Show Notes
In this daily update, Parker dives into a practical, punchy workflow for extracting better AI outputs with a single prompt idea, plus hands-on prompts, a fast book-outline exercise, and the Klein tool for planning versus acting. Real-world tests, cost notes, and quick takeaways sprinkled in.
Key takeaways
- A tight prompt structure plus source material can dramatically lift output quality in minutes.
- Treat prompts as a pipeline: source material → prompt generator → chat/completion → refine and reuse as presets.
- Use emotion/“stakes” prompts to nudge results, but use ethically and for clear, legitimate outcomes.
- Separate planning and execution when using AI for software/product work; Klein’s plan/act modes can help.
- You can draft a book outline in under 5 minutes with a structured, multi-prompt workflow.
- Track cost and model choice; small tweaks (plan vs act) can save tokens and dollars.
Workflow: how to get better titles and prompts fast
- Start with a source-driven prompt loop:
- Source material: reference a smart, concise guide or example you trust.
- Use a prompt generator (e.g., 03 mini High) to bootstrap prompts.
- Define a clear role, responsibilities, task, and constraints for the assistant.
- Include context and a concrete example of desired output.
- Core structure to copy/paste into prompts:
- Role: “You are a helpful YouTube title writing assistant.”
- Responsibilities: “Receive a first draft of a working title and a one-line description.”
- Task: “Return 10 increasingly more clickbaity titles.”
- Rules/constraints: “Incorporate curiosity loops, use specific language, punchy but concise.”
- Input/Output: “Input: draft title and one-line description; Output: 10 titles.”
- Tips:
- Use a prompt playground or companion to iterate quickly.
- Create a preset “YouTube title generator” for repeated use.
- Compare outputs side-by-side (model variation helps reveal what works).
Guilt/emotion prompts: adding stakes to prompts
- Concept: adding a psychological stake can improve output quality.
- Example prompt idea: frame outputs as if “it’s on the line for my career” or a historical/fictional threat (e.g., “The library burns if you don’t deliver”).
- Practical use: test with small, controlled prompts to see how outputs shift; refine with feedback.
- Caution: use responsibly; avoid manipulation or harm.
PRD planning vs execution: two camps, one goal
- Big idea: code and product work splits into two roles, often mapped to PMs (planning) and engineers (execution).
- Planning tools (e.g., Grok prompts) excel when you want structure and guarantees about the plan.
- Execution prompts focus on concrete code or implementation tasks; keep a tight scope to avoid drift.
- Rule of thumb: use planning prompts to outline PRDs, specs, and roadmaps; use execution prompts to implement.
Klein: plan vs. act for task-based work
- Klein offers a “Plan” mode (architect a plan, read data) and an “Act” mode (execute the task and report progress).
- Why it matters: context-window efficiency and accuracy improve when you break tasks into discrete steps.
- Practical demo notes:
- Create a task (e.g., add a new blog category called Daily Uploads).
- Use Plan to sketch the steps; switch to Act to execute with coding or database calls.
- Use the timeline to see before/after and auto-approve read/write commands as needed.
- Cost awareness:
- Plan steps can cost more per action; in Parker’s tests, plan runs charged around $0.80 for a broader styling update and less for simple reads.
- Use plan for complex changes; use act for routine tasks to save tokens.
Speedrun: speed-creating a book outline with prompts
- Objective: draft a short book outline and key themes in minutes.
- Process:
- Prompt 1: generate a detailed outline for a book about a topic (e.g., ChatGPT for professionals).
- Prompt 2: extract unique angles and a deep dive into key themes.
- Prompt 3: draft the first chapter intro and a kickoff for Chapter 1.
- Use a second window (or spreadsheet) to assemble the outline, themes, and chapter starts.
- Benefit: you end up with a working outline, a first chapter drift, and a ready-to-fill content plan in minutes.
- Next steps: feed the outline into a document or automation pipeline and iterate chapter-by-chapter.
Quick gains on the growth front
- School/paid product beta:
- Rapid early traction: 20 members in a few days; low-cost entry with plan to monetize later.
- Channel: niche AI education/bootcamp style, with cross-sell and outreach strategies in mind.
- Daily uploads channel:
- Early performance showing higher-perceived value per view; growth not linear but picking up.
- Focus: maintain consistent cadence, iterate on formats (shorts, cursor videos, quick prompts), and drive cross-channel referrals.
- Takeaway: stay disciplined on content quality and a tight value prop; experiment with paid/offers while growing the audience.
Practical takeaways you can apply now
- Build a reusable title prompt preset: role + responsibilities + task + rules + input/output.
- Always reference source material; use a prompt generator to spark variations before finalizing.
- Test model variations (e.g., medium vs high reasoning) side-by-side to find the best balance of quality and cost.
- Use emotion/stakes prompts sparingly to boost output quality when appropriate.
- Use Klein or similar plan/act tools to manage complex tasks with clear task boundaries and progress tracking.
- For content creation: create a rapid 5-minute outline workflow, then turn the outline into chapters and a workflow for automation or doc templates.
Links
- Cline (plan vs act workflow)
- Grok prompts (planning-focused prompts)
- OpenAI o3-mini prompt generator (prompt iteration helper)
- OpenAI Playground for prompt testing
- arXiv (source reference for "guilt/stakes" prompt concepts)
If you found value in this quick-fire session, drop a comment with what you’ll test first and any prompts you want analyzed. See you in the next update.