Weekly signal

A compact week for agentic creative tooling: multiple vendors shipped agent-first integrations that push AI agents deeper into production creative pipelines (editing, 3D/game assets, ad buying) and into the editors creatives already use. This week’s signal: agentic features are moving from experimentation to productized workflows with attention on identity, scoped tool access, and in-editor plugins. Key developments: Kino's code-first video editor for teams + agents; NOLGIA's agent and editor/plugin updates; Layer's Meshy 3D/production playables in its agent workspace; Synter's agentic ad platform; and a production-focused MCP/identity gateway (Scalekit AgentKit).

What changed

1. Kino launched a desktop, code-first video editor that exposes the timeline and every edit as addressable code and explicitly supports coding agents (Claude Code, Codex) and MCP access so agents can edit the same timeline as humans. This reframes agent work as collaborative, auditable edits inside the project rather than separate render outputs.

2. NOLGIA shipped agent and editor improvements across its agent studio and plugins (Photoshop, DaVinci Resolve, Blender), plus maker-focused UX changes and model upgrades — making it easier to brief an agent, run studio timelines, and keep uploaded files labelled as user inputs. These are product-level moves to put agents inside editorial tools creatives already use.

3. Layer published Meshy 3D guidance: its agent workspace now targets production-ready 3D (concept → mesh → retexture → rig checks → export for Unity/Unreal/Maya) and playable-ad generation inside an agentic flow, signalling agent support for full-engine pipelines in games and ad creative.

4. Synter positioned a universal ad MCP and "Canvas" product that lets agents build, adapt and (with approval gates) launch campaigns across 30–40+ ad platforms — shifting more of ad creative + media execution into agentic automation with explicit approval and pacing controls.

5. Scalekit launched AgentKit (production MCP gateway) aimed at identity/scoped access for agents so every tool call can execute as the authorizing user and be logged — a direct response to production security and audit needs for agentic creative workflows.

What to do with it

• Trial agent-in-editor plugins on small projects first: install NOLGIA or similar plugins inside DaVinci/Photoshop/Blender and test agent edits on non-production material to learn where agents save time and where they break the shot. Label and track user uploads to avoid attribution confusion.
• Evaluate code-first timelines for reproducibility: if you manage video pipelines, pilot Kino on a short-form ad project to test versioning, agent edit traceability, and how agent edits map back to human review. Treat the code representation as part of your VCS.
• For game teams, run a Layer Meshy 3D pilot for a single asset-to-engine pipeline to validate poly budgets, rig checks and exports before full adoption. Use exported artifacts in a staged build to measure QA overhead reduction.
• If you plan to let agents act on ad accounts, require an MCP with pre-flight approvals and scoped writes (e.g., Synter) and add per-user auth + audit (Scalekit) to avoid shared-service keys and uncontrolled spend. Run a dry-run audit for one campaign.
• Security and vendor checklist: (1) ask about per-user OAuth and audit logs, (2) insist on approval-before-spend for ad agents, (3) require clear provenance for generated assets and user uploads.