Bill Palombi

Embedded accounting tools for SaaS & Fintech product builders at Tight | Formerly @enigma_data, @prefectIO, & @fiscalnote

[Bill Palombi @BillPalombi]

@enigma_data 👊🇺🇸🔥

[Bill Palombi @BillPalombi]

Also, we're hiring: https://t.co/dBx4LVy6fr

[Bill Palombi @BillPalombi]

Me trying to work on on graph-model-1 launch

[enigma_data @enigma_data]

Come refine with us. https://t.co/bVtzjNnop4

[Bill Palombi @BillPalombi]

Its never been easier to go from an idea to prototype, but never been harder to know what information you can trust. Enigma's entity resolution platform & data model yield the most comprehensive, accurate, expressive data on the identity & activity of US businesses. Check it out.

[enigma_data @enigma_data]

Meet Enigma graph-model-1 (Research Preview)

In late 2023, we began transforming Enigma’s data and machine learning infrastructure to capture the complexity of the U.S. business landscape in far greater detail. Today, we’re proud to introduce the new version of Enigma, graph-model-1, an evolved entity resolution framework and data model powering Enigma that maps brands, legal entities, operating locations, and the people who power them—all woven together in a 2.4B-node knowledge graph.
Here, a “business” isn’t just a single branch or registration record; it’s the sum of everything we know: from its brand identity and legal frameworks to the physical footprints where it operates, the individuals who hold key roles or ownership stakes, and so much more.

graph-model-1 is the most comprehensive, accurate, and expressive business data set:

* Enhanced entity linking & accuracy: graph-model-1 drastically reduces inaccurate entity resolution (false positives and negatives)—even in complex entity relationship scenarios where traditional systems often fail.
* Expanded data coverage: now covering 49.8M brands and operating locations, 98M registered entities, and linking $7.1T in annual card revenues, we capture newly formed or fast-evolving businesses and operational changes in near real-time.
* Brand entity embeddings for semantic filtering: identify and group businesses based on nuanced brand relationships, enabling more intuitive queries and segmentation across franchise networks or complex corporate structures.

what you can do with graph-model-1:

* Target franchise networks: Pinpoint franchisor–franchisee relationships by focusing on a parent brand, then discover affiliated locations or legal entities.
* Discover emerging businesses: surface newly established companies using minimal input—such as a partial brand name or approximate location—and follow their trajectory as they evolve.
* Monitor operational footprints: combine location intelligence with card revenue data to detect expansions, closures, or operational shifts—ideal for market analysis and financial forecasting.
* Segment by behavioral insights: filter entities based on semantic understanding of their online presence, transaction patterns, customer feedback, or other key indicators—without relying on rigid, predefined search parameters.

We’re currently onboarding close partners and technical collaborators to validate graph-model-1’s capabilities. Over the coming weeks, we’ll share more about the additional features we’re rolling out—and demonstrate how we can unlock new levels of insight for the hungry algorithms that drive decisions off of Enigma.

[Bill Palombi @BillPalombi]

@jlowin

[Bill Palombi @BillPalombi]

But hey, I've got my own bias. Give my post about orchestrating python scripts a read and judge for yourself (https://t.co/ngI41gfsOM)!

[Bill Palombi @BillPalombi]

I love stuff like this! I wrote a similar post a couple of years ago (https://t.co/wHngZPaUc2). While we differ on some on the facts, Simon and I are broadly on the same page about how orchestration evolved. The most exciting part is that its still evolving!

[Simon Späti 🏔️ @sspaeti]

Excited to share a new chapter in my Data Engineering Design Pattern book on orchestration's evolution.

https://t.co/N9t6rPAqjF

[Bill Palombi @BillPalombi]

If I were a curious follower of the orchestration landscape, especially someone interested in native python orchestration, the decision to omit Prefect coupled with the knowledge of Dagster's sponsorship would make me wonder about potential bias here.

[Bill Palombi @BillPalombi]

Interactive workflows can wait for and receive input from humans, other workflows, and more!

[Bill Palombi @BillPalombi]

As with all Prefect features, you can get started quickly with the "it just works" version, exposing a form in the Prefect UI based on just a schema defined in Python, or you can compose the underlying methods to enable all sorts of interactivity.

[Prefect @PrefectIO]

Interacting with workflows at runtime can be hard to get right. To DIY you need to...

🎪 host an API
🤡 build a UI
🤹 run or rent a database

Interactive workflows are now natively supported in Prefect 💥 Check out this human-in-the-loop tutorial:

https://t.co/nmztmHw5Yh

[Bill Palombi @BillPalombi]

@sspaeti While it’s not as comprehensive as what you’re making, my brief history of workflow orchestration may have some useful anecdotes: https://t.co/6LxyLSgUJ7

[Chris @criccomini]

This is *awesome*. I built something @linkedin many moons ago called Reportal. This was its most used feature: user input for DAG execution. 👇

Opened it up to a ton of basic report generation use cases.

[Bill Palombi @BillPalombi]

@AAAzzam The Witness is also excellent

[Prefect @PrefectIO]

Hear about how Doug used Prefect to solve his team's 5 biggest problems:

- Scheduling remotely
- Error handling
- Monitoring
- Parallelization
- Result persistence

Listen to his story about going from unruly scripts to remotely running workflows 👉
https://t.co/tyUxkIN29p

[Bill Palombi @BillPalombi]

@sarahmk125 Totally agreed

[Bill Palombi @BillPalombi]

Don’t coerce your code into linear pipes. Elevate your code into dynamic workflows that react to their environment and the data they handle when they run.

[Bill Palombi @BillPalombi]

One of my favorite books, Surfaces and Essences, describes analogies as "constituting thought’s core." This post challenges the dominant analogy in our minds: workflows as pipelines.

[Prefect @PrefectIO]

Your data isn’t water, it’s mail.

Data intensive processes are inherently non-linear. Data is a reflection of the complex, asynchronous world around us.

How workflows should adjust to schema changes, manual intervention, or conditionals on the fly:
https://t.co/YGabu35iNa

[Bill Palombi @BillPalombi]

Legacy orchestrators, and the pipeline mental model embedded in them, have made us think in static terms. Prefect supports a new, dynamic way of thinking about workflows.

[Bill Palombi @BillPalombi]

Data folks have many ways of referring to the things we build - “jobs”, “pipelines”, “DAGs”, even “streams”. No matter what you call them, Prefect elevates Python scripts into dynamic, reactive, event-driven workflows. Check it out!

[Prefect @PrefectIO]

🚨 New website, new brand, new event-driven workflows. Take it for a spin: https://t.co/kpwLdhIvEM

At Prefect, we make complex workflows simpler, not harder. How, you might ask?🧵