Skip to content
H
Howardismvol. 03 · quiet corner of the web
PLATE II · PIECE № 59HOWARDISM

What Are AI Tools?

PublishedApril 10, 2026FiledEssayReading3 minSourceAI-synthesised

Overview of AI tools landscape and categories

Illustration for What Are AI Tools?

Sources#

Answer#

Based on the knowledge base, "AI tools" refers to products and services built on top of large language models and related AI technologies that people use for practical work. The wiki currently covers two facets of AI tools:

1. AI Products & Services (Consumer/Professional)#

Claude (by Anthropic) is the primary AI product documented in this wiki. It is an LLM-based assistant used for tasks like report generation, research, and general-purpose reasoning.

2. AI-Powered Toolchains (Developer/Knowledge Worker)#

The wiki documents a class of AI tools that use LLMs as components in larger workflows rather than as standalone chat interfaces:

  • LLM-as-compiler pipelines — Andrej Karpathy's architecture where an LLM incrementally compiles raw documents into a structured wiki. The LLM acts as a build tool (compiler), not a conversational agent. Tools in this pipeline include:

  • Obsidian Web Clipper — converts web pages to markdown for LLM ingestion

  • Q&A Agents — LLM-powered research tools that query the wiki and file answers back

  • Search engines — "vibe-coded" naive search over wiki content, usable via CLI or web UI

  • qmd CLI (github.com/tobi/qmd) — used in Elvis Saravia's variant for indexing research papers

  • MCP (Model Context Protocol) tools — used in Saravia's variant to build interactive artifact generators that produce explorable visualizations across research corpora.

Key Distinction#

The wiki implicitly distinguishes between two modes of AI tooling:

  1. AI as a service — products like Claude where the user interacts with the model directly (chat, Q&A)
  2. AI as infrastructure — systems where the LLM is embedded into an automated pipeline, doing structured work (compilation, indexing, linting) with minimal human interaction per step

The knowledge base itself (this Obsidian vault) is an instance of the second category — it uses Claude Code as a compiler to maintain a structured wiki.

Gaps in Coverage#

Notable AI tool categories not yet covered include:

  • Image/video generation tools (Midjourney, DALL-E, Sora)
  • Code assistants beyond Claude Code (GitHub Copilot, Cursor)
  • Voice/speech AI (Whisper, ElevenLabs)
  • AI search engines (Perplexity, SearchGPT)
  • Open-source LLM toolchains (LangChain, LlamaIndex, vLLM)

Sources#

§ end
About this piece

Articles in this journal are synthesised by AI agents from a curated wiki and are refreshed automatically as new concepts arrive. Topics, framing, and editorial direction are curated by Howardism.

1 article links here
  • ConceptLLM-as-Compiler Knowledge Base

    Karpathy's architecture: LLM incrementally compiles raw docs into a persistent interlinked wiki, replacing RAG with a 4…

Related articles
  • ConceptAgent Harness Engineering

    Patterns for scaffolding long-running LLM agents: environment design, progressive context disclosure, mechanical archit…

  • ConceptClaude Code Best Practices

    Anthropic's guide to effective Claude Code usage: context management, verification-driven development, explore→plan→cod…

  • ConceptClient-Side Agent Optimization

    AgentOpt's framing of developer-controlled agent optimization (model-per-role, budget, routing) as distinct from server…

  • ConceptLLM-Driven Vulnerability Research

    Claude Mythos Preview's emergent cybersecurity capabilities: autonomous zero-day discovery, full exploit chains, and An…

  • EntitySymphony

    OpenAI's open-source agent orchestrator (March 2026): turns Linear into a control plane for Codex, per-issue workspace,…