Smarter forecasting with Myst AI— Project Spotlight
No one can truly predict the weather, not even Punxsutawney Phil. With an accuracy rate of about 40%, the storied groundhog isn’t a great harbinger of Spring.
Humans, too, struggle with prognostication, which partially explains why more than 400,000 Texas energy customers woke up on Groundhog Day 2023 without power. A brutal ice storm hit the state hard, and utility providers didn’t factor climate change into the year’s projected energy consumption — at least, not precisely.
Traditional forecasting methods weren’t built for the type of extreme weather that turns Dallas roads into ice rinks, and prompts rolling blackouts across Northern California. It’s a particularly vexing challenge that’s only going to get more complicated as our climate continues to change.
Enter Myst, a time series forecasting platform leveraging artificial intelligence models that can help energy companies make better decisions about near-term energy needs. With Myst, customers can easily build and deploy forecasting models to increase accuracy, reduce energy waste, and lower financial risk. The company works with a variety of companies in the energy sector, including utilities and retail energy providers, as well as renewable generators, distributed energy resources (DER) providers, and grid operators.
Despite the varied audience for its product, Myst initially focused on a technical end-user base — the people comfortable with handling complex data sets. When they came to O/M to develop an MVP from scratch, we encouraged them to expand their vision for the product.
Instead of creating a product that requires a PhD in data science to operate, we recommended a visual canvas that allows non-technical users to build a technically sound data model. It pulls in third-party and proprietary data sources, inputs those into machine learning models, and outputs a time-series forecast that can be built upon like a tree to create a compound forecast.
Myst co-founder Pieter Verhoeven wasn’t immediately sold on the team’s no-code approach, but through research, we collectively determined that this was the most effective approach. “Look, you’re gonna have a bunch of people who are not writing code who are going to want to interact with this in some way — give them the tools to do it,” recalls OM senior designer Ted Schaller.
Myst’s trust in O/M’s perspective paid off. With our designs in hand, the startup quickly built its platform, secured $8 million in venture funding, and added several companies to its customer base who quickly saw efficiency gains. Perhaps most notably, they completed an exit in just two years. Cloud-based data warehouse company, Snowflake, acquired Myst with plans to expand the platform’s applications across a number of industries.
Such rapid success depends on a willingness to collaborate early and often. Schaller says clients must be willing to partner with O/M to achieve their biggest goals. For example, we enjoyed ongoing, open dialogue with Verhoeven throughout the engagement, including a two-week discovery process that required us to become experts in the energy and time-series modeling space. In a traditional client/agency relationship, the client gives an agency marching orders, the agency retreats and develops ideas, then the agency returns weeks later with recommendations that might be at odds with the client’s expectations. All of that runaround is avoided when client and studio are locked at the hip. Plus, when you work with O/M, you gain valuable, alternative perspectives that uplevel products from good to exceptional.
O/M goes to great lengths to help clients understand how their products might be received by a wide range of people. The studio is careful to understand each problem and advocate for the best possible solution, even if it means suggesting a path that might contradict a client’s initial assumptions.
AI is getting a lot of buzz right now thanks to Chat GPT, but O/M is hardly new to the technology. In fact, the studio has been working with clients whose offerings leverage AI in a variety of ways since O/M’s inception. Sometimes AI is factored in as a supportive technology to power routine tasks more effectively; increasingly, clients are building transformative AI products, including Myst.
According to McKinsey, the level of investment in artificial intelligence has increased alongside its rising adoption. The percentage of large enterprise organizations dedicating 5% of their digital budgets to AI has gone from 40% in 2018 to 52% in 2022. That number is expected to jump to 63% over the next three years.
The earlier we can get in the room with people who are thinking about these types of technologies, the more empathetic and inclusive the end products can ultimately be.
As the applications and implications of AI broaden and intensify, so too do ethical questions of how the technology might be abused. There are ways to embed guardrails in products to encourage responsible AI, like increasing friction to lower the chances of someone unintentionally misusing the technology. There’s no way to ensure someone won’t use their products with bad intentions, but by adding friction in the right places we can certainly make it harder.
“The more opportunities that design can get up close with AI, the better.” says Ted Schaller, “As an industry, the earlier we can get in the room with people who are thinking about these types of technologies, the more empathetic and inclusive the end products can ultimately be.”
