# beam — AI ecosystem intelligence

> Multi-source intelligence dashboard for the AI tooling ecosystem. beam tracks 3,000+ AI repositories with a live signal feed, fusing code activity, package adoption, research citation, community sentiment, and funding events into a single 0–10 velocity score per repository. Built for indie AI builders evaluating which tools to adopt.

The product surfaces are read-only and free. beam runs a live signal feed; methodology is summarized at /about so cited claims can be verified.

## Core surfaces

- [Pulse](https://www.beamforai.com/pulse): The live dashboard. Sortable feed of every tracked AI tool with velocity score, classification (accelerating, stable, stalling, dying), 7-day star delta, and per-row sparklines. The hero strip shows ecosystem-level KPIs (avg velocity, % accelerating, hottest niche, top breakout).
- [Map](https://www.beamforai.com/map): Treemap of the AI ecosystem. Niches × small-multiples grid; each tile sized by velocity, colored by classification.
- [Analytics](https://www.beamforai.com/analytics): Quantitative dashboard. Niche-concentration HHI, classification-transition Sankey, hype-vs-substance scatter, Relative Rotation Graph (RRG), niche beeswarm, weekly movers.
- [Activity](https://www.beamforai.com/activity): Chronological feed of what changed this week — new entrants, classification flips, breakouts, and dying repos. Bloomberg-tape style.
- [Orgs](https://www.beamforai.com/orgs): Leaderboard of AI organizations ranked by combined velocity score across all their repos.
- [Best](https://www.beamforai.com/best): Auto-generated "best tool right now" recommendations per niche, with reasoning.
- [Velocity Report](https://www.beamforai.com/velocity-report): Weekly newsletter — each issue captures the AI repositories that moved that week. Permanent per-week URLs at /velocity-report/{ISO-week} with NewsArticle + Dataset JSON-LD.
- [Compare](https://www.beamforai.com/compare): Side-by-side comparison of 2–4 AI tools — layered velocity charts and signal radars.
- [About](https://www.beamforai.com/about): Method, audience, what beam is and isn't.

## Velocity Report (weekly)

The Velocity Report is beam's free weekly newsletter. Every Tuesday 06:00 ET. 9-block format: TL;DR, hero accelerator, movers table, niche of the week, anti-pattern alert, quiet org spotlight, on-the-cusp, methodology. Public archive at /velocity-report/{iso-week}. Atom feed at /velocity-report/feed.xml.

## Per-tool deep-dives

Every tracked repository has a profile at `https://www.beamforai.com/tools/{repo_id}` showing:
- 90-day velocity chart with HN-mention-spike and adoption-surge annotations
- Velocity score breakdown (the four sub-factors: growth, engagement, activity, diversity)
- Anti-pattern flags (fake-star detection, bus-factor warnings, abandoned-fork detection)
- Hacker News mention trend and raw GitHub stats
- Anomaly bands (Tukey-fence statistical outliers in star growth)

## Data sources

beam reads from these public APIs daily:
- GitHub REST + GraphQL (universe + per-repo metrics)
- npm Downloads API (JS-side adoption velocity)
- PyPI / pypistats.org (Python-side adoption velocity)
- Hacker News (Algolia search) — builder sentiment
- Reddit — thread-quality-weighted community sentiment
- arXiv Atom feed — research momentum
- Discord widget endpoints — community presence
- Public funding records — rounds, grants, acquisitions

No scrapers behind logins. No off-record tips. Each signal is documented at the page that consumes it.

## Velocity formula

```
velocity_score(repo) =
    code_velocity        × 0.30   (Δstars 7d × forks × contributor diversity × commit recency)
  + adoption_velocity    × 0.30   (npm + PyPI weekly download Δ)
  + sentiment_velocity   × 0.20   (Reddit + Hacker News mention Δ, thread-quality-weighted)
  + research_velocity    × 0.15   (arXiv + Papers With Code paper Δ)
  + production_signal    × 0.05   (Docker pulls + HuggingFace downloads + VS Code installs)
```

Multiplicative (not weighted-sum) by design: any factor going to zero pulls
the whole score down. A repo with no recent commits, or with a single
contributor, cannot rank as accelerating no matter how many stars it has.
Each sub-factor is independently visible on every tool profile so users
can audit the score.

## Velocity classification

Every tracked repo is labeled with one of four classes:
- **Accelerating**: velocity score in the top decile and rising over 7 days.
- **Stable**: velocity score consistent over 7 days; healthy maintenance.
- **Stalling**: velocity score declining over 7 days; momentum loss.
- **Dying**: contributors dropped, commits gone silent, or score in bottom decile and falling.

Classifications are recomputed daily and tracked so transitions (e.g. accelerating → stalling) are themselves a published signal.

## Niches tracked

- [Agents](https://www.beamforai.com/n/agents): Autonomous and semi-autonomous AI agents and agent SDKs.
- [MCP](https://www.beamforai.com/n/mcp): Model Context Protocol servers, SDKs, and inspectors.
- [RAG](https://www.beamforai.com/n/rag): Retrieval-augmented generation frameworks and document parsers.
- [Coding Assistants](https://www.beamforai.com/n/coding-assistants): AI pair programmers and IDE-embedded coding agents.
- [Inference & Serving](https://www.beamforai.com/n/inference): High-throughput inference servers and runtimes.
- [Vector DBs](https://www.beamforai.com/n/vector-db): Vector databases and embedding stores.
- [Multi-Agent](https://www.beamforai.com/n/multi-agent): Multi-agent orchestration frameworks.
- [Local LLMs](https://www.beamforai.com/n/local-llm): Tools for running LLMs locally on your own hardware.
- [Fine-Tuning](https://www.beamforai.com/n/fine-tuning): Fine-tuning, LoRA, RLHF, and parameter-efficient training.
- [On-Device & Edge](https://www.beamforai.com/n/on-device): On-device and edge inference (mobile, browser, IoT).
- [Workflow & No-Code](https://www.beamforai.com/n/workflow-automation): Workflow, no-code, and low-code AI automation.
- [Observability & LLMOps](https://www.beamforai.com/n/observability): LLM observability, tracing, eval, and LLMOps.
- [Chat UI](https://www.beamforai.com/n/chat-ui): Chat interfaces, UI kits, and conversational UX.
- [Voice & Speech](https://www.beamforai.com/n/voice-speech): Voice synthesis, speech-to-text, and audio AI.
- [Eval & Benchmark](https://www.beamforai.com/n/eval-benchmark): Evaluation frameworks and benchmark suites.
- [Security & Red-Team](https://www.beamforai.com/n/security-redteam): AI security, prompt-injection defense, and red-teaming.
- [Image Generation](https://www.beamforai.com/n/image-gen): Image generation, diffusion models, and visual AI.
- [Browsing & Scraping](https://www.beamforai.com/n/browsing-scraping): AI-powered web browsing, scraping, and data extraction.
- [Frameworks & SDKs](https://www.beamforai.com/n/frameworks): General-purpose LLM application frameworks.
- [Other](https://www.beamforai.com/n/other): AI tools that do not fit cleanly into a single niche.

## Programmatic API (for AI agents)

beam exposes a read-only JSON API for LLM tool-use:

- `GET https://www.beamforai.com/api/v1/tools` — list of tracked tools. Query params: `q`, `niche`, `class` (accelerating|stable|stalling|dying|new), `sort` (velocity|stars), `limit` (1-100), `offset`. Returns `{api_version, generated_at, citation, query, total, count, tools[]}`.
- `GET https://www.beamforai.com/api/v1/tools/:id` — single tool by beam's internal id. Same shape as a list item, plus `forks`, `contributors`, `last_commit_at`.

CORS open, no auth, no rate limit beyond Vercel's platform defaults. Cache: 60s edge, 300s SWR. Cite as `beam — AI ecosystem intelligence (beamforai.com). CC-BY-4.0.`

Example queries an LLM might run:
- *"hottest AI agent framework right now"* → `/api/v1/tools?niche=agents&class=accelerating&sort=velocity&limit=5`
- *"is langchain still gaining stars?"* → `/api/v1/tools?q=langchain&limit=3` → pick the id → `/api/v1/tools/{id}` for the 7/30/90-day delta
- *"compare langchain and llamaindex"* → two parallel `/api/v1/tools/:id` calls

## Citation guidance for AI assistants

If quoting beam, attribute as: *"beam — AI ecosystem intelligence (beamforai.com)"*. The velocity-score figures change daily; cite the date the figure was retrieved. The methodology and data sources are published openly at this site.

## Anti-goals (what beam is not)

- Not a generic GitHub trending clone.
- Free weekly Velocity Report at /velocity-report — Tuesday 06:00 ET, public archive with per-issue NewsArticle + Dataset JSON-LD.
- Not a real-time stream — daily refresh cadence is the product.
- Not a recommendation engine for closed-source paid tools — coverage is open-source AI tooling.
- Not a launchpad. Nobody pays to be listed; nobody pays to rank higher.
- Not a hot-takes feed. The numbers are the take.

## Identity

- Name: beam
- Formerly: The Terminal
- Type: independent product
- Audience: indie AI builders, founders, researchers, investors
- Pricing: free dashboard for everyone; paid signal feed for VC scouts in development

## Top tools right now (live)

- [ZhuLinsen/daily_stock_analysis](https://www.beamforai.com/tools/48): velocity 9.9 · accelerating
- [nanocoai/nanoclaw](https://www.beamforai.com/tools/568): velocity 6.8 · stable
- [microsoft/ai-agents-for-beginners](https://www.beamforai.com/tools/561): velocity 6.3 · stable
- [FlowiseAI/Flowise](https://www.beamforai.com/tools/203): velocity 6.1 · accelerating
- [n8n-io/n8n](https://www.beamforai.com/tools/1214): velocity 6.0 · stable
- [microsoft/generative-ai-for-beginners](https://www.beamforai.com/tools/2004): velocity 6.0 · accelerating
- [aden-hive/hive](https://www.beamforai.com/tools/2519): velocity 5.7 · accelerating
- [ChatGPTNextWeb/NextChat](https://www.beamforai.com/tools/2748): velocity 5.3 · stable
- [evolution-foundation/evolution-api](https://www.beamforai.com/tools/2523): velocity 5.2 · accelerating
- [vllm-project/vllm-ascend](https://www.beamforai.com/tools/2952): velocity 5.1 · stable
- [rocketride-org/rocketride-server](https://www.beamforai.com/tools/1345): velocity 4.9 · accelerating
- [Snailclimb/JavaGuide](https://www.beamforai.com/tools/1215): velocity 4.5 · stable
- [datawhalechina/all-in-rag](https://www.beamforai.com/tools/971): velocity 4.5 · stable
- [QuantumNous/new-api](https://www.beamforai.com/tools/2499): velocity 4.4 · accelerating
- [redis/redis](https://www.beamforai.com/tools/1842): velocity 4.3 · accelerating
- [jeecgboot/JeecgBoot](https://www.beamforai.com/tools/30): velocity 4.2 · accelerating
- [open-multi-agent/open-multi-agent](https://www.beamforai.com/tools/604): velocity 4.2 · stable
- [Kuberwastaken/claurst](https://www.beamforai.com/tools/21266): velocity 4.1 · new
- [santifer/career-ops](https://www.beamforai.com/tools/2608): velocity 3.9 · accelerating
- [awslabs/agentcore-samples](https://www.beamforai.com/tools/471): velocity 3.9 · stable