Dec 31, 2023

Looking back on 2023

Hey there 👋 I'm building CodeTrail, which helps engineering teams document and share knowledge close to the codebase with no friction. If you're still using Notion, Confluence or Google Docs to document your engineering work, give it a try and let me know what you think!

Tonight, 2023 is coming to an end, and I’m still wondering where the last months went. I feel like a broken record repeating that time is accelerating year by year, so naturally I’m going to say that the past twelve months felt like the busiest yet. Platitudes aside, let’s recap what happened in 2023.

I started the year by departing Hygraph. Together with finishing my studies at TUM, this was part of an effort to wrap up old commitments to focus on the next arc. Making the decision to leave my job at Hygraph also forced me to critically judge the potential trajectory I could reach when trying something new: I am confident that staying in my previous position would have limited my pace of learning and by that, my growth.

So, what have I done with the time I freed up? Let’s dive into it.

Validating Anzu

Last year, Tim and I were building the first iterations of Anzu, a product offering building blocks to fast-moving teams, covering user management, product analytics, and CRM. We got in touch with Magnus from the venture team of TUM to discuss potential funding options to cover our operating costs. Ever since, Magnus has given us incredibly candid feedback on every step of the journey, which we’ve found invaluable.

Pretty early on, he convinced us that without clear customer interest, we wouldn’t be able to raise any funds. While the potential of a business shouldn’t always be judged by the ability to fund it, this prompted us to stop building and step back to think about our approach. We switched to customer validation and outreach and interviewed dozens of teams.

Originally, Anzu was created from the idea that teams would need strongly connected building blocks to save time connecting dozens of tools and reduce knowledge fragmentation as a side effect.

All teams we talked to had lukewarm interest, at best. Big teams already used one or more tools we tried to replace and weren’t interested in replacing the entire stack, small teams were harder to reach and had a high risk of failing to attract customers (we were in the same boat).

Since Engineering teams weren’t particularly interested in reducing fragmentation, we decided to switch the ideal customer to sales teams. By collecting data from product analytics and user management in one tool, we could offer more relevant signals to sales teams, alerting AEs about churn early on, highlighting highly qualified leads, and so on. Unfortunately, it’s harder for engineers to sell to sales and our target group didn’t appreciate the value of better information (one interviewee outright said they’d rather have their sales reps just pick up the phone and cold call people than spend time on segmenting leads).

In early summer, we acknowledged that offering Anzu to sales teams wouldn’t work either, so we stepped back and thought about the ideal customers we interacted with in our past jobs. Sales teams had just been disqualified, engineering teams experienced different problems altogether, but there was another customer group. See, Tim and I worked in software agencies early in our careers and experienced project management and customer interaction flows.

Uncovering Software Agency Problems

We flipped our previous script upside down and started interviewing agency owners and employees without writing a single line of product code. We quickly understood common pain points like handovers, communicating deadlines with customers, and aligning both sides of a project. Interviews repeatedly surfaced similar issues. Tim created face-to-face workshops that helped us get closer to the teams we were about to sell to.

In one workshop, we got an important signal. The problem we looked at originated in cultural issues and conflicts inherent in the customer/agency relationship.

Paying attention to this important insight saved us years of commitment to building a product that would ultimately fail to solve the very real issues those teams we talked to face in every project, so we put the agency project on hold.


Around July, we had just finished multiple attempts of ideating a customer profile, collecting potential customers, reaching out to teams, and preparing and running interviews and workshops, which all ended with missing signals to continue. We followed an incredibly lean approach and dissected problems to get the confidence to move on.

Interestingly, we always tried to bundle multiple product categories or think about first-mover products rather than creating a better iteration of an existing product with an existing market. More on that later.

Without any clear direction, we took some time to catch up on the latest developments in the ML space and released Gradients & Grit, a publication focused on bridging applied machine learning in production scenarios for software engineers. To this day, we released two full products deeply integrated with state-of-the-art LLMs and techniques like RAG.

One more try

In late September, Tim and I embarked on our annual offsite in the Netherlands. One of the key points I wanted to figure out was how we should continue. We could call it off and get back to working at some company to figure out more real-world pain points, or we could continue working toward discovering an unmet need.

Interestingly, once we discussed these options, we that our goals were fully aligned. Working in another company might work to discover different needs and problems, but I always wanted to build the next big thing together with Tim he felt the same way. We agreed to keep going.

We recalled our previous attempts at products and problems we experienced in our previous roles as software engineers and team leads. One area that kept coming up was sharing knowledge. LLMs can complete code by inspecting the existing codebase. But decisions and documentation not written down, cannot be used. Context is more important than ever. Complete, up-to-date documentation can turbocharge a repeatable onboarding experience for growing teams.

This is what we’ve been working on since October, and I cannot wait to share more details in the coming weeks.

Credits

As always, this year wouldn’t have been possible without the encouragement and support of friends and family. I’ll miss people if I try to list everyone, so let’s focus on the amazing people who kept supporting me through every pivot this year. First of all, thanks to Tim for being the best co-founder I can imagine. Thanks to Michael from Hygraph for being a mentor to lean on when tough decisions had to be made. And thanks to Magnus for never stopping to give honest feedback.

On to 2024.

Thanks for reading this post 🙌 I'm building CodeTrail, which helps engineering teams document and share knowledge close to the codebase with no friction. If you're still using Notion, Confluence or Google Docs to document your engineering work, give it a try and let me know what you think!

Bruno Scheufler

At the intersection of software engineering
and management.

On other platforms