Crisis

Podcast 2 — In conversation

Speaker A: The company is in full crisis management mode — the product recall announcement is going out today and every communication from here has to be coordinated, consistent, and honest. There's no margin for mistakes.

Speaker B: Crisis management as a professional discipline under maximum pressure — the product recall is one of the classic corporate crisis scenarios. Coordinated, consistent, and honest is the correct three-word crisis communication framework: inconsistency between different company spokespeople is fatal; dishonesty, when discovered, converts a crisis into a catastrophe. The margin for error in a product safety crisis is essentially zero.

Speaker A: He went through a genuine personal crisis in his mid-thirties — everything he'd built his identity around suddenly felt hollow. He came out the other side more certain of who he actually was, but the process was difficult.

Speaker B: Personal crisis as the Greek turning point in individual form — the mid-thirties identity disruption is a well-documented psychological phenomenon. The hollowness feeling marks the moment when an external identity structure — career achievement, relationship, social status — stops being sufficient. The coming out the other side with greater self-knowledge is the recovery the Greek medical crisis model predicted: the turning point resolves toward health or toward worse.

Speaker A: The climate crisis is unusual in that it develops on a slower timescale than political attention spans — it's hard to sustain crisis mode for decades.

Speaker B: The temporal mismatch of slow-developing crises with short attention spans is one of the most important challenges in crisis communication. Genuine crises that develop over decades rather than days or weeks are structurally incompatible with the crisis discourse that political and media systems have evolved to handle. The slow crisis requires sustained institutional response rather than emergency reaction.

Speaker A: People sometimes confuse crisis and emergency.

Speaker B: An emergency requires immediate action to prevent immediate harm — it is acute. A crisis is a broader state of severe difficulty or uncertainty — it may develop over time and require sustained management. All emergencies are crises; not all crises are emergencies. A financial crisis may unfold over months or years; the moment a bank run begins is an emergency within that crisis.

Speaker A: And crisis is different from disaster. A disaster has already occurred — the harm has been done. A crisis is the decisive period in which the outcome is still uncertain. A crisis that is mishandled may become a disaster; a crisis that is managed well may resolve without disaster. The crisis is the period of jeopardy; the disaster is the worst possible resolution of a crisis.

Podcast 3 — Prompt engineering

Speaker B: Scenario one — a crisis management schema.

Create a PostgreSQL schema for a crisis management platform. Tables: crises, crisis_timeline, crisis_communications, stakeholders, action_items. crises stores crisis_id, crisis_title, crisis_type enum operational, reputational, financial, safety, technical, natural_disaster, severity enum low, medium, high, critical, status enum monitoring, active, contained, resolved, onset_datetime, and crisis_owner_id. crisis_timeline stores event_id, crisis_id, event_datetime, event_description text, event_type enum trigger, escalation, decision, action, communication, resolution, and recorded_by. crisis_communications stores comm_id, crisis_id, audience enum internal_all, leadership, media, customers, regulators, message_text text, send_datetime, channel text, and approved_by.

Speaker B: Crisis_type enum covering operational through natural_disaster allows the platform to apply type-specific playbooks — a reputational crisis requires a different response workflow from a technical crisis or a safety crisis.

Speaker A: Crisis_communications with an audience enum and an approved_by field enforces the crisis communication discipline of coordinated, consistent messaging — no communication goes out without approval, and the record shows who approved it.

Speaker B: Next — a crisis communication playbook prompt.

Design a crisis communication playbook generator for a corporate communications platform. Given a crisis type (product safety, data breach, leadership scandal, service outage) and severity level, the playbook generates: a first 24 hours timeline with specific communications actions and their audiences, template holding statements for media enquiries that acknowledge the crisis without premature commitment to facts still being established, internal communication templates for staff that are honest about the situation and clear about what is and is not known, social media monitoring priorities specific to the crisis type, and the key questions that leadership must be prepared to answer at the first media briefing.

Speaker A: Holding statements that acknowledge without premature commitment is the crisis communication precision that prevents dishonesty-by-inaccuracy — saying we are aware of the situation and investigating is honest; saying there was no data breach before the investigation is complete risks a far worse later correction.

Speaker B: Right. Now an operational crisis monitoring prompt.

Build a real-time crisis detection and alerting system for a platform operations team. The system monitors: error rate above 5 percent of requests (crisis threshold triggering P1 alert), p99 response latency above 5 seconds for 3 consecutive minutes (performance crisis), database connection pool above 90 percent utilisation (resource crisis), payment failure rate above 2 percent (revenue crisis), and security anomalies including authentication failure spike above 10 times baseline (security crisis). Each crisis trigger creates an incident record with crisis severity, affected systems, time of onset, and immediately pages the on-call engineer with a crisis summary.

Speaker B: Payment failure rate above 2 percent as a named crisis threshold is the business-impact monitoring that pure technical metrics miss — a system that is technically performing well but failing payments is in a commercial crisis that technical dashboards alone would not surface.

Speaker B: Fourth — a mental health crisis resource prompt.

Design a mental health crisis resource directory for a wellbeing platform. The directory covers: crisis support services categorised by crisis type (mental health crisis line, domestic abuse crisis line, financial crisis counselling, bereavement crisis support, addiction crisis line), each with contact method (phone, text, online chat), availability hours, target population (all adults, young people, LGBTQ-specific), and language support. A crisis type quiz guides users to the most relevant resources for their current crisis. The directory is updated quarterly and each service has a last-verified date to ensure contact details remain accurate.

Speaker A: Last-verified date per service entry is the maintenance discipline that makes the directory trustworthy — a crisis resource directory with outdated phone numbers is worse than no directory at all.

Speaker B: Fifth — a product crisis decision tree prompt.

Build a product crisis decision tree for a product manager facing a significant production incident. The tree guides the PM through: first, assessing whether the incident is a crisis (is it affecting more than 10 percent of users, is it a safety or data issue, is it causing measurable revenue loss?) and escalating immediately if yes, second, determining whether to rollback (is the fix worse than the bug? Is rollback faster than the forward fix?), third, deciding on customer communication timing (communicate early with incomplete information, or wait for confirmed details?), fourth, the post-crisis review process. The decision tree produces a recommended action path with rationale for each decision point.

Speaker B: Communicate early with incomplete information versus wait for confirmed details is the crisis communication decision that has no universally right answer — the tree must surface the trade-offs of each choice rather than prescribing one path.

Speaker B: Sixth — a full small app.

Build a single-page web app called Crisis Log — a personal and team crisis and incident journal. The user logs crises and significant incidents with: title, date and time, crisis type (personal, professional, technical, financial), severity 1 to 5, a brief description of the situation at onset, actions taken, and the outcome and lessons learned once resolved. The app shows: all crises by date, severity distribution, a most common crisis type summary, and a lessons learned feed showing all outcome notes in reverse chronological order. A search allows finding past crises by keyword. All data in localStorage.

Speaker A: Lessons learned feed separate from the crisis list is the knowledge management output that makes the journal genuinely useful — the insights from resolved crises need to be surfaced independently so they can be reviewed without revisiting the full crisis records.

Speaker B: That's the lesson — a crisis is the decisive turning point when the outcome is genuinely unresolved. The Greek physician watching their patient at the turning point of an illness was managing a crisis in the original sense. Every operational team watching their dashboards during a service degradation, every communications director facing a media crisis, every individual facing a personal identity crisis — they are all at the same kind of decisive moment. How the turning point is navigated determines whether the outcome is recovery or disaster.

Speaker B: Prompt example: