Ask HN: What are tools you have made for yourself since the advent of AI?

Hacker News Discussion ↗

LLMs and coding agents are enabling developers and hobbyists to rapidly build highly personal tools, from home-automation dashboards, finance trackers and note systems to niche utilities like custom CAD pipelines, radio dashboards, game mods and even CNC controllers. Many describe “vibe coding” entire apps and frameworks in languages they barely knew before, using AI for boilerplate, integration work and debugging while still curating architecture, security and quality by hand. A recurring theme is that AI finally makes it worthwhile to create bespoke software and automation for narrow, individual needs—often replacing commercial SaaS—with trade‑offs around jankier UIs, hallucinations and the need for stronger guardrails and sandboxes.

Types of Tools People Built

  • Personal productivity: custom dashboards, task/goal trackers, habit and time trackers, personal finance/budgeting tools, invoicing and timesheet automation.
  • Developer tooling: agent harnesses, sandboxes, CLIs, Git helpers, TUI SQL and calendar clients, code visualizers, tmux replacements, debugging helpers.
  • Domain‑specific apps: CRMs for niche sales, scouting/volunteering organizers, education tools (language tutors, music theory, system‑design simulators), research/meeting summarizers.
  • Media tools: podcast and YouTube summarizers, RSS and news aggregators, music players/Spotify/YT replacements, image upscalers and background removers, audio editors and TTS pipelines.

AI’s Effect on Individual Development

  • Many say AI made them finally tackle “someday” projects (own language, CAD kernel, databases, static site generators, complex dashboards).
  • People with little or no experience in a language/framework shipped substantial apps (e.g., 300‑file Go CLI, SwiftUI/macOS apps, Rust tools).
  • A common pattern is “vibe coding”: letting agents write most code, then manual review/patching. Others deliberately push against this with strict harnesses, tests, and spec‑driven workflows.

Agents, Sandboxes & MCP

  • Heavy experimentation with local or containerized agent environments: podman/docker/VM‑based sandboxes, permission gates to prevent destructive commands, rootless container orchestrators.
  • Growing ecosystem around MCP and tool APIs: personal knowledge bases, search over logs, Obsidian/notes, business metrics, and email; some build general MCP servers and proxies to compose many tools.

Personal Data, Search & Knowledge Management

  • Strong interest in self‑hosted search and recall: indexing visited web pages, local files, ebooks, tweets, calendars, Slack/Mattermost/Asana/Dynalist archives.
  • Many tools export/import to Markdown, Obsidian, org‑mode, or SQLite so agents can operate over a consistent local corpus.

Home Automation & Physical World

  • Projects tie AI and scripting to Home Assistant, lights, blinds, HVAC, 3D printers, radios, drones, ESP32s, and CNC machines.
  • Some emphasize non‑digital tools (ceramics, woodworking, farm automation), using AI mostly to learn or prototype.

Attitudes & Skepticism

  • Enthusiasm: “personal app store” feeling; bespoke tools beat ad‑ridden SaaS; AI as “exoskeleton” for solo devs.
  • Concerns: hallucinations in legal/compliance contexts, fragile agent behavior, code quality slop, proliferation of low‑effort paid apps, and over‑reliance on opaque models.