Last week, we delved into the opaque and complex world of utility filings, exploring the hidden — and valuable — knowledge contained within integrated resource plans (IRPs) and rate cases.
We concluded that post by sharing some ways we are helping make this information more accessible to the clean energy developers and investors that use these documents to inform business decisions. Today, we’re going to share some examples.
Let’s say a hypothetical energy storage developer wants to find utilities with a growing peak load. These utilities may choose to add battery capacity to manage the peak days of the year, and avoid building an expensive new natural gas plant in the process. Our energy storage developer needs to see the load growth forecasts utilities publish in their IRPs!
Ordinarily, they would have to search each individual utility website, download a massive PDF, and control+F their way through hundreds of pages to find the answer. With Halcyon, you can just ask:
As usual, the source is cited and the information is easily verifiable. But we’ve found that investors/developers usually don't ask these questions in isolation. Rather, they often want to ask the same question of multiple IRPs that represent multiple utilities in various locations. Inquiring businesses aren’t just vetting one customer — they’re gathering intelligence across a range of utilities in order to get a sense of relative rates and identify potential partners that are the best fit for their specific conditions.
Solving for this instance, where investors and developers have the same type of question applied across markets, lends itself to a specific application of Halcyon’s technology. A user might be comfortable running the same query himself or herself 5 or even 10 times; it is inefficient, and an unfair ask, to make him or her do the same thing 50 or 500 times.
Instead, we automate that capability to interrogate any set of unstructured data (like IRPs or rate cases) so that we can rapidly run the same query many times. The term for this is batch processing and it is fast, high-resolution, granular in detail but also comparative across fields.
In addition to our natural language query interface, we’ve been working on automating these batch processing capabilities. What you see here is a structured output collated from unstructured inputs — in this instance, a table comprised of data pulled from text documents. Utilities and their respective IRPs are the Y axis, and the relevant information within those documents represents the X axis. Each data point contained within a specific cell is powered by an underlying query.
This is an early prototype, but it’s easy to imagine how one could extend this framework and customize outputs with any number of fields — any type of company or institution, any type of data. Once you’ve established the interrogable universe as rows, the specific questions themselves become the columns.
But as we know, the information velocity in energy and decarbonization is relentless. As of Wednesday, Oct 23, over the last 10 days, FERC has published a total of 348 newly created dockets. Once you factor in state and local-level agencies, any structured output you build will quickly be out of date.
What if there were a way to directly receive automated updates, delivered straight to you?
That’s exactly what we’re building. We’re now allowing people to receive a daily email notification when there’s new information added to our catalog, starting with Dockets, which are part of the rate case process. Eventually, we’ll make this customizable — there’s a lot of noise out there (and we’ve already written about how much we dislike bad emails), so the ability to filter and self-select is critical.
And while (good) emails are useful, wouldn't it be even better if you could create your own structured output like one of these tabular sheets that automatically updated with new information once it was published? After all, most analysts we talk to would use an email notification to stay current, but still have to click the link, find the right nugget, and then copy/paste that into their own work product.
We’re working on it! If that (or anything else you’ve seen here) sounds compelling, we’d love to talk.
Comments or questions? We’d love to hear from you - sayhi@halcyon.eco, or find us on LinkedIn and Twitter