All work

Case study 01 — AI agent / Content operations

Soontra

The AI agent for content teams.

What we did
Product design, full-stack development, AI agent, semantic video search, social publishing
Status
Live product
Soontra library search for “walking down hallway with backpack” returning two matching clips, with a video preview and an AI-written summary and timestamped scene description
LIVE PRODUCT · RECORDED IN THE APP, AUGUST 2026

The brief

From scattered tools
to one agent that does the work.

Content teams juggle a drive, a docs tool, a chatbot, a review tool, and a scheduler. We designed and built Soontra to replace that stack with one workspace and an AI agent that learns the brand, remembers the work, and executes instead of suggesting.

Every upload is indexed so teams can search footage by what they remember. The agent drafts scripts for approval, answers questions on the exact frame, publishes to five social platforms, and reads the analytics to plan the next post. An MCP server lets Claude, ChatGPT, and Cursor work in the same workspace.

  • Product design
  • Full-stack development
  • AI agent
  • Semantic video search
  • Social publishing

Ask for the clip

Search footage
by what’s in it.

Every upload is transcribed, tagged, and indexed. Type what you remember and jump to the exact moment.

5
Social platforms it publishes to
4
File types indexed: video, audio, images, docs
MCP
Works inside Claude, ChatGPT, and Cursor

Building something similar?

Start a project
NEXT CASE — 02 GROW Discover