Kredd

This is a complete SaaS application that allows users to create their own AI-powered inbound filter for cold pitches, requests, and opportunities.

Users describe the types of emails they receive, and Kredd generates a structured intake flow with relevant questions for senders to answer. Each submission is then scored, summarized, and ranked against the user’s criteria, helping them quickly identify the opportunities worth reviewing.

Instead of forcing high-volume inbound through a normal email inbox, Kredd gives users a dedicated front door: a unique shareable link they can place in their bio, email signature, website, or social profiles. People can still reach them, but every pitch arrives in a consistent, structured format with AI-generated relevance analysis.

Kredd is designed for anyone who receives more inbound than they can reasonably handle, including podcast hosts, literary agents, creators, investors, executives, advisors, recruiters, and other high-signal professionals. The goal is not to block opportunity, but to make good opportunities easier to find.

How it works

Step 1 - Describe what you're looking for

Tell Kredd what pitches you receive — AI generates smart filter questions automatically.

Step 2 - Share your unique link

Drop it in your LinkedIn, email signature, or website — anyone can pitch you through it.

Step 3 - Review ranked submissions

Every submission is scored, summarized, and surfaced with follow-up suggestions.

Installation Instructions

1. Install Docker and Docker Compose
2. Create a .env file in the root directory of the repository. This is used to store deployment-specific environment variables. Populate it with the environment variables from the below table.
3. Start all services with docker compose up. If you've made code changes, make sure to include the --build parameter.
4. Visit http://localhost for a local installation of Kredd.

Environment Variables

Following Twelve-Factor App principles, Kredd configuration is stored in environment variables.

Variable	Default	Description
KREDD_CERTBOT_DOMAIN	None	Domain name to issue a Let's Encrypt certificate for (e.g. kredd.io). Leave unset for HTTP-only local development.
KREDD_CERTBOT_EMAIL	None	Email address for Let's Encrypt certificate registration. When set alongside KREDD_CERTBOT_DOMAIN, HTTPS is enabled automatically.
KREDD_CLAUDE_API_KEY	None	API key for the Anthropic Claude API. If left empty, LLM-based relevance reasoning is skipped.
KREDD_CLAUDE_MODEL	claude-haiku-4-5-20251001	Claude model used to generate outreach fit reasoning.
KREDD_DB_NAME	kredd	Name of the database used by the application.
KREDD_DB_USER	kredd	Database username for authentication.
KREDD_DB_PASSWORD	add-a-long-password-here	Password for the database user.
KREDD_DJANGO_DEBUG	true	Enables or disables Django debug mode (should be false in production).
KREDD_DJANGO_SECRET_KEY	add-a-different-long-password-here	Secret key used by Django for cryptographic signing.
KREDD_GMAIL_ADDRESS	hello@kredd.io	Email address to use with Gmail's SMTP server.
KREDD_GMAIL_APP_PASSWORD	None	Application key for Gmail SMTP authentication. If left empty, no email notifications are sent.
KREDD_GMAIL_PORT	587	Port to use for SMTP authentication. For SSL, use 465. For TLS, use 587.
KREDD_GMAIL_SMTP_HOST	smtp.gmail.com	SMTP server hostname used to send outgoing email.
KREDD_MAX_ATTACHMENT_BYTES	10485760 (10MB)	Maximum size (in bytes) per outreach attachment.
KREDD_MINIO_ACCESS_KEY	kredd_minio	Access key (username) for MinIO object storage.
KREDD_MINIO_SECRET_KEY	add-a-different-long-password-here	Secret key (password) for MinIO access.
KREDD_MINIO_BUCKET	attachments	Name of the MinIO bucket used to store uploaded attachments.
KREDD_RABBITMQ_URL	amqp://guest:guest@rabbitmq:5672/	Connection URL for the RabbitMQ message broker.
KREDD_PASSWORD_RESET_TTL_SECONDS	86400 (24h)	Lifetime of a password reset token, in seconds.
KREDD_SEMANTIC_SIMILARITY_MODEL	/app/ml_models/stsb-TinyBERT-L4	Model to use for semantic similarity cross encoding. Either a path to directory within the worker container, or a name of the model. If a name is provided, the worker container will attempt to download the model as the service starts.
KREDD_SITE_URL	http://127.0.0.1	Base URL where the application is hosted.

Services

The Kredd application requires a number of different services to run. Each one has precise responsibilities. The service definitions are in docker-compose.yml.

Service	Description
db	MySQL database service that stores all persistent application data. It uses a Docker volume (db_data) for durability and includes a healthcheck to ensure readiness before dependent services start.
rabbitmq	Message broker used for handling asynchronous tasks and background job queues. Ensures reliable communication between the API and worker services.
minio	Object storage service compatible with S3 APIs, used to store file uploads and attachments. Data is persisted via the minio_data volume.
migrate	One-off service that runs Django database migrations (manage.py migrate) to set up or update the database schema before the application starts.
api	Main Django backend service that exposes the application’s API, handles requests, and coordinates with the database, message broker, and object storage.
worker	Background worker service that processes asynchronous jobs (i.e., via RabbitMQ), such as sending emails or handling long-running ML tasks.
nginx-public	Frontend web server that serves static assets and acts as a reverse proxy to the API service. In production (when KREDD_CERTBOT_DOMAIN is set), automatically obtains a Let's Encrypt certificate and serves traffic over HTTPS on port 443, redirecting HTTP.
nginx-spa	Nginx, only visible on the internal network, that serves the single page React application. This is seperate to the public facing nginx instance, so that we can enable maintenance mode, dedeploy the SPA, and users don't hit internal server errors during redeployment.

Development - Git Hooks

First install Ruff:

sudo snap install ruff

Then, configure git to use the .githooks directory. This will trigger the pre-commit formatting hook to run on git commit.

git config core.hooksPath .githooks

Development - Making Migrations

The simplest way to do this, is to use a Python environment outside of Docker.

1. Create and activate a virtual environment:

$ python3 -m venv env
$ . env/bin/activate

2. Install the Python requirements into your virtual environment:

$ pip install -r requirements

3. Generate the migrations

python manage.py makemigrations