REsolution 🧙🏻‍♂️

Chatbot that resolves the customer problems

Setup

1. Install AWS SAM CLI (tutorial)
2. Run Backend via

   docker run --env-file .env.prod resolution-api:latest

Developer Setup

1. Create environment file(s) Each environment has its own configuration file. .env.local for local, .env.prod for Production. Create each file in the root of the repo with all the necessary variables (look at settings.py).

2. Install dependencies

   pip install -r requirements.txt

3. Initialize the Vector Database with default data

   If you want to add your own data, provide the website URL via KNOWLEDGE_URL, like this:

   KNOWLEDGE_URLS=https://gosamurai.ai/,https://gosamurai.ai/payments python server/run.py vdb samurai --crawl_depth=1

   The first parameter is the type of the operation, the second is the name of the organization.

   Also, you can provide the crawling depth via --crawl_depth=N, where N denotes the number of nested levels of pages to crawl. By default, it's disabled.

4. Chat with the REsolution bot

   python server/run.py chat samurai

5. Chat with the Librarian bot

   python server/run.py librarian samurai

6. Save data to VDB from URLs

   a. Add the files to the server/upload directory

   b. Run the command

   KNOWLEDGE_URLS=https://link1.com,http://link2.com python server/run.py vdb <org_name> --crawl_depth <N>

   where KNOWLEDGE_URLS is the comma separated list of documentation links to scan, <org_name> is the name of the organization and <N> is the integer denoting the depth of links to follow. The command will add the data from the files to the VDB under the given org.

7. Save data to VDB from the file

   a. Add the files to the server/upload directory

   b. Run the command

   python server/run.py vdb <org_name> --store_files

   where <org_name> is the name of the organization. The command will add the data from the files to the VDB under the given org.

Development

• Docs
• Frontend instructions

Roadmap

0.8 alpha release

Planning.

We want to provide test environments to prospect clients. Simple chat window with the chat already configured for their company needs.

Basic components:

1. ✅ Org support
2. ✅ Chat is attached to current org It's easier to make accounts bound to the Org. Otherwise, we'll probably need to bind orgs and the databases to the URLs, which is complex and complicated. We should have boilerplate code for authentication available.
3. ✅ Save chat history in DB