After months of working with the platform, I noticed some of the most powerful features tend to get buried in changelogs. Here’s what’s actually changed my day-to-day workflow, hope it helps anyone onboarding or migrating pipelines.
PRATICAL TIPS
Global Branching — iterate without breaking prod
Global Branching is now available on all enrollments. You can make changes across Transforms, TypeScript functions, Ontology, Workshop modules, and AIP Logic on a single branch, test everything end-to-end, and only then merge into Main. No more accidental production breaks.
Worth exploring: Ontology Manager → Branching and the proposal approval options before merging.
Variable Lineage Graph in Workshop
If you’ve ever spent too long figuring out why a variable isn’t recomputing, the Variable Lineage Graph is your answer. It replaces the old dependency graph with a much clearer visualization of which widgets read or write each variable.
How to open it: Workshop → Edit mode → Variables panel → Graph icon (top right)
Ontology MCP — expose your ontology to external agents
Ontology MCP turns your Developer Console app into an MCP server. External agents, LangChain, CrewAI, Copilot Studio, custom Python/TypeScript, can discover and call your object types, actions, and queries via a standard protocol.
You can now manage all MCP servers across your enrollment from MCP Hub, including enabling/disabling individual tools and customizing descriptions without touching application code.
AIP Document Intelligence for structured document extraction
If your team processes PDFs or image-heavy documents, AIP Document Intelligence is worth a look. You can experiment with different extraction strategies (traditional OCR through to vision LLMs), compare quality, speed, and cost side by side, then auto-generate Python transforms to run at scale.
Enabled by default on AIP enrollments. Find it via AIP Document Intelligence in Compass.
Search logs in Workflow Lineage for production debugging
A small feature with a big payoff: the Search logs tab in Workflow Lineage aggregates logs from all executions of a function, action, or automation over the last 7 days. Perfect for tracking down intermittent errors without clicking through individual runs.
Migration heads-up: if you’re still editing Python transforms in Code Repositories, note that starting June 2026 that editor moves to legacy status. VS Code Workspaces is now the recommended environment, it brings AI-assisted coding, integrated dataset preview, and a built-in terminal. Worth switching before you get stuck on a deprecated path.
Which of these are you already running in production? Would love to hear use cases around Ontology MCP or Global Branching.
Felipe Montes
Development Advisor
Follow me @ linkedin.com/in/felipemontesl/ or GitHub: Brazillianerd