Filament is a data replication engine written in Go. It supports full loads, incremental replication, and change data capture (CDC). As data moves through the engine, Filament verifies Arrow batches at the sink handoff and, where the sink supports it, checks the exact serialized payload. You can run Filament as a standalone binary or embed it in a Go application.
Checksumming the Arrow batch and recomputing it at the sink is a good answer to the quiet failure in pipelines, where rows arrive but not quite as they left. Advancing checkpoints only after the destination commits makes replays safer too. How do you handle a schema change mid CDC stream, like a column type change on the source, without breaking verification or forcing a full reload?
Whoa this is cool! Congrats! 👏 One question: how does Filament handle schema changes mid-replication, like a column added or a type changed in the source? Or is the source locked-down so those changes can't happen?
Congrats to the whole team - @mitchbregs It's been great using the tool. Setting it up, configuring, testing, validating- great tooling.
Moving data is the easy part — trusting it arrived intact is the hard part, especially when an agent kicked the job off. I like that Filament verifies Arrow batches at the sink handoff, and that you can run it as a standalone binary or embed it in a Go app. Curious how it handles a long-running CDC stream if the sink drops briefly — does it resume cleanly from where it left off?
Congrats on the launch!
I know today is only day 1, but what is your medium to long term vision for filament? (1-3yr)
Hey @fmerian, congrats on the launch! Filament looks really interesting. I do have a bunch of data to migrate and this looks like the right tool for the job.
cdc plus verified arrow batches at the sink, love the focus on data integrity. congrats!
We let agents query UK B2B data through an MCP server, and the hardest part was trust: proving what the agent saw matched the source. Verifying batches at the sink handoff is the right instinct. Does the agent get a readable proof back, or just pass/fail?
this idea is so sweet! How do you verify the data arrived correctly?
Congratulations on the launch!
Appreciate the distinction between checking an Arrow batch at the sink handoff and checking the serialized payload where the sink supports it. Showing that boundary in the run result would help someone understand what was actually verified.
I understand the concept, but I want o learn more: Who uses this? Whats the primary uses cases for this?
Coming from PHP world, for a moment I was like "wow, that's a drastic pivot for Laravel Filament" :-) Congrats on the launch folks!
what happens to the route's slot when the primary fails over, does it get created with failover=true on pg17+
Congratulations @Filament by Galaxy team on their launch. It's a nice OSS project, and I'm following their growth story.
Hey everyone! I’m Saloni, i'm a student and I work on Filament’s open source side: building connectors and helping community contributors land their first PRs.
What’s surprised me most is how quickly people can contribute. Most SaaS connectors are just a YAML manifest plus a docs page. I shipped the Asana source this week and community contributors have already picked up Airtable and Jira within days of us opening them up!!
Is there a tool you’d want Filament to connect to or an idea for what we should build next? Drop it in the comments here or open an issue on GitHub!
If you try Filament on your own data, we’d also love to hear how it goes!!
Been lucky to play a small part in this journey with Mitch and Leon. Watching Filament go from repo to launch has been a buzz. Massive day lads, let’s goooo 🚀
Hi all - I'm Leon, one of the co-founders of @Filament by Galaxy. We are super excited about what filament can bring to your data stack, we've seen it reduce latency by up to 15x against some other tools in production.
After working in the data space for nearly the last decade we've seen countless tools (and pricing models... ha) for moving your own enterprise data into warehouses/lakes. We wanted to bring a fresh take on how to do this in the AI era with speed and trust (filament also converted me to a gopher, ✌️ python lol).
Major shoutout to @atterpac on the brunt of the ideation, design, and implementation of filament. If you struggle with cost, scalability, or disaster recovery in your current data replication tool we would love to help you test this out!
As always we would love feedback, stars, and for you to genuinely battle test filament on your own workloads and we're here to help!
About Filament by Galaxy on Product Hunt
“Move data using your terminal or agent. Fast & open source.”
Filament by Galaxy launched on Product Hunt on November 13th, 2026 and earned 295 upvotes and 66 comments, earning #2 Product of the Day. Filament is a data replication engine written in Go. It supports full loads, incremental replication, and change data capture (CDC). As data moves through the engine, Filament verifies Arrow batches at the sink handoff and, where the sink supports it, checks the exact serialized payload. You can run Filament as a standalone binary or embed it in a Go application.
Filament by Galaxy was featured in Open Source (69.1k followers), Developer Tools (523.1k followers) and Data (2.5k followers) on Product Hunt. Together, these topics include over 113.7k products, making this a competitive space to launch in.
Who hunted Filament by Galaxy?
Filament by Galaxy was hunted by fmerian. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how Filament by Galaxy stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
fmerian
Checksumming the Arrow batch and recomputing it at the sink is a good answer to the quiet failure in pipelines, where rows arrive but not quite as they left. Advancing checkpoints only after the destination commits makes replays safer too. How do you handle a schema change mid CDC stream, like a column type change on the source, without breaking verification or forcing a full reload?