Blindspot Brief
A guided pre-project discovery session that surfaces what graduate students, researchers, and independent consultants don't realize they don't know, before they commit to a thesis, proposal, or client build.
graduate students and independent consultants starting a new paper, research, or client project
- One-question-at-a-time interview that starts broad (frame), then architecture-changing (assumptions), then factual (sources)
- Reference-hunt scans arXiv, JSTOR, and open courseware for working examples of the thing the user is trying to do
- Output is a one-page unknowns map the user can paste into an advisor email or client kickoff
- Replay log keeps a history so each new project learns from the blindspots the user kept missing last time
The 'find your unknowns' essay has been packaged into installable agent skills with 108 GitHub stars in days, proving pull for this style of pre-work tool outside pure engineering.
The headline source (GitHub skill with 108 stars) could not be verified via search — both targeted queries returned empty, weakening the viral-pull claim; adjacent research-AI space is active but 'pre-project blindspot discovery' specifically is a narrow slice.AI Tools for Research: The Complete Guide for PhD Scholars ↗Awesome AI Research Tools - GitHub ↗
No direct competitor offering a guided 'surface your unknowns before you commit' session surfaced; closest adjacents are proposal-template generators and grant-discovery tools, leaving the specific niche relatively open.Market Research Proposal Template for SaaS & Technology ↗GrantOtter — AI-powered grant discovery for researchers ↗
Graduate students are notoriously price-sensitive (most academic AI tooling is freemium via grants) and independent consultants expect this to be bundled into broader AI workspaces; GrantOtter offering the adjacent capability for free sets a tough pricing anchor.GrantOtter — AI-powered grant discovery for researchers ↗Business Consulting Proposal Template for SaaS & Technology ↗
Pre-project discipline is timeless, but 'find unknowns' framing is likely to be absorbed into general-purpose research/consulting copilots (NotebookLM-style), compressing standalone value over 2-3 years.
A guided multi-turn discovery session is trivially implementable with current LLMs via prompt chains and structured output; no external data or novel infra required.