Just went through a "seamless" migration from Jenkins to Claw. Their sales pitch made it sound like a weekend job.
We had to bring in three external consultants for two months to handle:
* Pipeline translation – their converter ignored all our shared libraries.
* State migration – build histories didn't map. The "automated" tool failed on our artifact naming scheme.
* Team onboarding – the Claw "expert" knew less about our workflows than we did.
Ended up with a parallel run for six weeks. The config drift between old and new was a nightmare.
Our final working pipeline config looked nothing like their examples:
```yaml
# Their 'simple' example
flow:
- task: build
# What we actually needed
flow:
- task: custom_sonar_check # bespoke plugin
- matrix: envs # dynamic matrix they said was 'easy'
with:
- env: prod-us
vault_path: secret/legacy/prod # non-standard path
```
Anyone else hit this? Was our experience an outlier or is their tooling just not ready for complex, real-world pipelines?
Benchmarks or bust.
Oh man, this hits home, just swap "Jenkins" for "our old CloudFormation monstrosity" and "Claw" for any "next-gen infra platform" vendor. Their sales decks always assume greenfield, cookie-cutter setups. The moment you have *shared libraries* or a *non-standard path*, their "automated" tooling falls over.
You mentioned the config drift during parallel run - that's the silent killer. We saw the same during a "lift-and-shift" that turned into a rewrite. The vendor's cost projection didn't include the 400 hours of engineer time spent reconciling state because their tool couldn't handle our S3 versioning schema. Suddenly that 30% estimated savings was a 15% cost overrun for the first year 😅
Their 'simple' example vs. reality yaml is the perfect artifact. Frame it and send it to your finance team for the next "seamless migration" budget approval.
> Their sales pitch made it sound like a weekend job.
Of course it did. That's how sales works.
Honest question: why did you expect a migration tool to handle your custom Sonar plugin and non-standard vault paths? Those are your bespoke choices, not Claw's fault. The "seamless" pitch assumes you're running a vanilla pipeline. You weren't. You had technical debt dressed up as "complex workflows."
The real cost here isn't the consultants. It's the 400 hours of engineer time you spent building and maintaining that custom shared library in the first place. Maybe simplify your pipeline before blaming the tool.
Simplicity is the ultimate sophistication