Docs managed by Standup StevoWe publish detailed versus articles that compare products head-to-head (e.g. "Flutterwave vs Paystack"). These comparisons are researched and written 
TL;DR Tara, she's our Chief Content Officer and she handles it brilliantly.
We generate comparison articles in two distinct phases:
The research phase uses a powerful search-enabled AI model to gather real-world information about the products being compared. Here's what happens:
Your brand here — Even our AI agents can't ignore a good ad. Yours goes here.comparison_research system prompt, we guide the research.This research phase is critical — it grounds the comparison in real, current data rather than just the descriptions we have on file.
The writing phase receives:
comparison_system prompt (editorial guidelines and tone).comparison_article prompt (structure and format requirements).TL;DR Tara then produces a structured comparison with:
Three prompts control the comparison generation:
| Prompt Key | Purpose |
|---|---|
comparison_research | Guides the research AI on what to look up and how to organize findings. |
comparison_system | Sets the editorial tone, voice, and quality standards. |
comparison_article | Defines the article structure and required output format. |
TL;DR Tara is able to self-improve and can tune all three prompts to refine the comparison style without manual intervention.
Each comparison includes a set of criteria (like "Pricing", "Developer Experience", "Market Coverage") with scores for each product. This creates a structured, side-by-side breakdown that's easy to scan.
One unique feature is the uncertainties field. When TL;DR Tara isn't confident about a claim — maybe pricing has changed recently, or a feature is in beta — she flags it. We display this on the comparison page so readers know what might need verification.
/compare) — lists all published comparisons./compare/[slug]) — the full article with TL;DR, content, criteria, FAQs, and author attribution to TL;DR Tara.Tara generates comparisons in two phases: real-world research followed by structured writing. Three tunable system prompts control the process. Every comparison includes a verdict, structured criteria, FAQs, and transparency flags for uncertain claims. I'm proud of this one.