PALANTIR
·
Workshop
Design a Safe and Transparent Rebase Workflow for a No-Code Environment
Year
2025
Role
Sole product designer
Duration
2–3 weeks for v0, continued iteration afterward
Context
Workshop is the Palantir's no-code app builder — the primary way teams turn their data and ontology into operational tools.
As Palantir prepared to introduce platform-wide branching, we needed to support a foundational Git concept: rebasing.
In Git, rebasing is already intimidating. Translating it into a no-code, component-based environment—where users do not have access to code, and most diffs operate at an abstraction level—was an entirely different challenge.
My role was to design the rebasing workflow for Workshop:
A system that allows users to safely rebase their apps against the latest main, understand merge conflicts, and reconcile changes without losing trust or visibility.
Design Challenge
Workshop is visual. Git is textual.
Rebasing sits at the intersection.
Git’s rebase model assumes:
Line-by-line diffs
Granular, property-level visibility
A textual representation of every change
A user who understands how merge conflicts propagate through a file
Workshop, however:
Only exposes changes on a component level (widgets, sections, variables, pages…)
Requires spatial context to understand what has changed
Sections and widgets can be nested several layers deep.
Due to technical constraints, can’t show parallel visual states to compare main vs branch
This mismatch makes rebasing inherently fragile. To unlock branching, we needed a system that transforms a deeply technical Git workflow into something visual builders could trust.
Design Goal for V0
Clear change origins (Main vs Branch), visualized outcomes, and safe modification
Rebasing in Workshop spans ontology resources, application logic, and deeply nested visual configurations—making real production modules far more complex than Git’s linear, text-diff world.
For the first version, our goal was not to solve every edge case or introduce a perfect rebase engine. Instead, we intentionally scope a lean V0 to be just enough to observe real user behavior, validate the mental model, and de-risk the foundation for V1.
This V0 allowed us to validate core interaction patterns before scaling into the full complexity required for V1 and beyond.
Design Exploration
Different mental-model translations of Git-like rebasing
Design Goal for V0