About Cagdas

Cagdas Ucar is VP of Technology at FundCanna, the leading provider of financial products for the US cannabis industry, which has funded roughly 4,500 applications to date. A software engineer of 20-plus years, he oversees the technology behind FundCanna's revenue-based lending products, serving a market of only around 50,000 eligible businesses nationwide — where banks won't lend, Google won't run ads, and every customer counts.

What we talked about

A market too small for "big data" thinking — Why FundCanna tracks individual customers in real time instead of aggregate funnels, in an industry with only ~150–200 fundings a month.

Financing that isn't a loan — How FundCanna's revenue-based "pay advance" model works, and what happens to repayment when a customer's revenue collapses.

Growing without Google Ads — Why word of mouth, dispensary marketplaces like Apex Trading and LeafLink, and a partner network of 200+ carry almost all of FundCanna's growth.

When Plaid breaks down — The banking-data verification problem that pushed FundCanna to build its own bank-statement analysis pipeline.

Schedule III and what changes — How reclassifying cannabis could unlock interstate transport and bank lending — and reshape FundCanna's competitive edge.

Building for AI agents, not browsers — Why 90% of FundCanna's code is already AI-written, and what an MCP server for a lending company actually looks like.

Sharpest moments

On why scale doesn't work like normal e-commerce:

For us, it's like every single customer counts.

On the safety net built into FundCanna's model:

If you don't have revenue, you don't have to pay us back.

On engineering in 2026:

At this point, 90% of all software is written with AI.

On what Plaid does behind the scenes:

What Plaid is doing apparently is something really, really nasty.

On where commerce is heading:

We're not gonna be building websites anymore. We're gonna be building MCP servers.

Key takeaways

• FundCanna funds ~150–200 cannabis businesses a month (~$10M) in a market capped at roughly 50,000 eligible businesses — every customer and every partner relationship matters.
• Its lending model is revenue-based, not fixed-term: if a borrower's revenue drops to near zero, the obligation to repay stops.
• With Google and Meta ads off-limits, growth runs almost entirely through partner marketplaces and word of mouth in a trust-starved industry.
• Plaid's bank-verification service works fine with major banks but breaks down with smaller credit unions — forcing FundCanna to build its own bank-statement fraud-check pipeline.
• Cagdas expects Schedule III reclassification to open interstate transport and bank lending, triggering consolidation — but sees FundCanna's data and trust advantage as durable.
• The team already writes ~90% of its code with AI and is building an MCP server so AI agents can shop, compare rates, and finance purchases directly through FundCanna.