Skip to content
Tim van der Wijst · AI guides

Changelog

Public changelog: every substantive change to lessons, tools and comparisons, with dates.

AI moves fast, and so does this site. Here’s what we changed and when. Every lesson also shows its own changes at the bottom.

November 2026

  1. AI cost calculator: what does the API cost per month?

    Added security block: key abuse and hard limits.

  2. Best AI assistant for beginners in the Netherlands (2026)

    Filled in the ChatGPT price and the training status of Copilot (off in the new app) and Grok (on by default).

  3. Best AI assistant for beginners in the Netherlands (2026)

    Added security block: what not to paste, training and shared links, account security.

  4. Best way to make money with AI (2026)

    Added security block: the key security point per route.

  5. Creating SSH keys

    Expanded security block (agent forwarding, separate deploy key) and adding a passphrase to an existing key.

  6. Hardening your server

    Verified and expanded the ufw prompt, IPv6 rule, sudo-rs on 26.04, Hetzner menu paths and the backup price (20% of the server price).

  7. Hardening your server

    First version.

  8. Honest numbers: what do you really earn with AI automation?

    Verified Make's free plan (1,000 credits per month) via an archived copy of its pricing page.

  9. Honest numbers: what do you really earn with AI automation?

    Added security block: patching n8n, keys per client and approval for AI steps.

  10. Honest numbers: what do you really earn with AI automation?

    Fact-check applied: numbers, rates, sources and legal rules corrected.

  11. Honest numbers: what do you really earn with AI bots?

    Checked the Fiverr entry price for voice bots: basic packages start at $50-90 instead of $90.

  12. Honest numbers: what do you really earn with AI bots?

    Added security block: toll fraud, keys in the web widget and verifying webhooks.

  13. Honest numbers: what do you really earn with AI bots?

    Fact-check applied: numbers, rates, sources and legal rules corrected.

  14. Honest numbers: what do you really earn with AI content?

    Confirmed the ±13% drop per job with Upwork's own press release.

  15. Honest numbers: what do you really earn with AI content?

    Added security block: fake AI tools, fake sponsorship emails and keys per tool.

  16. Honest numbers: what do you really earn with AI content?

    Fact-check applied: numbers, rates, sources and legal rules corrected.

  17. Honest numbers: what do you really earn with apps and SaaS?

    Expanded security block: spend cap before launch and scanning for leaked keys.

  18. Honest numbers: what do you really earn with apps and SaaS?

    Fact-check applied: numbers, rates, sources and legal rules corrected.

  19. LinkedIn and X: from posts to clients

    Added security block: fake recruiters, app permissions, 2FA.

  20. LinkedIn and X: from posts to clients

    Fact-check applied: quotes, numbers and sources corrected; uncertain claims flagged.

  21. Malicious skills, MCP servers and extensions

    First version.

  22. Managing API keys and secrets

    Checked the push protection menu paths in the GitHub docs and clarified that the default applies to public repos.

  23. Managing API keys and secrets

    First version.

  24. Model names decoded

    Added security block: fake apps, wrappers, API keys.

  25. Model names decoded

    Trimmed to the essentials; added insider tips, short answer and FAQ.

  26. n8n vs Make vs Zapier: which one to pick in 2026?

    Added Make's paid price and noted that the Make and Zapier starting prices assume yearly billing.

  27. n8n vs Make vs Zapier: which one to pick in 2026?

    Added security block: who patches, locking down the editor and approval for AI steps.

  28. Personal AI agents compared: Hermes, OpenClaw, OpenBot and Grok Bot

    Added security block: OpenClaw sandbox, Hermes approvals and write access to your email.

  29. Privacy and account settings per assistant

    Verified and updated facts on ChatGPT, Grok, Copilot, Meta AI, Mistral and Claude: Copilot no longer trains on chats in the new app, Meta AI has Incognito Chat.

  30. Privacy and account settings per assistant

    First version.

  31. Prompt injection in practice

    First version.

  32. Security check for your app

    First version.

  33. SEO and GEO: get found by Google and ChatGPT

    Added security block: Search Console, domain, shared chats.

  34. SEO and GEO: get found by Google and ChatGPT

    Fact-check applied: quotes, numbers and sources corrected; uncertain claims flagged.

  35. Short video: TikTok, Reels and Shorts that stick

    Verified the TikTok rule on comment bait with the exact wording from the Community Guidelines.

  36. Short video: TikTok, Reels and Shorts that stick

    Added security block: channel hijacking, 2FA, schedulers.

  37. Short video: TikTok, Reels and Shorts that stick

    Fact-check applied: quotes, numbers and sources corrected; uncertain claims flagged.

  38. The dangerous combination: data, input and actions

    First version.

  39. Tokens and the context window

    Filled in the free Claude context window: up to 1 million tokens, varies by model.

  40. Tokens and the context window

    Added security block: context, prompt injection, API keys.

  41. Tokens and the context window

    Dutch-English gap updated with our token index: 24 to 37% more tokens.

  42. Tokens and the context window

    Trimmed to the essentials; added insider tips, short answer and FAQ.

  43. Vapi vs Retell vs ElevenLabs Agents: which voice platform to pick?

    Added security block: widget keys, toll fraud and logging per platform.

  44. What AI is and isn't

    Added security block: retention, work data, shared links.

  45. What AI is and isn't

    Added the Hallucination detective game.

  46. What AI is and isn't

    Trimmed to the essentials; added insider tips, short answer and FAQ.

  47. What do AI models cost via the API? (price comparison)

    Added security block: leaked keys, hard limits and one key per application.

  48. What is an LLM, really?

    Added security block: training, memory, feedback.

  49. What is an LLM, really?

    Trimmed to the essentials; added insider tips, short answer and FAQ.

  50. What you never paste into a chat

    First version.

  51. Which assistant should you choose?

    Added ChatGPT prices in euros (Dutch App Store, 2 November 2026).

  52. Which assistant should you choose?

    Added security block: account, phishing, extensions.

  53. Which assistant should you choose?

    Training switches now named exactly.

  54. Which assistant should you choose?

    Trimmed to the essentials; added insider tips, short answer and FAQ.

  55. YouTube and your newsletter: an audience that's yours

    Added security block: channel hijacking, newsletter takeover, phishing.

  56. YouTube and your newsletter: an audience that's yours

    Fact-check applied: quotes, numbers and sources corrected; uncertain claims flagged.

  57. Model names decoded

    All facts checked against sources and corrected.

  58. Tokens and the context window

    All facts checked against sources and corrected.

  59. What AI is and isn't

    All facts checked against sources and corrected.

  60. What is an LLM, really?

    All facts checked against sources and corrected.

  61. Which assistant should you choose?

    All facts checked against sources and corrected.

© 2026 Tim van der Wijst · Free AI guides for everyone. Made in the Netherlands.