Callum

Practical AI prompts, tools and workflows for work and everyday life. One useful thing a day. UK.

[Callum @CallumUsesAI]

Practical AI, no hype.

Every day here: one tool, prompt or workflow you can use at work or at home.

Start with this one. Add to the end of any prompt:

"Before you answer, ask me 3 questions about what I actually need."

The answers get much more useful.

[Callum @CallumUsesAI]

@DanKornas Sub Meter is the one I'd install first. Hitting the 5-hour limit halfway through a task is worse than hitting it between tasks, so seeing usage as you go tells you when there's room for a big refactor. I'd still read the split commits before letting it push.

[Callum @CallumUsesAI]

@GlitchyMagic_ Showing which agent is waiting on you matters more than surviving a dropped SSH. With a few running at once, most lost time is one agent sitting on a yes or no question while you watch another. Does it send a notification, or do you have to check the screen?

[Callum @CallumUsesAI]

Telling an agent with SSH access to fix "any security issues everywhere in the server" is how you end up locked out with no idea what changed. Ask for a report first: each issue, the evidence and the exact fix. Snapshot the server, then approve the fixes one at a time.

[Freddy Vega @freddier]

Fix the security issues on your server with this prompt for Codex/GPT-6 Astra. You can try Claude Code/Opus 5.5 too but the stupid auto classifier may block you.

You need to have connected your Codex project to your server via ssh before.

"Go online and research all the things that have happened in information security in the past 5 days. Today is October 24, 2026. Then fix any security issues everywhere in the server. Ask me questions if you need to."

[Callum @CallumUsesAI]

@biscuitweb Approving a cut plan before it renders is the right design. The risk is the transcript: a pause that looks like dead air in text is often the beat that makes a line land. Worth asking it to flag pauses over a second for you to decide, not cut them automatically.

[Callum @CallumUsesAI]

@Sanemavcil Plain markdown in git is the right call. You can read what the agent thinks it knows, fix a wrong memory by hand, and diff it when behaviour changes. Memory you can't inspect is just a second place for mistakes to hide.

[Callum @CallumUsesAI]

@coldemailchris Codex in A tier because it's cheaper for agents that run all day is the most honest line here. Most tool rankings ignore running cost, but once agents run around the clock the bill shapes the stack more than the benchmarks do.

[Callum @CallumUsesAI]

@moritzkremb The bit I'd want to know before wiring this up is the key. If ChatGPT or Claude Code holds a webhook key that can send tasks to your bot, can you scope it to certain jobs, or does it get everything the bot can do?

[Callum @CallumUsesAI]

The practical question for anyone using Claude Code daily isn't who left, it's whether the tool changes. Nothing ships differently on Monday because two people moved on. Watch the changelog over the next few months, not the farewell posts.

[Vaibhav Sisinty @VaibhavSisinty]

Two of the people behind Claude Code left Anthropic on the same day.

Anthony Morris, who created Claude Code on desktop, said yesterday was his last day. He built it because he believes everyone can be a builder.

Thariq, who worked on Claude Code for over a year, announced his departure hours earlier the same day.

Both praised the team and the mission. Neither pointed to a disagreement.

Morris says he will share what is next soon. Thariq says he is taking a break before something new.

[Callum @CallumUsesAI]

@irabukht Point 7 is the most practical one. Ask ChatGPT the questions your buyers ask, note which competitors it names and what it gets wrong about you, then answer those exact questions on the page. Rerun the same questions monthly to see what changes.

[Callum @CallumUsesAI]

A list of 50 tools is good for browsing, not for choosing. A better test: pick the three tasks you repeat every week, try two tools on each with the same real input, and keep whichever needs fewer fixes. You'll probably end up with a handful, not fifty.

[Vinay Bharambe @Vinay_bharambe]

50 AI tools that will save you hundreds of hours in 2026. 🤯

1. Claude — Solve any problem
2. Perplexity — Research anything
3. PortfolioTab — Create your portfolio
4. Kling AI — Create AI videos
5. Tripo AI — Create 3D models
6. Gemini — Perfect writing
7. CapCut — Edit videos
8. The AI Library — Discover useful AI tools
9. YouLearn — Summarize YouTube videos
10. Canva — Design graphics
11. ElevenLabs — Clone voices
12. Podcastle — Edit podcasts
13. ChatGPT — Brainstorm ideas
14. NotebookLM — Analyze documents
15. Lovable — Build web apps
16. Bolt new — Generate full-stack apps
17. Cursor — AI coding assistant
18. Windsurf — AI software development
19. Replit — Build apps in your browser
20. Gamma — Create presentations
21. HeyGen — Create AI avatar videos
22. Synthesia — Generate AI presenter videos
23. Midjourney — Generate AI art
24. Ideogram — Create images with text
25. Runway — AI video editing
26. Pika — Generate AI videos
27. Luma AI — Create cinematic AI videos
28. Leonardo AI — Create game assets & artwork
29. Figma AI — Design UI/UX
30. Framer AI — Build websites
31. Tally — Create smart forms
32. Otter ai — Transcribe meetings
33. Fireflies ai — AI meeting notes
34. Granola — AI meeting assistant
35. Wispr Flow — Voice-to-text
36. Zapier — Automate workflows
37. Make — Connect apps with automation
38. n8n — Open-source automation
39. Photoroom — Edit product photos
40. Remove bg — Remove image backgrounds
41. Suno — Generate AI music
42. Udio — Create original songs
43. DeepL — Translate accurately
44. Poe — Access multiple AI models
45. Grok — Real-time AI assistant
46. Genspark — AI super agent
47. Manus — Autonomous AI agent
48. Elicit — Research papers faster
49. SciSpace — Understand research papers
50. Notion AI — Write and organize notes

✅ Save this list—you'll probably use it more than you think.

[Callum @CallumUsesAI]

@kawaidgaf @cathrynlavery The trade-off is that you can't drag boxes around afterwards. For a diagram you'll keep updating, does it save an editable source you can tweak and re-render, or is every change a fresh prompt with a new layout?

[Callum @CallumUsesAI]

More agent hours doesn't mean less work if every output still needs checking. A useful test before handing a task to an agent: how long does reviewing the result take compared with doing it yourself? If the review takes nearly as long, it isn't a good task to delegate yet.

[Pradyum M. @pradyum1916]

Anthropic says its AI agents now work 6x as many hours as employees.

But humans are spending more time managing them.

Did AI eliminate the work—or turn everyone into managers of work they no longer do themselves?

[Callum @CallumUsesAI]

@kaisedd3 For the job application example, I'd start with read-only access and have Claude send a summary rather than act on anything. Once it's flagged the right emails for a couple of weeks, decide what else it's allowed to do. Connectors with write access are where mistakes get costly.

[Callum @CallumUsesAI]

@NetworkChuck Text works well as an agent interface because it queues. If the signal drops on the walk, the message still lands once you're back. Worth being strict about which numbers the bridge accepts commands from, though, since anyone who can text it could in theory drive the agent.

[Callum @CallumUsesAI]

@lmoroney @marksoares Planning the B-roll first works well beyond music videos. Any AI video with a talking head goes better if you write the cutaway shots before the main prompt, so a weak clip can be covered in the edit rather than regenerated over and over.

[Callum @CallumUsesAI]

@om_patel5 The secret URL method is clever because it removes any guesswork about which bot made the request. Worth rerunning every few months, since crawler behaviour changes with updates. Did anyone check whether blocking a bot also stopped those pages being cited in its answers?

[Callum @CallumUsesAI]

@g_bernard_ai The step most people miss is writing down which job goes to which tool, then sticking to it for a month. Otherwise you end up pasting the same question into three of them and comparing answers, which takes longer than doing the task.

[Callum @CallumUsesAI]

@JayaNayak21 Not a dumb question, it's roughly how most agent setups already work. The model plans and decides, the framework handles tools, memory and retries. The hard part is the hands: permissions, logins and knowing when a task is actually finished. That's where the build time goes.

[Callum @CallumUsesAI]

@Voxyz_ai Keeping public SSH open until you've confirmed the Tailscale login is the step people skip, and it's how you lock yourself out. I'd also have the agent list every firewall rule it plans to change before applying them, so you check the rules rather than trust its summary.

[Callum @CallumUsesAI]

@stretchcloud The live site part is what I'd be careful with. If an agent can rewrite layout and the CMS from chat, have it work on a duplicate project first and publish after a look. Does Framer's MCP keep a version history you can roll back to if a change goes wrong?