Agent and infinite canvas
I worked on the Agent and canvas features, defining how users configure generation and analysis nodes, reference assets, follow execution and preview results.
ClawLu launched in China and internationally. I helped deliver Agent, infinite canvas and asset-library features, and defined account rules for individuals and teams.
I defined what each workflow needed to support, worked through feature scope with engineering and followed the changes through testing.
I worked on the Agent and canvas features, defining how users configure generation and analysis nodes, reference assets, follow execution and preview results.
I mapped which assets creators could reuse and how they could access them. I defined personal and team permissions, entry points and credit rules.
I delivered sign-in, payment and order-email specifications for the international version. I also tested the domestic canvas and personal and team features and followed up on gaps.
The decisions concern interaction structure, development scope and how the interface communicates product rules.
This diagram shows my design scope. Creation capabilities belonged to the existing product; this work focused on the interactions connecting nodes, assets and results.
Design scope · From assets to results
Retain categories, upload and previews
Match controls to generation or analysis
Cover states, expanded views and return paths
Retain current actions and assess extensions separately.
Use prompt controls for generation and in-node configuration for analysis.
Include execution, success, failure and preview states.
An illustration of node relationships using fictional assets and local simulation. It does not demonstrate product performance or release outcomes.
Enable JavaScript to try the conceptual demo.
The AI short-drama platform went live in China and internationally. The commercial module completed testing and went live.
I delivered product specifications for email sign-in, payment and order emails.
I completed feature testing for the infinite canvas and personal and team workflows.
ClawLu is an AI short-drama platform. Its Agent supports creative tasks, and the infinite canvas brings assets, generation and analysis into one workspace. My work focused on the rules and handoffs between these steps.
Design configuration, execution and previews for text, image, video and analysis nodes, alongside asset-library interactions.
Define entry points, entitlements and actions for individuals, teams and enterprises, connecting payment, credits and management flows.
I applied this workflow to canvas, account and enterprise work. AI outputs were checked against scope and existing capabilities before review and handover.
Use the product, code and historical requirements to set the objective, retained capabilities and priorities.
Create editable prototypes, link states and prepare interaction notes, rules and delivery documents.
Check added features, permissions and exception paths, revise decisions and coordinate engineering and testing.
Examples include retaining existing asset categories, keeping analysis nodes self-contained, and distinguishing account menus by actual role. Each required a concrete product choice from me.
Open to product lead, senior product manager and full-stack product opportunities.