How the pipeline works

Code counts. The model refactors. A person decides. Everything is logged.

01 · Deterministic code

Collect

Parsers read your repository, OSGi configurations, dispatcher files, package filters, Oak indexes, BPA report and content statistics. Output is a fact sheet: every finding with file, line and rule. No model is involved, so the inventory is complete and repeatable.

02 · AI on our own servers, with the facts as input

Judge

For each finding, the model decides the target pattern (Cloud Service or Edge Delivery block), the risk, and whether it can be fixed mechanically. It never invents findings: it can only reason about items in the fact sheet.

03 · Agents, one pattern per pull request

Refactor

Agents make small, single-purpose changes, add or update unit tests, and open a pull request. Adobe's own migration tools are used where they cover a pattern.

04 · A senior AEM engineer

Approve

Every pull request is reviewed and approved by a named engineer before merge. Rejections and edits are fed back so the next change is better.

05 · Deterministic checks

Verify

Cloud Manager build and code quality gates, the AEM Analyser, generated regression tests, visual comparisons of key pages and an SEO parity check of URLs, redirects and metadata.

06 · Audit log

Record

Every model call, check result and approval is appended to a hash-chained log stored in your repository. Editing or deleting any entry breaks the chain, so you can prove who approved what, and which model produced each change.

Where the AI runs

Client code is processed by AI models running on servers we operate. It is not sent to third-party AI services. If a task would benefit from an external model such as Anthropic's Claude, we ask first and use it only with your written consent, recorded in the audit log.

Running our own models also means AI time costs us almost nothing per change, which is part of why the Blueprint is free and migration prices are fixed.

What we do not automate

Authentication, security and integration design.
Product decisions, such as replacing AEM Communities or moving a site to Edge Delivery Services.
Approving anything that reaches your main branch.