Capability

Podcast 2 — In conversation

Speaker A: We're doing a capability review across the team this quarter.

Speaker B: Mapping who can do what?

Speaker A: Exactly. Current capabilities, gaps, and where we need to build.

Speaker B: Capability mapping is a useful exercise. People often have hidden capabilities nobody's tracking.

Speaker A: Right. And we found three people with data-analysis capability that we weren't using at all.

Speaker B: Classic capability gap in reverse — people are capable but not deployed.

Speaker A: The other direction is also common. We've got a capability gap in machine-learning model evaluation.

Speaker B: Is that a hire or a build?

Speaker A: Hire short-term, build long-term. You can't close a technical capability gap with wishful thinking.

Speaker B: True. Capability takes time.

Speaker A: I was reading about Amartya Sen's capability approach too. Very different use of the same word.

Speaker B: Yes — the idea that human development is about expanding what people can actually do, not just giving them resources.

Speaker A: People sometimes confuse capability with capacity.

Speaker B: Important difference. Capacity is about volume — how much can be processed. Capability is about type of skill — what kind of thing can be done.

Speaker A: So you can have high capacity but limited capability, or vice versa.

Speaker B: And don't confuse capability with potential. Potential is theoretical; capability is demonstrated.

Speaker A: Save capability for things that can be shown and measured.

Podcast 3 — Prompt engineering

Speaker B: Scenario one — a model evaluation prompt.

Evaluate the coding capability of this language model across five dimensions: syntax correctness, algorithm choice, edge-case handling, test coverage suggestion, and code readability. Use ten representative tasks from our production codebase. Score each dimension 1 to 10 and flag any capability gaps versus our previous model.

Speaker B: Capability gaps versus the previous model forces a comparative output.

Speaker A: And scoring each dimension makes the capability measurable, not just described.

Speaker B: Next — a database schema.

Create a PostgreSQL schema for a learning management system. Tables: learners, capabilities, assessments, capability_scores. capabilities has a name, domain enum technical, soft_skill, leadership, and a target_level integer 1 to 5. capability_scores records learner_id, capability_id, current_level, assessed_at. Index on learner_id and capability_id.

Speaker A: Capability_scores as a separate table lets you track growth over time.

Speaker B: Right. Now a UI side panel.

Design a Capability Radar side panel for an HR dashboard. Show each team member as a spider-chart across six capabilities: coding, communication, delivery, architecture, testing, and mentoring. Highlight capability gaps against the team's target profile. Include a 'plan development' action.

Speaker B: Capability Radar is a specific chart type that the model renders correctly when named.

Speaker B: Fourth — a strategic planning prompt.

Write a 300-word capability gap analysis for a fintech startup planning to build an in-house ML team. Cover current capability, required capability in 18 months, skill-gap for each role, build versus buy for each gap, and the three highest-priority hires.

Speaker A: Capability gap analysis is a recognised strategic document — the model knows the format.

Speaker B: Fifth — a technical prompt.

Audit this microservices architecture for missing observability capabilities. Check for distributed tracing, structured logging, metric collection, alerting, and canary analysis. For each missing capability, output the gap, the recommended tooling, and an implementation effort estimate in engineering-weeks.

Speaker B: Missing observability capabilities structures the audit around a precise checklist.

Speaker B: Sixth — a full small app.

Build a single-page web app called Capability Mapper for engineering managers. Inputs: team members and six rated capabilities per person. Output: a heatmap showing team strengths and gaps, a top-three recommended hires, and a download of the map as a PNG. Persist data in localStorage.

Speaker A: Capability Mapper is a practical tool with immediate team-planning value.

Speaker B: That's the lesson — capability is measured, specific, and demonstrable. Use it in prompts where the quality and type of ability matter more than raw volume.

Speaker B: Prompt example: