Toward Dataflow Automation
Jun. 10, 2019

Prefect Logo - Toward Dataflow Automation

Jeremiah Lowin
Jeremiah LowinCEO

Prefect has a culture of transparency, which involves sharing news — both good and bad — with our employees and investors. In keeping with those values and the spirit of open-source, we’d like to also include the broader community when possible. This post is excerpted from our May 2019 investor update.

There’s a difference between what a company makes and what it sells.

The thing it sells delivers value to customers; the thing it makes is the vehicle for that value. Notice that selling doesn’t necessarily imply “revenue.” For example, Google makes a search engine and sells attention; Chevrolet makes Silverados and sells Corvettes (more practically: Chevy makes cars and sells leases); Stripe makes a payments API and sells convenience. The greater the difference between what a company makes and what it sells, the more “premium” — and defensible — its products become, because the company is necessarily creating value greater than the sum of its parts.

This has become a guiding principle for our company. Properly designed tools can unlock disproportionate value, but only rarely is the tool itself so valuable. The history of open-source business models is littered with companies that never figured this out. Fortunately, we learned it quickly from our very first market research sessions back in 2017, and it’s impacted our strategy ever since.

What we make is an amazing workflow engine; what we sell is confidence. We provide a risk management platform that minimizes our users’ effort at exactly the moment they need to maximize their response. By leveraging Prefect Core, we are able to free time and energy that would otherwise be focused on exponentially complex maintenance. We call that “negative engineering” and our company’s mission is to eliminate it.

When we introduced Prefect to the public, we described a product vision that included two components:

One is our open-source framework, which operates like a hardware store: stocked with all the necessary components for building great data applications. The other is our platform logic, which we think of as the store manager: guiding users to the right tools and making sure their projects are successful.

Today we can say more succinctly: we make Prefect Core, the best tool for building data applications; and we sell Prefect Cloud, a powerful service for dataflow automation.

This “make/sell” distinction is particularly important at this moment, because we’re halfway through our platform launch. Prefect Core is out the door. Prefect Cloud is being prepared for its initial release to our Lighthouse Partners.

The open-source release of Prefect Core in March exceeded even our be

st expectations. In its first 24 hours, Prefect was the #2 Python project on GitHub and the #12 project overall. Since then, growth by any metric has been steady and positive, closing in on 1,000 GitHub stars.

We have developed an active user community and already welcomed code contributions from 10 of its members, not to mention issues, discussions, calls, and in-person sessions. Both online and off, we continue to discover that Prefect has been deployed into data stacks from startups to multiple Fortune 100 companies. We would like to find ways to better engage these “shadow” users who may not have had cause to open issues on our GitHub, but we are overjoyed that our efforts have resonated positively.

One encouraging datapoint has been almost immediate requests for a platform providing many of the features that we will deliver in Prefect Cloud, most notably a UI and distributed scheduling — both of which will be powered by Cloud’s GraphQL API. Accordingly, we have turned our attention to the Cloud launch, ensuring that even its free tier delivers everything our users expect.

A familiar pattern has emerged: when we prepared for the Core launch, we planned to work with 3 partners and ended up with over a hundred. Similarly, we hoped to prepare the Cloud launch with 5 partners and have already been approached by more than 40. When we actually start advertising the Cloud Partner opportunity, we expect that number to climb. Moreover, if future partner use cases continue to be as high-impact as the ones we’re seeing already, we may elect to exhaust that waitlist completely before we release Cloud publicly.

Our team has been working incredibly hard to ensure that Cloud’s initial release exceeds all expectations, and I’m extremely proud of the progress everyone has made. We expect that our first partners will begin using the product in June. Though we’ve kept quiet about some of its features, we’re excited to share a design study that captures the look and feel of the new UI:

Dashboard study


We can’t wait to get this in your hands.

Happy engineering!

Posted on Jun 10, 2019
Blog Post
Dynamic DAGs
Error Handling
DevOps & CI/CD

Love your workflows again

Orchestrate your stack to gain confidence in your data