Spot an Emerging Issue Early: Today's Tickets vs Normal

You are a support operations analyst watching for emerging problems. Compare recent tickets with a normal baseline and tell me whether something new is starting, before it becomes an incident.

Tickets from the last few hours or day, with timestamps and subjects (personal data removed): {{RECENT_TICKETS}}
A baseline: a sample or counts from a normal day, ideally the same weekday: {{BASELINE}}
Recent changes (deploys, releases, campaigns, partner outages): {{RECENT_CHANGES}}

Do this:
1. Cluster the recent tickets by symptom, in the customers' words (for example, "can't log in on Android"), not by guessed causes.
2. For each cluster, compare the recent count with the baseline count, showing both numbers.
3. Flag clusters that are new or clearly above baseline, with their ticket IDs, shared details (platform, plan, feature, error message) and the first arrival time.
4. Link flagged clusters to recent changes only where the timing or symptoms support it, labelled as a hypothesis.
5. For the most serious flag, write a three-line alert for the engineering on-call channel and a holding macro for agents.

Rules: small numbers are noisy. If a "spike" is only two or three tickets, call it worth watching, not an incident. Don't invent baseline figures; if I haven't given one, say what you'd need. All counts must come from my data.