A Platform One Person Carried, Now One a Team Can Own
SavATree had acquired a business whose apps ran across five third-party hosts, with key pieces in one developer's personal accounts. We moved them onto AWS, on accounts SavATree owns.
SavATree, a tree, shrub and lawn care company, had acquired a business whose collections, irrigation and sales apps ran across five third-party hosts. They were deployed by hand, with key pieces sitting in one developer's personal accounts. After the acquisition, SavATree reviewed the platform and made de-risking it the priority. We moved the apps onto AWS, on accounts SavATree owns.
- We moved five API services, a worker, cache, queue and file storage onto AWS.
- We set up automated pipelines for development and production, with secrets out of code and no long-lived keys.
- Authentication data and proposal templates moved off personal accounts and onto company accounts.
One platform SavATree owns
- Agents handled code, migrations, audits, testing and deploys. Scope and merge decisions stayed with people.
- Every change started as a tracked story and was reviewed against its acceptance criteria.
- We rebuilt the full development environment in a new region in one day.
Human-led, AI-amplified
Migrate Everything vs. Migrate What Pays
| Migrate everything | Migrate what pays | |
|---|---|---|
| Scope | Every line of the plan, as written | Each item re-checked for what it buys the client |
| Proof | Found out at cutover | Proven in a dev environment first |
| New requests | Absorbed into the plan | Dated, approved and tracked on their own |
| Testing | Saved for the end | The client's own developer, before handoff |
The Outcome
SavATree owns its platform now. Changes get tested before production, and someone new can pick up the work.
- No key-person dependency on the original developer
- Knowledge can now transfer to another person
- Staff sign-in no longer rests on one person's account
- SavATree completed its own production cutover a month later
- Satisfied, and has since brought us in for new work
- Production errors now surface when they happen
- Every in-scope service running in development within two weeks
- Week three: testing with the original developer, plus the fixes found
Carrying a platform that depends on one person?
Challenge us. 30 minutes, no fluff.