Tomorrow's Valentine's Day. Our gift to you? You can now chat with your canvas. Ask your Railway environment what's running, debug failing deploys, and stage changes through a new AI agent that understands your infrastructure. We're also shipping Postgres metrics and network flows to GA after battle-testing them in Priority Boarding.

Chat with your canvas to Priority Boarding

dev.new - Talk to your Railway environments through the new canvas agent

Ask about your environment’s status and resources

Starting now, you can talk to your Railway environments through our new agent. Think of it as a 

, but it actually understands your infrastructure.

 Tell the agent what you want to change, and it'll stage the updates for you to review before applying

 Ask what's running, what's healthy, and what needs attention

 Instead of digging through logs yourself, ask the agent what went wrong and get a clear answer

 The agent checks your metrics and logs to help pinpoint problems, so you spend less time investigating and more time fixing

 Get recommendations for improving your serverless functions

 Ask questions about Railway and get answers backed by our documentation

enabling the feature flag in your account settings

, you can head to dev.new and tell it what you want to build. Already have a project? Click the chat icon in the navbar to open the agent as a side panel.

This is our first step toward making the canvas conversational. Instead of clicking around to find what you need, just ask. We're shipping this to Priority Boarding first so we can iterate on your feedback before it hits GA.

Give it a try and tell us how to make it better on 

View database metrics for your Postgres database

Your Postgres databases now have their own dedicated metrics view in the dashboard. You can see connection breakdowns (active, idle, idle in transaction), cache hit ratios, and table-level stats like row counts, data sizes, and index usage.

The best part is the Query Statistics panel. It surfaces your most expensive queries with call counts, rows returned, and timing percentiles, so you can find the slow ones before your users do. If you're running Postgres on Railway, this gives you several database observability features so you don’t immediately need to reach for external tools.

As always, you can drop your feedback on 

Network flows between services + network flow logs

Your canvas now shows live network traffic between services — also graduating from Priority Boarding this week.

Connections between services appear as animated lines, with thickness proportional to throughput. You can flip between this view and reference variables using the controls in the top left corner of your canvas.

What you're looking at is layer 4 network data, aggregated every 5 seconds. Each flow
tells you where traffic is coming from, where it's going, how fast, and whether
anything got dropped along the way. For the full picture, check the new Network Flow
Logs tab in any service's logs — it has everything: protocol, addresses, volume,
latency, and status.

We shipped telemetry in the Railway CLI to help us improve the developer experience. No project source code or environment variable values are collected. You can opt out by setting 

We shipped the ability to see the scope level for your API tokens, so you can quickly tell what each token has access to

https://res.cloudinary.com/railway/image/upload/v1771028836/2026-02-13_temxwe.png

Chat with your Canvas, DB Metrics, Network flows

Changelog #0277