"Captain's Log, Stardate 2022.6. The GISKARD’s sensors are detecting a new life form! Could it be a new ally in our battle against ML Bugs?”
Yes, Giskard is changing identity! We are keeping our name, but the way we present ourselves is changing. We have a new value proposition and a new logo.
Why did we decide to change our identity? How did we manage such transformation?
Giskard is all about building in the open, so here is our story. We hope it is interesting to the ML community and useful to fellow entrepreneurs on the Product-Market-Fit journey.
💡 Finding our new value proposition
A company starts with a value proposition. When we started Giskard last year, this proposition was “Quality Assurance for AI”.
A few months ago, we released our Open-Source product and we started to work with our first design partners. Getting user feedback on real Machine Learning use cases has been incredibly useful to better understand the priorities of ML teams today.
We ran over 100 interviews with ML practitioners, documenting all the insights that we received, to rebuild our value proposition from the ground up. We believe the best way to build a product is bottom-up, starting from users’ feedback.
We discovered that the Number 1 challenge of ML teams today is Time:
- At the development stage, getting feedback & buy-in from business stakeholders on ML models is very time-consuming.
- Deploying a new ML model version involves a lot of repetitive manual actions.
These inefficiencies in the ML development process create mounting frustrations among Data scientists, ML Engineers and business sponsors. This can lead to staff departure, and ML projects not meeting expectations.
As a consequence, the challenge of Quality should come after fixing the issue of process efficiency. You can think of this as two steps in the “Time-Quality space” below:
First, make your ML development process more efficient, at the same level of model quality. Then, you will be able to reach higher ML model quality, quicker!
If you do that in the reverse order, you risk spending too much time & resources for only a small increase in quality.
After we understood this, we realized that “Quality Assurance” is not the best way to describe our value proposition. While Quality is the end goal, what matters to ML teams is the process to get there.
Following that analysis, we understood that our value proposition to ML teams should be to help them deliver ML products, better & faster.
This is exactly what Continuous Integration & Delivery (CI/CD) is meant to do for software products:
CI/CD bridges the gaps between development and operation activities & teams by enforcing automation in building, testing and deployment of applications. [...]
Automated tests verify the software functionality, and automated deployment services deliver them to end users. The aim is to increase early defect discovery, increase productivity, and provide faster release cycles. [...]
The CI/CD practice forms the backbone of modern day DevOps operations.
That's why we chose to explain our solution as an Open-Source CI/CD platform for ML teams.
🎨 Designing our new logo
In parallel of this search for a new value proposition, we worked on a new visual identity. We wanted to keep the name Giskard, which comes from a unique robot in Isaac Asimov’s science-fiction novels.
But we were not satisfied with our logo, a green robot.
We wanted a logo that was:
- Very distinctive
- Less robotic & more relatable
- In line with our value proposition
After a lot of brainstorming, we understood that we needed to find an animal mascot. Symbolically speaking, animals naturally carry meaning & values. Most of our favorite developer tools company have an animal mascot as logo: GitHub has its Octocat , Airbyte its Octopus, Dataiku its bird, Datadog its 🐶, etc.
This was easier said than done. We actually went through 8 weeks and 4 cycles of iterations before having our Eureka moment!
During Iteration 1, our idea was to emphasise vision, wisdom & collaboration:
- Lynxes have sharp eyes
- Owls are wise and see in the night
- Meerkats spot predators and collaborate to protect their tribe
We brainstormed for several weeks without being able to converge. We wanted to reach a consensus among co-founders, without any doubt. Fun fact: we discovered new animals during that ideation process. Do you know the Dipodoid?
One morning, our co-founder Jean-Marie had a brilliant idea and proposed the Turtle. It was an instant hit, and we all voted in favor!
The turtle expresses wisdom and knowledge. It resonates with robustness. And, if you read The Tortoise and the Hare Fable from Aesop, you know its the animal who wins the race! This fable symbolizes the quest of ML teams for both efficiency & quality.
After our minds were set on the Turtle, 3 more iterations were needed to find the perfect shape & style.
First, we changed from a land to a sea turtle, in order to emphasize speed. Our designer found the perfect species: the Leatherback Sea Turtle. It is the fastest-moving turtle, at 35 km/h in the water! And that’s how we landed on this beautiful, fast-swimming turtle logo:
You can see our new value proposition and logo in action on our new website:
We are actively working on the website to make it even more crisp and convincing. Please don’t hesitate to tell us what you think and report any bugs at email@example.com.
All of this would not be possible without these amazing people:
- Jonathan Da Costa at Index Studio, our incredible designer, who thoughtfully accompanied us and delivered throughout the entire creative process.
- Arthur Vu, our Webflow genius, who translated the design vision and put his attention to every pixel of our website.
- Tan Yucel, our brilliant illustrator, who draw all the turtles on our website.
- Vincent Huguet, CEO of Malt, who kindly advised us during our animal ideation process, by sharing his stories of Dromadaire.
- Chris Schagen, our amazing Go-to-Market coach, who gave us the methodology and advice we needed to better define our positioning statement
We thank you from the bottom of our turtley hearts!
Also, I would like to thank all 1445 of you for following our company on LinkedIn. A special note for the 164 who starred our Open-Source repository. Stars really help make our project more visible and encourage contribution. We are very grateful for this initial traction. Every star counts!
If you read this and haven’t left us a GitHub star yet, all it takes is one click!
👉 Our GitHub: gisk.ar/github
Our focus this summer is to polish our product, and to work closely with our design partners to ensure it performs well and creates value. We are hard at work to smooth the installation process, squash the bugs, and improve the user experience.
We have a very rapid release schedule, to make sure these enhancements are rapidly in the hands of our beloved users. We actually just released the new version of our Python client (1.2.0) today. It greatly improves model and dataset upload validation!
A great product is the result of a great team.
So our second focus is also to recruit top R&D talent. Are you interested in joining the adventure? Do you know someone who could be a great fit?
👉 Apply here & refer candidates: https://gisk.ar/jobs