Roadmap

Where Negitama is going, milestone by milestone — what each goal means, what has shipped, and what comes next.

Milestone 1

Highly curated AI news

Shipped

Get highly curated news items into readers' hands every day — manual curation by an editor at the center, and public APIs exposing the everyday news items so Negitama works as a platform, not just a site.

Technology achieved

  • Always-on data warehouse — twice-daily ingestion of ~18 AI companies, GitHub releases, and 5.7M+ company records, self-healing scrapers
  • Editor workspace — manual curation with sticky filters and upvote-driven selection
  • LLM-in-the-loop summaries — multi-provider fallback chain with per-model test buttons
  • Everyday newsletter system — select → generate → publish → email, with a public issue archive
  • Public APIs — contribute/ingest API (Swagger + OpenAPI), newsletter items API, MCP server
  • Distribution — dark-theme email broadcasts and a human-approved LinkedIn automation heartbeat
Milestone 2

AI tools catalog & hybrid search

Targeting now

Specifically target all AI tools: capture each tool's metadata and give every tool its own page. Attach its blogs, tutorials, know-hows, latest releases, and latest capabilities — then a hybrid search so anyone can query the top tools with a natural-language question, with community voting feeding the ranking.

Already in place

  • Tool search — full-text with pricing/badge filters and paginated cards
  • Tool voting and comments via the community layer
  • Blog platform (Markdown/GFM) — the foundation tutorials will thread onto

Building

  • Per-tool canonical pages — metadata, official links, related news, linked tutorials, kept live as the tool evolves
  • Tool metadata harvesting at warehouse level
  • Tutorials & know-hows linked to each tool, refreshed on every update
  • Hybrid search — keyword + semantic, so “best open-source coding agent” just works
  • Vote-informed ranking of query results
Milestone 3

Living topics

Half built

News items are ephemeral. Topic modeling merges related items — early and late — into one single evolving source: a tool, a company, or an AI theme. Multiple AI agents enrich each topic so any reader can see when it started, what happened next, whether it subsided, and if there was hype — whether it was justified or called off.

Shipped half — the community of news

  • Community feed with trending & latest sorting
  • Voting and comment threads on every news item and tool
  • Experience posts — first-hand write-ups attached to the feed

Building

  • Topic modeling & merging — cluster related items across time into one topic
  • Topic pages with an origin-to-now timeline
  • Multi-agent enrichment — AI agents append context, updates, and verdicts
  • Hype ledger — was the hype justified, or called off?
Milestone 4+

Open — your input wanted

Open for feedback

More milestones are being considered — this list is deliberately open for feedback and suggestions. Current candidates:

  • Open registration and Google OAuth
  • Tool discounts for community members
  • Community leaderboard & reputation
  • Marketplace & early access — maker advertising, sponsored launches
  • Access tiers — gated early access, premium data & API
  • Signal-grade news intelligence — Signal Score, “Why Featured”, EN/中文