Model ML

Launched "Table mode" for better table reading accuracy.

Model can now read information from pasted URLs.

Crunchbase Partnership - For quick finding of similar businesses/competition.

1. Improved transcript querying
2. Launched "Table mode" for better table reading accuracy.
3. Model can now read information from pasted URLs.
4. Crunchbase Partnership - For quick finding of similar businesses/competition.

Transcript Summarization and Improvements

We've significantly improved call summarization. Model ML can also now better distinguish who’s on the call. Enabling for queries like, “What did they ask me?” or “give me a list of questions I asked on the call along with their answers in a table”.

You can now search for competitors based on our Crunchbase partnership in two different ways:

If the company already has a profile on Crunchbase, we will use this description to find similar companies. Alternatively, you can provide a description of the company as Crunchbase descriptions can be slightly outdated.

Give me the competitors for the following company from Crunchbase: {Company Name} {Optional filter: Raised X in funding} {Optional filter: HQ in X location}

Give me a list of companies that match this description from Crunchbase: {2 sentence description} {Optional filter: Raised X in funding} {Optional filter: HQ in X location}

1. Search by Company Name:
 Give me the competitors for the following company from Crunchbase: {Company Name} {Optional filter: Raised X in funding} {Optional filter: HQ in X location}
2. Search by Description:
 Give me a list of companies that match this description from Crunchbase: {2 sentence description} {Optional filter: Raised X in funding} {Optional filter: HQ in X location}

Remember to mention Crunchbase as the source in your prompt.

Will send back first 10 results along with FULL csv.

1. Remember to mention Crunchbase as the source in your prompt.
2. Will send back first 10 results along with FULL csv.

Check out the video below and read the linked article for more information.

We've also made significant improvements to the model, including longer outputs, ability to read from URLs you give it and better summarizations. Our focus remains on accuracy, so please report any issues.