Gideon

What is this?

• An attempt at AI-first legal tooling.
• Frontend companion (treat it as a sidebar) runs at port 1111.
• Backend API file parser + ai functions run at port 3000
• Postgres DB running at port 5432

Run it Locally

1. Create a file in the root of this repo called .env, and copy values from the gist, notion page, or whever it's being stored atm. It'll be minimal vars, the rest are fetched at runtime
2. Before we spin up services, let's install some helper packages
   1. Docker (for running all the services!): Check out docs https://docs.docker.com/get-docker/
   2. JQ (for json manipulation in bash): brew install jq
   3. AWS (for fetching secrets, deploying services): View instructions here
3. Spin up the frontend/backend from the root of this repo with docker-compose up (when api requirements change, rebuild the api with docker-compose build api)
4. Once the cluster & database are running, go to root and run migrations with the command bash manage.sh local migrate-app head
5. Check you've done your migrations correctly via CLI or a nice UI like Postico!
6. Then from the same directory, seed the database with the command bash manage.sh local seed-app
7. You're good to go! Just go to localhost:1111, click through the login/case views, and start uploading files + querying!

Production Deploys/Migrations

Want to deploy some changes for others to use?

• To deploy, we'll specify an environment and service with our deploy.sh script (ex: bash deploy.sh <TARGET_ENV> <SERVICE>)
  • bash deploy.sh production api
  • bash deploy.sh production defender
  • bash deploy.sh production frontdoor
• To migrate a database with some new revisions, we'll use our manage.sh script (ex: bash manage.sh <TARGET_ENV> <COMMAND> <DIRECTION>)
  • bash manage.sh local migrate-app head
  • bash manage.sh local migrate-app rollback
  • bash manage.sh production migrate-app head

---

---

Demo

The pretty multi-modal tool interface, letting you query against video, image, audio, documents for a case and across the organization. Also did auto-summarization of facts and important points for the case. I saved a recording I did of my hacky MVP in the docs folder.