Consultant

Podcast 2 — In conversation

Speaker A: We have just brought in an external consultant to review our API architecture. My team is not sure how to work with them effectively.

Speaker B: That is a really common situation. The key is to treat the consultant as a temporary expert with access, not an outsider with a brief. Give them real context — past decisions, existing constraints, the history of why things are the way they are.

Speaker A: So they need the full picture, not just the pretty version?

Speaker B: Exactly. A good consultant is only as valuable as the information they receive. If you show them a sanitised version, you get a sanitised recommendation. The word consultant implies they are being brought in for honest diagnosis — not a rubber stamp.

Speaker A: What about the difference between a consultant and a contractor?

Speaker B: Good distinction. A contractor executes work — they write the code, build the feature, deliver the output. A consultant advises — they assess, recommend, and guide. Some people combine both roles, but the core distinction is: contractor does, consultant advises. You bring in a consultant when you need thinking, and a contractor when you need doing.

Speaker A: And in the NHS — senior doctors are called consultants. That is a different use entirely.

Speaker B: Yes, though the root meaning is the same. In British medicine, a consultant is the most senior hospital doctor in a specialty — the person others refer patients to for specialist opinion. They are consulted when a case exceeds general practice. The advisory meaning is absolutely still there; it is just a formal job title in that context.

Speaker A: What about close synonyms?

Speaker B: Adviser is broader and less formal — your financial adviser is not necessarily an external expert. Specialist emphasises expertise but not the advisory role. Expert implies skill but again not the engagement model. Consultant uniquely combines expertise, independence, and the structured advisory relationship.

Speaker A: Consultant — the outside expert who sees what the inside has stopped noticing.

Podcast 3 — Prompt engineering

Speaker B: Today we are looking at consultant as a role assignment in AI prompts. When you assign the AI a consultant identity, you unlock a very specific mode: independent analysis, frank assessment, and structured recommendation. It is the most powerful role you can assign for complex technical decisions.

Speaker A: I have used "act as an expert" before, but consultant feels more precise. It implies a brief, an output, and accountability.

Speaker B: Exactly right. A consultant has a deliverable. Let us start with UI design.

Speaker B: PROMPT_1: Act as a senior UX consultant hired to audit a SaaS product dashboard. The client has three problems: low feature discovery, high user drop-off on the settings page, and inconsistent colour use across the navigation. Produce a structured consultant report with an executive summary, prioritised findings, and specific redesign recommendations for each issue. Include one wireframe description per recommendation.

Speaker A: That is a full engagement brief packed into one prompt. The AI is being held to consultant-level output.

Speaker B: And that changes the quality of the response significantly. Now for database design:

Speaker B: PROMPT_2: You are a database consultant engaged by a fintech startup. Their PostgreSQL schema for transaction history is causing slow reporting queries at ten million rows. Review the current schema, act as a consultant who identifies the root cause, and deliver a remediation plan that includes index changes, a partitioning strategy, and a query rewrite example. Frame your output as you would present it to a non-technical CTO.

Speaker A: Non-technical CTO framing — that is smart. The consultant role forces the AI to translate technical findings into business language.

Speaker B: Good observation. Now for application development:

Speaker B: PROMPT_3: Act as an independent software consultant brought in to review a Node.js monolith before a planned microservices migration. Identify the top five risks of the migration, recommend which services to extract first and why, and propose a phased migration roadmap. Include your reasoning as a consultant would — showing trade-offs rather than just prescriptions.

Speaker A: Showing trade-offs — that is a uniquely consultant output. Not just "do this" but "here is why this and not that."

Speaker B: Precisely. Now for a full application build with consultant framing:

Speaker B: PROMPT_4: You are a technology consultant engaged to deliver an inventory management system for a wholesale distribution company. Begin with a consultant-style discovery summary covering user roles, core workflows, and integration requirements. Then produce the full application using React, Node.js, and PostgreSQL. Structure the code as you would hand it off to an in-house team — with a README, folder conventions, and inline comments explaining architectural decisions.

Speaker A: Discovery summary first — like a real consulting engagement starts with a needs assessment before any delivery.

Speaker B: Exactly. Now for security:

Speaker B: PROMPT_5: Act as a security consultant performing a pre-launch assessment of a REST API. Review the authentication mechanism, input validation approach, and error disclosure behaviour. Produce a consultant-style risk register with each vulnerability rated by likelihood and impact, followed by a remediation roadmap ordered by priority. Include code examples for the top three fixes.

Speaker A: Risk register and remediation roadmap — proper consultant deliverables, not just a list of issues.

Speaker B: One final prompt — for team process consulting:

Speaker B: PROMPT_6: You are an Agile delivery consultant engaged by a software team that is consistently missing sprint goals. Interview the team by asking me five diagnostic questions one at a time. Based on my answers, produce a consultant report identifying the root causes and a tailored process improvement plan. Include specific ceremony changes, tooling recommendations, and a thirty-day milestone plan.

Speaker A: Interactive consulting — the AI asks questions before it advises. That is incredibly realistic.

Speaker B: And that is the power of the consultant role in prompts. It sets expectations: you are not here just to generate, you are here to assess, understand, and then deliver something the client can act on. Consultant is the word that turns the AI from a tool into a partner.