2022 in Review

As 2022 comes to a close, we at Jamie have a lot to reflect on. It has been a year filled with growth and change for our company, as we launched two new products and welcomed talented new team members. But it has also been a year of learning and adaptation, as we navigated the challenges and opportunities that arose along the way. In this blog post, we'll look back at the highlights of 2022 and offer a glimpse into what the future holds for Jamie as we move forward.

Did you know?

Jamie creates the most accurate meeting notes for free. No catch, just catch-up on what you missed!

Try Jamie

TeamTap: the start of our new chapter

In early 2022, Benedikt and I started working on a new idea we believed was solving some real problems people in the distributed era of work were experiencing. This started out as a concept we discovered when studying the world's best remote teams and the way they worked. Companies like Github, Zapier, and Automattic were leveraging internal microblogs to establish more transparency and accountability within their companies. As they were relying on custom-built solutions for this, we discovered that there was no good ready-to-use solution on the market. Easier said than done, we started digging deeper.

As any startup playbook suggests, talking to users and deeply understanding their problems and needs is the first step before starting to commit yourself to one idea. And that's exactly what we did. We talked to more than 100 companies about the problems they faced in the distributed era of work and eventually pitched our first version of TeamTap. The result? More than 70 signups to our waitlist. We were ready to start building.

Fortunately, as we were seeking feedback from many people, some ended up being interested to invest in TeamTap. The commonly shared presumption to "Ask feedback, get money. Ask money, get feedback" turned out to be true in our case. So by mid-2022, we closed our pre-seed round with great angels on board who closely supported us on our journey to this day and beyond. Also, we got the chance to pitch TeamTap at the ATEC startup conference and were able to win 1st place - further validation in our pursuit of this idea.

After building the first version of TeamTap within a couple of months, we launched it to a select cohort of folks from our waitlist. After closely watching how they explored the product, we realized that there was one key action we had to focus on to make TeamTap take off and gain adoption within a team: getting people to share posts. As one of TeamTap's core use cases is bridging the gap between product and sales teams by sharing insights from customer calls, we started thinking about how we could make the act of sharing those insights from calls easier. That's when the idea of building a feature that writes summaries of your calls automatically first crossed our minds.

After validating our idea for a meeting summary feature in a couple of customer calls, we quickly decide to build a first prototype of this. It started out as a chrome extension that copied the captions from a Google Meet call and then used this to generate a summary with GPT-3. For the first prototype, this worked well and we were excited as we believed this could significantly increase the adoption of TeamTap as more people would start sharing by removing the friction of manually needing to write summaries.

Pivoting to jamie, an AI meeting summary assistant

When sharing this idea in customer calls, we noticed a pattern in the way calls went: every time we mentioned the automatic meeting summary feature, the conversation completely changed. Customers were no longer talking about TeamTap or their challenges in distributed work. All they cared about was how they could get automatic meeting summaries in their life. After gathering those data points and also discussing this internally and with our investors, we came to the conclusion that this meeting summary feature is not a feature, but a product itself.

Despite the first data points clearly pointing towards the fact that a meeting summary product has enormous demand in the market, we wanted to ensure to not pivot too quickly away from our initial idea with TeamTap as this slowly started to work out as well. Hence, we decided to do a company off-site with our small team to launch an experiment and gather more data on how much automatic meeting summaries were actually wanted in the market.

Our plan was quite simple: create a landing page for this new product paired with a social launch strategy on LinkedIn while optimizing for the # of signups generated for this product.

Launching jamie: an ai-powered meeting summary assistant

On Thursday, the 13th of October it was time. After just 5 days of preparing this experiment, we launched jamie to the world. Within 48 hours, we got hundreds of signups. This exceeded the threshold we set ourselves for abandoning this idea so the way forward was clear: we had to build this product quickly.

Combined with the launch, we offered people who signed up to reserve a personal onboarding 8 weeks out. This allowed us to test how badly people wanted this kind of product while also providing us with an ambitious deadline to build a working product that we can give to customers by then. More than 30% of signups reserved their onboarding immediately so the game was on.

The first version of jamie

A good rule of thumb for building minimal viable products (MVPs) quickly is to focus on the product's essence. This will allow you to quickly learn whether you are onto something. In our case, this meant that we wanted to build the first version with a carefully selected set of features:

• A native application for macOS & Windows
• Audio capture of the microphone
• Transcription of audio to text
• Summarisation of long-form text
• Delivery of summary to users in an easy-to-edit way
• Rating mechanism for the quality of summaries

Easier said than done, we got busy building this first version. In early December, we were able to onboard dozens of early customers to this first version.

One lesson we were able to learn (again) was that despite all the good intentions of great UI/UX, nothing beats the learning you get when giving your product into the hands of real people. After just the first days, we already had a good idea of how we needed to improve our product based on the feedback we have gotten. Many improvements of which are already live in the current product. The underlying feedback we got, however, was that despite the quality of summaries already being impressive in some cases, it is still not perfect - a finding, that had to be expected. So moving forward, our primary focus is on improving our deep learning pipeline to deliver the world's best automatic meeting summaries to our customers.

Did you know?

Jamie creates the most accurate meeting notes for free. No catch, just catch-up on what you missed!

Try Jamie

The way forward

As you've seen, 2022 has been filled with a lot of action, learning, and iteration over our core products. Despite not having reached product-market-fit entirely, we are closer than ever and confident to be on the right track. Together with a talented team, we are excited to push the boundaries of how artificial intelligence can be used by anyone in a value-adding manner. All in the pursuit of delivering the world's best automatic meeting summaries to our customers. Moving from our overall intent to more tactical projects, the weeks and months ahead will be filled with exciting projects.

On the product side, we will ship two major new features to jamie next to countless smaller, yet important, improvements.

Firstly, headphone support will come to jamie in Q1 2023. This means that now not only the microphone input but also your system audio can be captured to generate a summary. This is crucial when using headphones or AirPods in your calls. Further, this will improve the quality of summaries as the transcription of the audio will become more accurate as the audio from other participants can be captured in higher quality.

Secondly, we will be releasing a major improvement to our summarization deep learning pipeline: speaker diarization. Also in Q1 of 2023, we will include a powerful way to detect the identity of speakers in a meeting based on their voice. This yields better structure in transcripts that will in turn improve the accuracy of the summaries significantly. The best part is that this is entirely based on audio so you can still use jamie in hybrid or offline meeting settings. Next to this, we are experimenting with ways to add a "memory" to jamie for every individual user so that the names of speakers can be remembered and hence the experience of using jamie becomes more magical by the day for users.

For new customers, the beginning of 2023 will look similar to what it has until now. We are continuing to follow our invite-only onboarding plan to ensure we carefully craft the highest-quality product together with our customers. Once we believe that jamie is robust enough to perform well in a range of settings, we will open it up to anyone. But for now, the best way to get access is by signing up for our waitlist. If you want to have an early preview of what jamie is capable of, feel free to DM any of our team members. We are happy to give you a bump on our waitlist.

All in all, 2023 will be the most important year in our company's history. We are grateful for all the support from customers, investors, and partners along this journey and can't wait to continue building a magical experience for our customers.