Backboard

• 

  The core dashboard

Inspiration

With teammembers fundraising and many conversations with fund managers, VCs and founders we know that 99% of deals happen through warm intro.

Ultimately the email inbox is the graveyard of every introduction and cold inbound. There has to be a way to filter out the noise and have a high level understanding.

What it does

Backboard is like an AI dataroom that requires no active maintenance or updates - it monitors your email inbox, infers which emails are worthy of ingesting and then summarises these deals.

Backboard AI let's you then ask questions about all deals that have come across your desk in the last day/week/month/year and supports asking questions relative to other private companies and data both on your Backboard and in Crunchbase/Pitchbook.

How we built it

We use Gmail API and (if needed) n8n to hook into email and pass this to Command R on Bedrock to determine if the email contents warrants saving the attachment.

We then store the deck/documents in Supabase Storage and at regular intervals use AWS Bedrock and multimodal models to infer our structured output to make all information visually consistent.

Users (VCs) can then use our chat to speak with Command R and if needed use MultiOn to crawl Crunchbase for related companies.

Challenges we ran into

• A team member left
• Problems with identifying strengths for teammates
• Using Gmail API is cumbersome so we scrapped this part of the project

Accomplishments that we're proud of

• Our design, brand and story are visually appealing and strong
• We delivered a solution with real-world utility

What we learned

• Communicate more frequently
• Divide tasks and actions more accurately based on team strengths

What's next for Backboard

• Finish the project
• Ensure email ingress is functional, secure and compliant (a hurdle like Superhuman have)
• Have VCs test the project

Built With

• amazon-web-services
• anthropic
• bedrock
• cohere
• multion
• supabase
• svelte
• vercel