Discovery questions to understand fit:

• "Have you worked with equity crowdfunding campaigns before, or mainly product DTC?" UNANSWERED
• "What's your experience with investor acquisition funnels versus customer acquisition? How different do you see the mechanics?" ANSWERED: He engaged immediately with a framework — look at how much they've given, who they are, what their capacity is, and likelihood to give more. He also flagged the accredited vs non-accredited distinction as the key segmentation question, showing he understands the mechanics even if he hasn't done crowdfunding specifically.

Capability questions for the Cytonics engagement:

• "If we brought you a biotech crowdfunding campaign as a client, what would your first 30 days look like?" UNANSWERED
• "What data sources would you need access to? Do you have experience integrating with platforms like Deal Maker, Wefunder, StartEngine, or Republic?" UNANSWERED
• "Your anomaly detection for CAC and ad spend — how quickly could that spin up for a campaign that's already live?" UNANSWERED

Partnership/structure questions:

• "Would you be open to a subcontracting arrangement where Deep Sight manages the client relationship and you provide the analytics layer?" UNANSWERED
• "What does a typical engagement look like — retainer, project-based, rev share?" UNANSWERED

Strategic value questions:

• "What's the biggest lever you typically find for DTC brands that are already scaling? Is it usually acquisition efficiency, retention, or something else?" ANSWERED: For this specific situation, he said "it's going to be far easier to find more people like the people they raised money from" — meaning lookalike targeting on acquisition is the lever, not upselling existing investors or retention.

Tech Stack:

1. "How are you pulling data from all these platforms — custom integrations or using something like Fivetran?"ANSWERED: Custom build. He's pulling from Shopify, Amazon, TikTok, ad channels, TV data, and scraping — all into a central data warehouse he created. Not using off-the-shelf ETL.
2. "Where does AI/LLM fit in your stack currently?"ANSWERED: More advanced than his website suggests. Agents run on top of his data warehouse. He said he moved away from specialized tools toward agents that write Python code dynamically: "The actual work is fucking Python code that it writes with everything else, and it's so much better... it's literally night and day."
3. "What happens when a client has a data source you don't already support?" UNANSWERED