They're pushing us to migrate from our current stack to Braintrust. "Modernize the workflow," they say. So I spent an afternoon actually reading the docs and made this.
It's not a sales slide. It's a translation from marketing speak to what it means for our actual, boring, needs-to-work-on-Monday systems.
| Thing We Actually Use | Braintrust Calls It | What It Actually Means |
| :--- | :--- | :--- |
| Jenkins pipeline in a repo | "Git-native pipelines" | YAML in your repo, but now vendor-locked. Can't run it locally without their CLI. |
| kubectl apply -f k8s/ | "Kubernetes-native execution" | It generates a Job/Pod spec and runs `kubectl` for you. You still debug YAML. |
| Ansible playbook runs | "Declarative automation" | Wrap your Ansible call in their YAML. Adds a layer. |
| Grafana + Prometheus alerts | "Built-in observability" | Another dashboard to check. Alerts likely less granular than what we've tuned for years. |
| Bash scripts in cron Jobs | "Scheduled workflows" | Cron, but with their scheduler and more points of failure. |
The value prop seems to be "one pane of glass." But we already have glass. It's called a terminal and a browser with tabs. This adds another abstraction layer and another vendor.
To run a simple shell command, you now need their config:
```yaml
tasks:
- name: "Why?"
bash: |
echo "Migrating this would take a week"
echo "For what benefit?"
```
We'd be trading simple, debuggable, proven tools for a "platform" that does the same thing with more complexity. I'm not seeing the win. Convince me I'm wrong.
If it ain't broke, don't 'upgrade' it.