Prospective customers naturally want to be nice. They may even get excited about your product. But until they have skin in the game, like a signed contract or significant time commitment from their team, it’s meaningless. Yes, this sounds harsh, but as you’ll later read, you must be brutally honest with yourself to succeed.
After various attempts at marketing and sales, we had no real traction. We had talked to over 80 potential customers, and nearly all of them saw load testing as an occasional need, not a “must-have.” Our team had poured almost a year into developing a product, coding like crazy, but not seeing meaningful usage—an emotionally draining place for any team to be. Thankfully, we’d kept our burn rate low, so we still had over two years of runway. That cushion gave us room to make a drastic change: to acknowledge the signals and pivot. One of the most important things an early-stage startup can do is stay hyper-attuned to what customers are telling you and adapt quickly when it’s time to change course.
A successful pivot hinges on three essentials: a concrete plan, transparent communication with your team and investors, and brutal honesty about what’s not working. Our plan was straightforward:
- Run an aggressive customer and problem discovery process. Meet with every engineering team we know. Meet with at least 5 new teams each week.
- Experiment and hack on our own ideas. The next generation of LLMs came onto the market around this time, so we had plenty of topics to explore.
- Establish a tight timeline for the discovery process, the pivot, and finding pilot customers. Parkinson’s law always applies.
We built great relationships with folks during the load-testing days, and many of those teams were happy to meet again. During these discovery interviews, we kept the discussions open-ended and gave people space to share about their most pressing issues. I ran through these topics in the interviews:
- Can you share more about your role and your team’s responsibilities?
- What are the most critical projects you’re taking on? Why are they important and what are the outcomes you’re aiming for?
- What are the company’s technical or engineering initiatives this year?
- What are the challenges or pain points your team is facing? What would be the impact of solving each one?
- What is the most annoying thing you/your team have to deal with every week?
- If you had a magic wand and unlimited resources, which single problem would you fix first and why?
Only after exploring these points would we float our own concepts and ideas. We took meticulous notes during every conversation. Every Friday, our team had a “Weekly Wrap” session, where we’d discuss learnings from the week. This feedback loop and open-ended conversations would reveal where the real opportunities are.
This discovery blitz combined with a perfect storm of market forces and user feedback pointed us toward what is now Jolt. We heard the same complaints over and over: engineers were frustrated by AI’s lack of codebase context. Copy-pasting files into a chatbot or manually picking context files is tedious and a non-starter for larger projects. The existing AI tools simply did not understand or work on real-world codebases. At the same time, most engineering leaders had an initiative to define an AI strategy for software development. Coincidentally, our own internal research was focused on AI and complete codebase understanding. We ran experiments to figure out how to get AI to understand a large codebase in its entirety, and then explore what we could achieve with it.
Realizing we may have tapped into a hugely important problem to solve, we assembled a small group of early adopters to build alongside us. Every decision was validated by their feedback; if it didn’t help solve a problem they mentioned, we didn’t build it. They deserve a ton of credit for their belief in us and for putting up with alpha versions that were rough around the edges. After months of tight feedback and iterative development, we had something that felt truly special. That milestone led us to a soft launch of Jolt, and the early response confirmed we were onto something.
Repeated usage is the clearest sign that you’re building something people genuinely need. Pricing optimization, marketing campaigns, and fancy features will not make a difference if people do not care about what you’re building. If your product solves a real, painful problem, early adopters will overlook friction or bugs. Users' willingness to endure inconveniences is a powerful signal that you’re on the right track. With our load-testing tool, we never saw this phenomenon. But when we released Jolt, people kept coming back. They were excited about our updates. No amount of marketing or optimization can replicate that kind of genuine, organic engagement. It’s the first glimmer of the startup holy grail: product–market fit.
Today, we’re in a vastly different position than we were a year ago. Usage for Jolt keeps climbing—not just the number of users, but on a per-user basis. Getting here took relentless iteration around our users’ needs. Our earlier challenges shaped us into a stronger, more user-centric company.
Case in point: one early assumption I made turned out to be completely off: I didn’t think developers would be interested in a “Chat” feature. I was convinced our “Implementation Plan” would be a winner. But our early adopters kept asking for a chat workflow. And they were right. Now, 98% of Jolt’s usage is through that chat feature, and honestly, 99% of my own Jolt usage is too. That single feature shift was a complete game-changer.
We launched our public beta last month, and the response has been beyond what we hoped for. We owe that success to the hours we spent with our users, validating every step to ensure we built something that truly solves their problems.
To sum up our key takeaways:
- Be aware of false starts. Just because a potential customer expresses excitement or interest does not mean they will buy your product. Get to a concrete yes or no as quickly as possible.
- Invest in problem and customer discovery. Have open-ended conversations, let people tell you what truly matters to them. Save your own ideas for later in the discussion.
Build hand-in-hand with real users. - Don’t develop anything in a vacuum. Your earliest adopters should be your guiding light for product direction.
- Usage is the ultimate litmus test. If people keep coming back, you’re onto something. If they don’t, revise.
We’re going to keep iterating closely with our users; every new feature or project must address a genuine customer request. I hope our story has been helpful and sheds some light on the rollercoaster pivot process. If it resonates or you’d like to connect, please reach out. And if you’d like to give Jolt a try, sign up here.
