Doza Assist

Product Introduction

1. Definition: Doza Assist is a local-first, open-source AI narrative intelligence application and editorial assistant designed for documentary filmmakers and spoken-word editors. It functions as a specialized AI layer for non-linear editors (NLEs) that automates the labor-intensive process of transcribing, logging, and building rough cuts from unscripted footage. Technically, it is a macOS-native application that leverages local Large Language Models (LLMs) and speech-to-text engines to analyze narrative structure without requiring cloud connectivity.

2. Core Value Proposition: Doza Assist exists to solve the "blank timeline" problem in documentary and interview-based post-production. Its primary value lies in its ability to convert hours of raw interview footage into a structured narrative sequence in minutes, using "Narrative Intelligence" rather than simple keyword matching. By prioritizing privacy through local processing and offering a personalized AI that learns a specific editor's style, it provides a high-efficiency workflow for creators who handle sensitive or high-volume spoken-word content.

Main Features

1. Local Narrative Intelligence & AI Story Chat: This feature utilizes a local AI engine to act as a collaborative partner. Unlike standard search tools that look for specific words, the Story Chat understands thematic concepts, emotional subtext, and narrative arcs. Users can query the footage with complex questions such as "What is the emotional climax of this interview?" or "Find moments of internal conflict." The system generates responses linked directly to playable, time-coded clips, allowing for immediate verification and placement in the timeline.

2. Automated Story Builder & Sequence Generator: The Story Builder acts as an automated "paper edit" tool. By providing a prompt—such as "create a 60-second social cut"—the AI analyzes the transcript to identify hooks, mission statements, and narrative resolutions. It proposes a chronological sequence of clips, complete with editorial notes explaining the selection. These sequences can be rearranged via a drag-and-drop interface and then rebuilt or refined until the structure is finalized.

3. Style Calibration (Machine Learning Integration): This is a sophisticated personalization feature that allows the AI to move beyond generic editorial suggestions. Users can import their previous finished projects (XML/EDL files), which the AI analyzes to learn specific editorial habits: how the editor builds tension, their preferred pacing, and which types of soundbites they typically prioritize. This "Style Profile" then calibrates future Story Builder suggestions to mirror the editor’s unique creative voice.

4. Multi-NLE Workflow Integration: Doza Assist is built for professional interoperability. Once a story sequence is finalized within the app, it can be exported as a native timeline file for Final Cut Pro, Adobe Premiere Pro, and DaVinci Resolve. This eliminates the need for manual clip matching and allows editors to move immediately into the fine-cut stage within their preferred professional environment.

Problems Solved

1. The Time-Intensive Sifting Process: Documentary editors often spend days or weeks "mining for gold" in hundreds of hours of interviews. Doza Assist reduces this discovery phase from days to minutes by surfacing the most impactful story beats automatically.

2. Cloud Security and Privacy Risks: Many AI transcription and editing tools require footage to be uploaded to external servers, which is a dealbreaker for investigative documentaries, legal depositions, or corporate projects under NDA. Doza Assist runs 100% on the user's Mac (Apple Silicon or Intel), ensuring that no data ever leaves the machine.

3. Generic AI Output: Standard AI tools often produce "flattened" edits that lack character. Doza Assist solves this by allowing the AI to be trained on the user's own body of work, ensuring the output feels like a human-led rough cut rather than a machine-generated summary.

4. Target Audience:

• Documentary Filmmakers: Who need to find narrative threads in massive amounts of footage.
• Podcast Producers: Looking to turn long-form interviews into structured episodes or social clips.
• Corporate & Branded Content Editors: Tasked with creating testimonial-driven stories under tight deadlines.
• Legal Videographers: Processing depositions for specific thematic evidence.
• News & Investigative Journalists: Who require fast, private transcription and story assembly.

Unique Advantages

1. Local-First Architecture: By running entirely on-device, Doza Assist bypasses subscription costs associated with cloud processing and offers the highest possible level of data security. This also means it works offline, making it ideal for editors working in the field or in secure environments.

2. Personalized Narrative Training: Most AI editing tools treat every user the same. Doza Assist’s ability to ingest and learn from an editor's portfolio (Style Calibration) creates a feedback loop where the tool becomes more useful and "familiar" with every project it processes.

3. Open Source & MIT Licensed: Unlike proprietary SaaS models, Doza Assist is free and open-source. This transparency allows the community to audit the code, contribute to its development, and ensures that the tool is not subject to "feature sunsetting" or sudden price hikes.

Frequently Asked Questions (FAQ)

1. Does Doza Assist require a monthly subscription? No. Doza Assist is free and open-source software under the MIT license. There are no recurring fees, no per-minute transcription costs, and no account requirements to use the application.

2. Which NLEs are compatible with Doza Assist exports? Doza Assist supports direct exports for the industry's leading non-linear editors, including Final Cut Pro, Adobe Premiere Pro, and DaVinci Resolve. It generates the necessary XML and metadata files to rebuild the sequence instantly in your chosen timeline.

3. How does the "Local AI" handle heavy video files? The AI primarily analyzes the audio/transcript data for story structure, which keeps the processing overhead manageable. Because it is optimized for macOS (Apple Silicon and Intel), it utilizes the Mac's local Neural Engine and GPU to handle transcription and narrative reasoning without the lag associated with cloud-based tools.

4. Can I use Doza Assist without an internet connection? Yes. Once the application is installed, all transcription, analysis, and story building happen locally on your hardware. No internet connection is required for the tool to function, making it a secure choice for air-gapped workstations.