The easiest way to use AWS for agents and humans. Manage and monitor Terraform resources, builds, and deploys in your own AWS account.
- Tweets 942
- Following 37
- Followers 4,674
- Likes 5,847
Hello world. All systems are go!
🔥 ANNOUNCING RAVION 🔥
The command center for your cloud infrastructure
For startups: the easiest, most powerful way to self-host in your cloud
For scaleups: a framework for internal developer platforms
Result of 14 months. 1M LoC. Everything we learned from Flightcontrol.
1
5
16
3,976
PrivateLink can now share a network segment with a vendor, not just individual resources.
Tunnel Endpoints let you share a CIDR range through one Resource Configuration instead of creating a configuration for every resource.
ALT Vendor app sends GENEVE traffic through a tunnel endpoint and customer resource gateway into a shared CIDR. No per-resource configurations.
1
1
86
This is not a drop-in endpoint for ordinary app traffic: the vendor must use GENEVE encapsulation.
Hourly endpoint and per-GB charges apply. Keep the shared CIDR deliberately narrow.
aws.amazon.com/about-aws/wha…
56
AWS Backup no longer stops at 1,000 S3 buckets per account.
It now matches the account’s S3 bucket quota, so large fleets can protect every general-purpose bucket.
aws.amazon.com/about-aws/wha…
1
2
83
Still running small T3 instances? T8i offers up to 30% better price performance with the same CPU-credit model.
Unlimited remains the default, so sustained CPU can add charges. Benchmark before switching.
aws.amazon.com/about-aws/wha…
ALT EC2 T8i: up to 30% better price performance than T3. AWS claim; workload dependent.
1
2
196
GPT-6 Sol and GPT-6 Luna are now on Amazon Bedrock.
Sol is aimed at coding and complex tasks. Luna is built for high-volume extraction, summarization and routing.
aws.amazon.com/about-aws/wha…
ALT GPT-6 Sol and Luna are now on Bedrock
1
3
324
Claude Opus 5.5 is now available on Amazon Bedrock and Claude Platform on AWS.
aws.amazon.com/about-aws/wha…
ALT Claude Opus 5.5 is now on AWS
1
1
274
The practical win: teams with Canadian residency or latency constraints can evaluate AWS's latest Intel generation without moving regions. Benchmark gains are workload-dependent.
aws.amazon.com/about-aws/wha…
71
EC2 Fleet and Auto Scaling can target the group, prioritize reservation types, and fall back to On-Demand. Previously, groups supported only On-Demand Capacity Reservations. There is no added charge.
aws.amazon.com/about-aws/wha…
56
Releasing the hold does not make the object immediately deletable. An existing later retain-until date still wins.
No extra feature charge, but PUT requests and retained storage are billed. Test the release workflow before using compliance mode.
aws.amazon.com/about-aws/wha…
50
Use an imported certificate or an ACM-managed AWS Private CA certificate, not an ACM public certificate.
Renewals/reimports propagate without redeployment. Imported certificates still need your renewal workflow.
aws.amazon.com/about-aws/wha…
1
70
AWS-to-Azure private networking is becoming a managed object.
AWS Interconnect removes customer-managed routers, cross-connects, and BGP. Preview links are 1 Gbps and will be removed before GA.
aws.amazon.com/about-aws/wha…
2
255
Agents and MCP servers need a service catalog.
AWS Agent Registry is now GA with cross-account sharing, IaC support, audit history, and organization-wide discovery of AgentCore resources. Approval workflows are still yours to build.
aws.amazon.com/about-aws/wha…
2
170
Graviton5 is here for memory-heavy EC2 workloads.
R9g delivers up to 25% better compute performance per vCPU than R8g, with 5x the L3 cache and up to 2x the network and EBS bandwidth on the largest sizes. R9gd adds local NVMe. Benchmark before migrating.
aws.amazon.com/blogs/aws/ama…
1
181
Ravion retweeted
Managing AWS through @Ravion is a real pleasure. In preparation for our EKS deployments, I needed to add another domain name to our API; very simple operation, but touches about 3 different AWS services:
1. ALB
2. ACM
3. Route53
Usually, you need to find the terraform (or your Iac) for each of those and update tfvars (inputs) and run pipelines. Or have your agent go rogue and do all of the changes in AWS CLI, and none gets tracked.
With Ravion, single prompt:
- Add new domain to API service
The agent went its way, figured which modules needs updates, made the changes in ravion.yaml - all done in few minutes.
Ravion’s ability to allow discoverability and state management in a single plane of glass is really amazing. Agents do not need to spend tokens trying to discover your infrastructure, then your IaC, then your pipelines structure, everything is discoverable in a single yaml file.
2
3
5
472
Get multi-region AWS with Ravion’s out-of-the-box modules.
Build once. Deploy the same image across regions. Route traffic with Route 53 latency routing.
In your AWS account, via config and CLI for humans and agents.
ravion.com/docs/guides/aws/m…
ALT One Ravion pipeline builds once and deploys the same image digest to regional ALB and ECS stacks in Ohio and Ireland. Route 53 provides latency-based DNS selection using ALB target health, not request proxying. In your AWS account; data replication configured separately.
1
1
2
296