Project Description
I want to roll out an on-premise AI assistant that becomes the first place my team turns whenever they need information. Its core jobs are clear: instantly search our file system, answer everyday work questions, and surface multimedia assets—especially images and videos—without leaving our secure environment.
Typical requests it must handle:
• “Find the latest Expo 2025 PR article.”
• “Show me the apple-tea campaign photos.”
• “Pull the internal training video on onboarding.”
Scope of work
1. Index and retrieve data from shared drives and cloud buckets while respecting existing permissions.
2. Support conversational querying so the assistant can answer questions in natural language, not just return file paths.
3. Recognise and preview multimedia files, with a focus on PR materials, campaign photos and internal training videos.
4. Run entirely inside our company network; no data may leave our servers.
5. Provide a lightweight desktop or browser interface that every employee can launch from their computer.
Acceptance criteria
• Queries return accurate results in under two seconds on a 1 TB test set.
• Image/video previews render inline.
• Access controls mirror our current Active Directory groups.
• Deployment passes a penetration test and code review by our infosec team.
If you have experience combining large-language models with vector search, OCR for images, and secure on-prem deployment, let’s discuss your approach and timeline.