UDP ❤️ Serverless | Proxylity UDP Gateway

What Teams Find on the Other Side

One production stack. Deployed once a week or two — not because the team was slow, but because the infrastructure made every deploy something you had to plan around. Schedule it, coordinate it, hope nothing broke.

"It's not a fair comparison. We went from deploying our one stack once every week or two, to multiple times a day over real dev, staging and production stacks."
— A.G., Senior Core Developer

Migrating off MQTT mid-project, with a live customer already in flight, and no appetite — and no budget margin — for an all-in rewrite that might not work.

"Moving away from MQTT was challenging and the risk of going all-in with a rewrite unacceptable. The Proxylity platform allowed multiple rounds of iteration and engagement that solidified the approach we needed to move forward with our customer."
— B.R., CTO, IoT Systems Integration

What Changes — Day to Day

This is not a feature list. It's what it actually feels like to work with Proxylity.

Every developer gets their own stack. No waiting on a shared environment. No "who broke staging." No merge conflicts in infrastructure config. Spin one up, use it, tear it down. The cost when idle is effectively zero.
Your branch environments stop being a compromise. Dev → staging → prod becomes a CloudFormation parameter, not a negotiation with your infrastructure team. Feature branches get real, isolated environments. PRs get tested on stacks that match production exactly.
Wire up new flows without touching old ones. Adding a new destination or swapping a Lambda handler doesn't require a deployment plan or a change window. Update a parameter, redeploy. Everything else keeps running. Your architecture finally matches how your team actually works.
Nothing to patch. No CVEs in a UDP server you own. No 11pm incident where someone asks "who owns this box." No compliance documentation for infrastructure that doesn't exist. The server layer is gone — and so is the entire category of problems it brought with it.
Your audit log is already being written. Every API call Proxylity makes into your account is automatically recorded by AWS CloudTrail. Your ops team has the access history they need. Your next security review starts with evidence, not excavation.
De-risk migrations by running old and new side by side. New and legacy endpoints can share the same listener while you validate your approach — one destination for the old path, one for the new. You don't have to bet the project on a full cutover. Move one flow at a time, on your terms.
UDP packets become native events in your architecture. Proxylity routes directly to EventBridge, SQS, Kinesis, Lambda, Step Functions, and 14 AWS services — no bridge function, no protocol translator, no adapter layer in the middle. If you already use EventBridge for your system events, your IoT devices and UDP clients join that same bus with the same routing rules, the same replay, and the same tooling. You don't adapt your architecture to UDP. UDP adapts to your architecture.
One packet fans-out to multiple destinations simultaneously. Configure a single listener to deliver to Lambda for real-time processing, Kinesis Firehose for archiving, and CloudWatch Logs for debugging — all at once, before any of your code runs. It's a configuration choice. And because you only pay for packets that reach a destination, filtered and blocked traffic — port scans, DDoS noise — costs nothing.
Going global doesn't become a project. Your listener is present in every supported region the moment you create it. The only question is where you want your packets processed. When your team gets asked "can we reduce latency for our European customers?" — you deploy your workload stack to eu-west-1, register one CloudFormation resource, and packets arriving in Europe now route to your Dublin function automatically. No regional load balancers to provision, no DNS weights to configure, no health check rules to write. And if a region goes down, traffic automatically routes to the nearest healthy region you have deployed — no runbook, no pager, no failover configuration. That's just the default behavior.

A Note from Our Founder

"The heart in the graphic above isn't ironic. UDP packets are events — discrete, timestamped, produced by something that happened in the world. Serverless is event-driven by design — your functions wake up, do one thing, and release. UDP Gateway is built around that alignment.

Event-driven architecture has always been the natural model for UDP workloads: sensors, games, telemetry, IoT, control planes — all of it is just events at high volume and low latency. When built that way, the resulting systems are simple, observable, and genuinely reliable. Not because you engineered heroically around the complexity, but because the complexity wasn't there to begin with.

Now add AI-assisted development on top of that. The leverage is enormous — but only if your architecture is coherent enough for an AI to reason about it. A tangle of servers, shared state, and hand-rolled UDP infrastructure is hard for a human to hold in their head. It's equally hard for a model. Clean event-driven systems, where each component has a single job and boundaries are explicit, are exactly the kind of codebase where AI tooling starts to feel like a superpower.

I built Proxylity because I kept watching teams spend their best engineering hours on plumbing that had nothing to do with what they were actually trying to build. When that layer disappears, something real changes — not just in your sprint velocity or your AWS bill, but in how your team feels about the work. That's what I wanted to give you. That's what the love is about."
— Lee Harding, Founder, Proxylity

The best infrastructure is the kind you stop thinking about.

What Teams Find on the Other Side

What Changes — Day to Day

A Note from Our Founder

See it for yourself