This is a demo of startupradar.co's API combined with OpenAI embeddings to find similar startups.
The code works as follows:
- load the list of startup domains in
domains.txt
- fetch descriptions of these startups with startupradar's API
- create embeddings for all startups with OpenAI
- compute cosine similarities between all pairs
- output a similarity matrix as
similarity_matrix.csv
The formatted output looks like this:
and the provided sample can be found in a public Google Sheet.
Install the dependencies into a virtual environment.
pip install -r requirements.txt
Create a config.py
file and add the credentials for startupradar and OpenAI:
STARTUPRADAR_API_KEY = "your-key-here"
OPENAI_API_KEY = "your-key-here"
OPENAI_ENGINE = "text-similarity-davinci-001"
An API key for OpenAI can be created online. Please note that embedding a lot of startups can result in significant charges. Make sure to set budgets upfront!
Run with
python cli.py run
Looking for a ready-to-use solution? Skip the setup and get instant results with our Competitor and Lookalikes API.
This demo shows the technical approach, but running it yourself means:
- Managing OpenAI API costs and rate limits
- Maintaining startup data freshness
- Handling embedding computation at scale
- Building your own similarity algorithms
Our API handles all of this for you with a simple REST endpoint.
- Instant competitor detection - Submit any company domain or identifier
- Fresh startup data - Always up-to-date company information
- Pre-computed similarities - No waiting for embeddings or calculations
- Multiple similarity types - Find direct competitors and lookalikes
- Scalable pricing - Pay only for what you use
Use cases
- Market research - Quickly map competitive landscapes
- Investment analysis - Find comparable companies for valuation
- Sales prospecting - Identify similar companies as potential customers
- Partnership discovery - Find complementary businesses
- Competitive intelligence - Stay updated on similar companies
Sign up at markets.apistemic.com and start finding similar companies in seconds, not hours.