🔬 Product & Research — go live in one command

This stack is a product-research department:

                         ┌──────────────── Caddy (auto-HTTPS) ────────────────┐
   your team ──────────▶ │  https://research.yourdomain.com   (GPT Researcher)│
                         └───────────────────────┬────────────────────────────┘
                                                 ▼
                                         GPT Researcher
                                         (cited briefs)
                                                 │
                                         web search + LLM

You get an autonomous research agent that turns a question into a sourced brief — behind real HTTPS on your domain.

Replaces: a junior market/desk researcher + a "Deep Research" subscription
Time to live: ~30–60 minutes
You need: a server with a public IP, a domain, Docker, a search-provider key, and an OpenAI-compatible LLM endpoint

Step 1 — Get a server

Any Ubuntu 22.04+ VPS with 2 vCPU / 4 GB RAM (minimum) works — Hetzner, DigitalOcean, Vultr, AWS Lightsail, etc. SSH in, then install Docker:

curl -fsSL https://get.docker.com | sh

Open ports 80 and 443 in your provider's firewall.

Step 2 — Point your domain at it

Create a DNS A record pointing at your server's IP:

research.yourdomain.com → GPT Researcher

Verify it resolves before continuing (HTTPS issuance depends on it):

dig +short research.yourdomain.com   # should print your server IP

Step 3 — Get the stack

git clone https://github.com/arunrajiah/ai-workforce.git
cd ai-workforce/departments/product/deploy
cp .env.example .env

Step 4 — Fill in `.env`

In .env, set at minimum:

DOMAIN → research.yourdomain.com and ACME_EMAIL → your email (for Let's Encrypt)
TAVILY_API_KEY → a search-provider key from tavily.com (GPT Researcher's default retriever)
LLM_BASE_URL and LLM_API_KEY → your OpenAI-compatible endpoint and key (see below)

Pointing at your LLM (provider-neutral)

GPT Researcher speaks the OpenAI API, so any OpenAI-compatible server works — a self-hosted model server, Ollama, or an OpenAI-compatible gateway. Set:

LLM_BASE_URL → the endpoint's base URL (usually ends in /v1)
LLM_API_KEY → the key that endpoint expects (some local servers accept any value)
FAST_LLM / SMART_LLM / STRATEGIC_LLM / EMBEDDING → models in provider:model form. Keep the openai: prefix for any OpenAI-compatible server (e.g. openai:gpt-4o-mini), or use ollama:<model> when running against Ollama.

Step 5 — Launch

docker compose up -d

Caddy will fetch a TLS certificate automatically. In ~1 minute, open https://research.yourdomain.com and ask your first research question.

Watch it come up: docker compose logs -f gpt-researcher caddy

Step 6 — Run a research brief

In GPT Researcher, enter a question (e.g. "competitive landscape for open-source survey tools in 2026"), pick the report type, and it will search the web, read sources, and return a cited report you can export.

You're live. 🎉

Operate it

Task	Command
View logs	`docker compose logs -f gpt-researcher caddy`
Update everything	`docker compose pull && docker compose up -d`
Restart the researcher	`docker compose up -d gpt-researcher`
Stop everything	`docker compose down`

Troubleshooting

No HTTPS certificate — DNS isn't pointing at the server yet, or ports 80/443 are blocked. Check docker compose logs caddy. The research. record must resolve.
Research runs fail / no answer — check TAVILY_API_KEY is set and that LLM_BASE_URL/LLM_API_KEY point at a reachable OpenAI-compatible endpoint. Inspect docker compose logs gpt-researcher.

Swap a component

Prefer a different research agent or an internal-knowledge base? Every piece is replaceable — see the Product & Research directory for alternatives (Open Deep Research, deep-research, Onyx).