Peak Money Support Tooling Discussion — Summary & Action Items

Main Takeaways#

• Peak Money's current support setup is unsustainable. They use Zendesk with the Finn AI bot (set up by Britney, who has since left). No one on the team is familiar with the tooling, the bot can't access the gated app or codebase, and Felipe is manually triaging tickets every 1–2 days.
• Volume is low but quality is poor. ~44 tickets in the system; some Finn responses are inaccurate (e.g., a yield/APY question where the bot misdiagnosed the issue and a user nearly churned despite depositing ~$30k).
• Peak Money wants their own internal agent with access to the codebase, database, and the ability to open bug tickets / PRs automatically — not a third-party tool that lacks internal context.
• The AZ team's prototype (built for Toby) is a strong fit. It generates draft responses with human-in-the-loop review, connects to internal docs and databases via Console GPT (BigQuery/Postgres), surfaces customer context, and self-improves hourly from resolved tickets.
• Strategic alignment: Felipe sees value in a shared "central intelligence" approach across Dapper rather than each team building their own — a win for both sides as a case study and potential future product.
• Felipe's feature suggestions:
  • GitHub / PR integration so the bot knows when fixes ship
  • MCP support so agents can interact with the tool programmatically
  • A chat interface (like Customer.io's "ask agent") for natural-language operation
  • Reports / context exports the bot can push back into Peak Money's repo for their own AI context
• VIP user handling — power users / large depositors should get differentiated, more personal treatment vs. fully automated responses.

Action Items#

• Felipe to sync with Bart about onboarding Peak Money onto the AZ support tool and report back by tomorrow (May 5).
• Felipe to flag Dan (Dapper support) and Alan & Rena (Octopus) as additional people for the AZ team to talk to.
• AZ team (G / Ben / Simon) to prepare for a follow-up sync to scope changes needed for Peak Money's use case.
• AZ team to consider / scope:
  • GitHub integration
  • MCP interface
  • Chat UI for the agent
  • Report generation back into customer repos
• AZ team to think through how Peak Money's adoption fits the productization path (internal user vs. external customer model).