What Sourcegraph shipped
Sourcegraph shipped Agentic Batch Changes, a major update to their code search platform that automates vulnerability remediation across massive codebases. By deploying frontier agents to handle cross-repository updates, the tool shifts platform engineering from manual tracking to automated PR generation.
The challenge with any code-at-scale product is making invisible, distributed work feel tangible. This film has to prove that an AI agent can navigate multiple languages, write regression tests, and submit reviews without overwhelming the viewer with terminal windows.
How the motion works
Devtool launches face a common hurdle: making complex, multi-repository workflows visually legible. Sourcegraph tackles this by avoiding sprawling, unguided screen recordings. Instead, the edit relies on a series of rapid punch-in maneuvers, driving the viewer's eye directly to the relevant lines of code and pull request statuses. The camera doesn't drift aimlessly; it snaps to precise UI coordinates.
To visualize the AI agent's autonomous work, the motion design employs scramble-text effects during code generation sequences. This avoids the tired trope of a blinking cursor typing out characters one by one. The code blocks resolve instantly after a brief, cryptographic scramble, communicating machine speed and scale rather than simulated human effort.
The rhythm of the piece mirrors the product's core value proposition: eliminating friction. Transitions are executed via hard-cut editing, jumping from Python to Go to Java environments without transitional fluff. Combined with stark kinetic-typography to introduce key phases of the workflow—find, write, test, review—the film maintains a clinical, authoritative momentum from start to finish.
What to steal from it
- Use hard cuts to transition between different UI states to convey speed and system efficiency.
- Replace slow typing animations with text scrambling to visualize machine-speed code generation.
- Punch in aggressively on specific UI elements to prevent viewer fatigue when showing complex developer dashboards.
- Anchor abstract AI concepts in concrete, recognizable developer environments like pull request screens.
Want a video like this for your product?
Impractical cuts one from a single prompt — the same motion craft, in about twenty minutes.
