DIGITALOCEAN
AI/ML Product Strategy
TL;DR
A quarter-long research effort aiming to more clearly define the problem space and business opportunities for DigitalOcean within the burgeoning AI and Machine Learning industry.
The Opportunity
My team, the Design-Led Engineering team, was tasked with helping identify opportunities within the AI and Machine Learning development space to help inform DO’s long-term strategy within this subject matter. Our team conducts a variety of evaluation types, with the most in-depth one being what we call a Thematic Evaluation. The thematic evaluation aims to deliver a broad understanding of the problem space, business and user opportunities, an overview of the landscape, and next steps.
Evaluation Exercises
We had two major evaluations going on in Q2 2023, so I took on the bulk of the AI & ML evaluation tasks. For this evaluation, I conducted the following exercises:
Problem statement development
Comparative landscape (with the help of this amazing interactive landscape by Matt Turck)
State of the industry / Current events
Vocabulary
User journey mapping
Opportunities within the DO ecosystem for each product/service we offer
Integration opportunity mapping to business outcomes
In collaboration with research partner, designed user survey which resulted in {N} completed surveys
Outcomes
The output of this evaluation was a 67-page long detailed report covering everything above, as well as a presentation to stakeholders, that now lives in the DigitalOcean strategy library for anyone to reference.
Additionally, the most important outcome of this effort was that findings directly influenced the due diligence period happening in tandem during the quarter for DigitalOcean’s eventual acquisition of AI and ML Ops platform Paperspace. While I was not directly involved in the acquisition and due diligence period, it’s exciting to know my research directly influenced the decision to acquire Paperspace and continues to influence how we choose to prioritize integrating Paperspace’s offerings into the DO ecosystem.
Some interesting information we discovered
Even in its nascent state, we identified the future as not being training and fine-tuning, but in applications and inference. Specifically, we identified GenAI as an opportunity that was welcoming to both new-to-AI as well as seasoned AI professionals seeking to embed AI into their organizations and existing products.
With the industry moving at the speed of light, users need trusted documentation to know what type of AI solution is the best fit for their goals, why AI is or is not a good fit for their org, how to get started, and when they need to scale.
Unexpectedly, small-to-medium sized businesses are price conscious and want to know they can trust the products they’re shelling out for.
Partnerships with industry leaders, such as the foundation model developers, model repositories like HuggingFace, and partnering with vendors who offer AI services on top of existing DO infrastructure, are key opportunities to offer a variety of solutions for our users.
What happened next?
The following were projects or initiatives that immediately stemmed from this work: