The pitch sounds ambitious, maybe overly so: a desktop application that commandeers professional creative software through scripting hooks and window automation, routing generative AI models to execute editing tasks inside the same programs Hollywood studios and advertising agencies already use.
Voyager, which emerged Tuesday from a six-person team in New York, is banking that creators will prefer finishing AI-assisted work in Premiere Pro or DaVinci Resolve rather than exporting from standalone web generators. The startup describes its software as an "open harness" that wraps frontier models and pipes their output into industry-standard editors, preserving full editability.
Whether that represents a structural advantage or just adds complexity to an already crowded field remains an open question. Competitors including Runway and OpenAI have spent the past year aggressively pitching similar agent-driven workflows, though most route users through proprietary web interfaces.
Scripting APIs and fallback automation
Voyager works by calling official APIs in Photoshop, Illustrator, After Effects, and Premiere Pro, with capabilities extending to DaVinci Resolve, Final Cut Pro, Blender, Unity, and Ableton Live through official scripting bridges, according to company documentation. When those scripting interfaces don't expose the necessary commands, the software resorts to direct window control. Projects stay in their native file formats.
For users without licenses to professional tools, Voyager bundles three built-in engines. Voyager Design handles vector, raster, and motion graphics with export to PNG, SVG, PDF, PSD, Lottie, and ProRes 4444. Voyager Video provides timeline editing with multicam, color grading, and keying. Voyager DAW arranges audio with VST3 plugin support and broadcast-standard mixing.
The system routes requests to a rotating stable of image, video, and voice models that currently includes Veo 3.1, Kling 3.0, MiniMax H3, Seedance, FLUX.2, and ElevenLabs. Users can specify which model to invoke or let the agent decide based on the task.
Team and capital
CEO Anvisha Pai previously co-founded Dover and worked at Dropbox. COO Ravi Parikh co-founded Heap and Airplane. CTO John Holliman joined Dover as its first engineer after time at Expanse. The three appear on Y Combinator's roster, though the exact vintage of their batch isn't specified in available materials.
The funding picture gets murkier. Nullframe, the parent entity, announced a $7.5 million seed round on March 24, 2026, for an earlier product called Moda. An October article on explainx.ai cited founder claims of $16 million raised for Voyager from YC, General Catalyst, Pear, and angels at Dropbox, Segment, Stripe, and Scale AI, though neither the outlet nor this publication could verify the figure independently. Neither amount has been confirmed through SEC filings visible in public databases.
Credits, not subscriptions
Voyager's pricing model sidesteps the usual SaaS tiers in favor of a credit system pegged to underlying model costs. The company charges one cent per dollar that providers bill it, claiming zero markup. An eight-second MiniMax H3 clip with audio runs roughly 48 credits. Seedance 2.5 for the same duration costs around 277 credits. ElevenLabs voice synthesis runs about nine credits per minute.
The free plan includes 1,000 monthly credits limited to agent chat and still images; video, voice, and cloud rendering require paid accounts. Starter runs $19. Creator costs $49 and delivers 4,700 monthly credits at current rates, though the FAQ mentions an initial launch allocation of 7,050. Ultra runs $149 with 14,400 credits. GigaMax costs $500 with 52,000 credits and priority support. Those credit bundles have fluctuated since launch, which may reflect promotional periods or model repricing.
Users holding Claude or ChatGPT subscriptions can connect those accounts to handle conversational orchestration without double-paying. The company also allows passthrough for Higgsfield credits, letting users spend existing balances inside Voyager's interface.
Crowded timing
Runway launched its own agent product in what company materials indicate was May of this year, with an Agent 2.0 update following in June. The feature promises to carry users "from idea to finished, ready-to-publish video in a single conversation." OpenArt Director, announced June 23, targets longer vibe-driven films with consistent characters. Higgsfield added an AI Motion Designer for After Effects on October 8, the same day Voyager went live.
Adobe has folded Firefly and expanded AI Assistant capabilities across Premiere and After Effects through successive updates this year, though those features remain tightly integrated within Adobe's walled ecosystem. Voyager's positioning hinges on finishing work inside real timelines rather than exporting from web-based black boxes, a distinction that matters intensely to professional editors and perhaps less to casual creators.
Early release constraints
The macOS version for Apple silicon running macOS 15 or later is signed and notarized. Windows 11 and Linux builds exist in alpha, with screen recording, certain computer-use features, and portions of Resolve integration still unavailable as of version 0.1.87 released October 10. The FAQ promises feature parity across platforms "very soon," a timeline that remains unspecified.
The company is hiring a filmmaker in New York at $100,000 to $140,000, according to its Y Combinator profile. That salary band suggests the startup is still assembling its go-to-market motion rather than scaling an established playbook.
