The framing is right — most support bots hand off to a human the moment things get real, so an agent that actually understands context and resolves issues is the harder, more useful problem to solve. Customizing to brand voice and policies is the part that usually makes or breaks this.
The open question is permissions and guardrails. "Handling real customer problems" means the agent needs access to take actions — refunds, account changes, bookings — and that's where the risk lives. Would love to see how it scopes what the agent can do, where it escalates to a human, and how every action is logged before trusting it in production.
DORA metrics are something a lot of teams either track by hand or skip entirely, so pulling them automatically from GitHub activity is a practical angle. The useful part is turning raw git data into exec-ready reports a manager can actually act on, instead of another dashboard nobody opens. Looks like a genuinely handy take on engineering visibility.
The scoping is smart. GA can technically show signup funnels, but for a solo founder it's overkill — you fight event setup, sampling and cookie-consent just to answer one question: where are people dropping off. fnel collapses that to one snippet and a real-time funnel, which is exactly the level most indie makers actually need.
The flip side of that focus is the "why not just use GA or PostHog for free" question. To justify being another paid script, it needs a sharper edge than "simpler GA" — deeper drop-off diagnostics, retention over time, or being genuinely faster from install to first insight. Curious how it holds up once someone needs more than a single funnel.