TAG Weekly Roundup #8

Editor's Note

Two articles this week addressing different sides of the same problem - getting useful content from AI. The articles:

1. Perplexity Pages automating the research-to-publication pipeline, and

2. Systematic prompting frameworks  - turn vague requests into specific, contextual instructions.

🧰Tools

Perplexity Pages - Research to Publication in One Click

Perplexity's Pages feature transforms research threads into polished, shareable articles with professional formatting and multimedia integration. Users can generate content from simple prompts or convert existing research into publication-ready material.

Pages creates structured content that includes source citations, customisable formatting, and multimedia elements.

Like all AI tools, Pages output quality depends heavily on input quality. Vague prompts produce generic articles regardless of formatting sophistication. The platform's strength - automated structure and citations - can't compensate for unclear instructions or missing context. 

📽️Video

Write Prompts Like a Pro - Solving the Input Quality Problem

Everyone claims to have the "best" prompt framework, but most focus on structure rather than psychology.

RICECO works differently - it thinks about perspective, context, and boundaries before attempting solutions.

Why This Framework Actually Works

Unlike generic prompt templates, RICECO tackles the core problem: AI models fill gaps with generic responses when information is missing. The framework plugs those gaps by giving the AI the same contextual information a human expert would need.

The video proves this with identical prompts producing completely different results based purely on role assignment: 

• "Give advice on better sleep" (1.16), yields standard tips

• "You are a board-certified sleep doctor" (1.24)delivers evidence-based medical guidance

• "You are a sleep-deprived parent writing to other parents" (1.30), creates conversational, emotionally resonant advice suitable for blog content

The RICECO Components

• Role: Assigns expertise and perspective - transforms generic responses into specialist insights by defining who the AI should think like

• Instruction: Demands specificity over vague requests - "write an engaging YouTube short" becomes "write a 60-second script using curiosity gap hook and scroll-stopping visual anchor"

• Context: Provides background most people skip - audience, business scenario, purpose, and tone requirements that prevent generic outputs

• Examples: Shows rather than tells through few-shot prompting - demonstrates desired structure, formatting, and quality standards

• Constraints: Eliminates AI bad habits - wordiness, corporate buzzwords, repetitive phrasing by setting clear boundaries

• Output Format: Specifies deliverable structure - bullet points, tables, tweet threads, mind maps for immediate usability

Real-World Application

The video's real estate automation example shows systematic business thinking. Rather than asking "How can I implement AI in real estate?", the framework builds:

• Role: Business growth strategist specialising in AI adoption

• Context: Detailed business overview including lead sources, time allocation patterns, current operational challenges

• Constraints: $400 budget limit, maximum 3 hours weekly maintenance, non-technical solutions only

• Output Format: Prioritised action playbook with quick wins, core systems, and long-term growth phases

Result: customised recommendations targeting specific pain points with projected 8-12 hour weekly time savings, rather than generic advice.

Combining RICECO with Perplexity Pages - A Real Example

Take the quantum machine learning example from Perplexity's own Pages announcement. While the original provides solid technical coverage, applying RICECO methodology can produce more targeted results:

• Role: Business technology analyst preparing executive briefing for enterprise AI strategy team

• Context: Fortune 500 company evaluating emerging AI technologies, audience includes CTO and business unit heads

• Instruction: Analyse quantum machine learning focusing on business readiness, implementation timeline, and competitive implications

• Constraints: Executive-level language, include specific vendor examples, avoid theoretical physics, 2000 words max

• Format: Executive summary, business case analysis, vendor landscape, implementation roadmap

The result: A targeted business analysis covering market readiness, vendor landscape (IBM, Google, Microsoft), realistic deployment timelines, and enterprise use cases.

The difference: business-focused intelligence versus educational overview. Same research foundation, completely different audience value.

Condensed ICC & EIO approaches

The condensed ICC approach (Instruction, Context, Constraints) covers 80% of prompting needs. The follow-up EIO process (Evaluate, Iterate, Optimise) transforms initial outputs into polished results through systematic refinement cycles.

Most business analysts will find immediate value in applying role-based prompting for stakeholder communications, using constraints to eliminate corporate jargon, and specifying output formats that match deliverable requirements.

Till next week.