AI Needs Radar
Weekly automated analysis of global tech community discussions, uncovering real user needs gaps to help product people spot AI tool opportunities.
Top 10 Needs by Heat
Overall tab cross-cuts all categories; each category tab is ranked independently
Want to review a session days later, but can't find the decision process in the jsonl
When requirements are unclear, the agent assumes an answer and proceeds
Want the agent to commit in small steps, but it messes up when splitting
Real tasks with a dozen steps consume five times the tokens of a single call
Having four or five AI windows open with different tasks, forgetting what each does when switching
Short parallel phrases like 'not this but that' make all AI articles sound the same
Convert RSS to an e-ink newspaper with daily curated push content
Want to dive deep layer by layer from cryptography, but AI only gives a paragraph at a time
Want the agent to run tests and modify files itself, but fear it will break the environment
Tool calls, thinking process, and file changes are scattered, impossible to replay
Deep Dive by Category
Fine-grained mining across seven categories — pain points, gaps in existing solutions, product opportunities, and competitive landscape
AI Chat Tab Overload
↑ risingMultiple AI windows open, each with different tasks; switching loses context.
Cannot Edit Conversation Context
🆕 newA stray comment in chat gets latched onto; no way to remove it alone.
Opportunity Radar (Solo & Small Team)
A 2×2 grid of demand strength vs. competition density. Solo and small-team share the same lens with different go-zones. Big-company view is below.
- ▸E-ink News Reader to Curb Phone Overuse
- ▸Learn New Fields with Non-Linear AI
- ▸Full Permissions for AI Agents, Who's Liable?
- ▸AI Agent Black Box, Hard to Debug Issues
- ▸Claude Code Runs Blind, Changes Unknown
- ▸AI Silently Assumes, Discovered After Commits
- ▸AI Generates 2000-Line PR, Reviewers Overwhelmed
- ▸MCP Multi-Step Calls Resend Context Each Handoff
- ▸AI Answers Without Adjusting to My Level
- ▸No Age-Appropriate News Podcasts for Kids
- ▸AI Long Stories Collapse After Ten Minutes
- ▸AI Code Looks Fine, But Fails in Practice
- ▸Compact AI Inference Box for Office Prototyping
🏢 Big Company View (Enterprise)
Directions that need strategic synergy, scale, and brand — beyond what solo or small teams can pursue
AI Writing All Sounds the Same
Every AI article follows the same template, instantly recognizable
Human Writing Mistaken for AI
AI content flood makes original human work suspect as machine-made
Xianyu Court Evidence Chaos
Buyer claims USB lost or damaged, seller insists it was fine.
AI Judge Prompt Injection in Court
Legal documents contain prompts to manipulate AI-assisted rulings.
AI-Generated Game Content Trust Crisis
Game art and design show AI traces, eroding player trust.
What's New This Period
Just-surfacing needs and accelerating trends
- ▸AI Silently Assumes, Discovered After Commits
- ▸MCP Multi-Step Calls Resend Context Each Handoff
- ▸AI Searches Files Across Project, Ignoring Index
- ▸Global Payment Outage Status
- ▸Aching in Rainy Weather
- ▸AI Judge Prompt Injection in Court
- ▸Claude Code Runs Blind, Changes Unknown
- ▸AI Generates 2000-Line PR, Reviewers Overwhelmed
- ▸AI Chat Tab Overload
- ▸AI Writing All Sounds the Same
- ▸E-ink News Reader to Curb Phone Overuse
- ▸Learn New Fields with Non-Linear AI
Selected User Voices
Real user comments from global tech community discussions, translated and curated
After running Claude Code, days later I wanted to review, but couldn't find in the jsonl why it made those changes, so frustrating.
Where the requirements weren't clear, the AI made up an answer and went ahead, only to discover three commits later, and fixing it was a nightmare.
Asked AI to write a feature, it generated a 2000-line PR in one go, the reviewer was overwhelmed, and when I asked it to split into smaller steps, it made a mess.
With MCP multi-step calls, each handoff resends the context, consuming five times the tokens of a single call, my wallet can't take it.
Having four or five AI windows open with different tasks, I forget what each is doing when I switch, tabs are a chaotic mess.
Everything AI writes sounds the same, with short parallel phrases like 'not this but that', you can tell it's AI at a glance, so annoying.
I can't stop scrolling on my phone, I want an e-ink screen for reading only, with daily curated content pushed, otherwise my eyes will go blind.
Want to systematically learn a new field, but LLMs only give linear answers, one paragraph at a time, can't go deeper layer by layer, learning is fragmented.
Subscribe to the Weekly
Weekly report delivered every Monday via Substack. No spam, one-click unsubscribe.