In operations-heavy environments such as last-mile delivery, e-commerce, or marketplaces, 90% of incidents are detected only through customer complaints. This reactive approach forces more than 80% of operations teams to perceive their work as constant firefighting. Having experienced the challenges of managing operations in fast-growing companies, we set out to build a software platform that monitors and automates business operations – think Datadog meets Zapier. We’ve personally felt the pains caused by rapid growth, the dynamic nature of business operations, inadequate tooling, disparate data, and limited engineering resources. Through hundreds of interviews, we found validation for our vision. This led us to believe that a low-code, self-service tool leveraging the latest advances in the modern data stack and AI can alleviate the daily struggles of operations managers in high-transaction environments.
While we’ve managed to raise seed funding, build a great product, and satisfy more than a dozen customers, over time we’ve learned that several of our core hypotheses were incorrect, leading to a commercially unsustainable business model.
Hypotheses and learnings
Hypothesis 1: A low-code tool for operations monitoring & automation is a painkiller for urgent struggles, not a vitamin.
‍Despite nearly every operations manager we spoke with mentioning that they and their teams are overworked and constantly firefighting, we realized that most companies are adept at managing their core business operations. Particularly in today's economic environment, where growth has slowed, businesses sustain their current demand and pace. The remaining issues that drive the aforementioned perception are mostly nuisances without significant top- or bottom-line impact or very custom needs that are difficult to solve. To put it differently, in our experience, the most crucial monitoring and automation needs are already resolved. Essential, complex issues are handled by (mostly internally developed) custom software; and simple, standard ones are solved with off-the-shelf tools, sometimes connected via no-code integration tools such as Zapier. What’s left are “nice-to-haves” with no real urgency to solve, or important but “special cases” requiring too much customization for a low-code tool.
‍Hypothesis 2: Operations managers & teams are empowered to utilize a low-code tool to help themselves.
Due to the aforementioned firefighting, operations teams often lack the time to step back, assess challenges and craft solutions, especially for “nice-to-have” use cases. Even if they do find the time to work on important ones, those are often custom challenges (see above), for which they tend to lack access to relevant data, as it’s dispersed across multiple systems and databases. Assuming the data access and aggregation hurdles can be overcome, the complexity of the associated data schemas often poses another challenge for operations managers when it comes to monitoring and automation. As a result, we’ve found ourselves dependent on our customers’ product, data, and engineering teams to help empower operations teams to get value from our product. This added significant complexity, time, and the need for professional services – all factors that are challenging in the sales cycle of a SME-focused SaaS product.
‍Hypothesis 3: (Gen)AI is a game changer for supporting custom monitoring & automation needs.
‍One planned way to overcome some of the challenges mentioned above was AI. We were very optimistic that integrating data sources, aggregating custom data, and comprehending multi-source data schemas could be significantly automated given the recent advances of LLMs. While we found that AI performs fairly well with simple, standard data schemas consisting of a handful of tables, we struggled to achieve reliable results with varying data schemas containing hundreds of tables. Similarly, with integrations, we have seen unified APIs and LLM-aided products making initial strides in simplifying the process, but we have also experienced their limitations with custom or niche applications. However, for a horizontal platform that relies on bespoke operational data sources, the manual effort remains substantial. While we believe that some of these challenges, particularly the understanding of complex data schemas, will advance significantly over the next 24 months, this potential partial easing of efforts isn’t convincing enough for us to base the sustainability of our business model upon it.
‍Thought process
In the last few weeks, there were moments when we questioned whether we were throwing in the towel too early. However, after making multiple smaller and bigger pivots – across product, positioning, industry, and more – over the past two years, we have ultimately disproved our key hypotheses, and thus no longer have conviction in the operations monitoring & automation space. Even though our remaining funding would have allowed further pivots over the next 18 months, we are convinced that closing down is the right thing to do. Building a company requires resilience and stamina, but it also requires full conviction and the ability to learn. Continuing despite having disproven our core hypotheses wouldn’t make us better or more resilient entrepreneurs. Instead, it would mean prioritizing short-term face-saving over the long-term interests of our team, investors, and customers.
Hindsight is 20/20. While this harsh reality now seems obvious to us, it took many attempts and learnings along the way. These insights were only possible because of some diligent efforts that we are proud of. Firstly, we’ve been relentlessly hypothesis- and learning-driven. Every pivot, such as adding workflow automation to our monitoring platform, was based upon clear, often customer-inspired, hypotheses followed up by nimble experiments and quick MVPs. Fail fast, learn faster. Secondly, we’ve built a great team and culture, despite – or perhaps because of – our fully remote setup. A blend of personal team-building during quarterly get-togethers and well-defined routines, such as async and transparent communication via Slack and Notion, created a highly effective, inspiring, and fun environment. Thirdly, we were living and breathing the needs of our customers, whose ample feedback via both in-person or online meetings drove our product development.
As such, we want to thank our customers for trusting us during these early times. We are very thankful for the feedback you’ve provided and hope that – despite stopping our services soon – we’ve left a positive mark. To our investors, we are grateful for your unwavering support and belief in our vision throughout this journey. Your commitment and advice have been invaluable. Last but not least, we want to thank our team – Amal, Andrei, Csongor, Daniil, Flora, Juraj, Viktor, and Zalan – for joining us on this exciting and memorable ride. To anyone reading this: please do reach out in case you need a rockstar product & engineering team!
‍
What's next?
For now, our primary focus is on winding down Eversync and assisting our customers in transitioning to other solutions. As we navigate this process over the summer, we welcome conversations about our experiences with operations monitoring and automation or low-code/no-code tools, and leveraging AI & LLMs for supporting integrations and data comprehension. Additionally, we are happy to discuss the valuable insights we’ve gained into the e-commerce, last-mile delivery, and logistics industries. We are eager to share our learnings and equally excited to hear about any interesting projects you might have in mind.
Please feel free to reach out!
Zoli (LinkedIn, Email) and
Wolfgang (LinkedIn, Email)
