Hey everyone. I've been a Craft user for over two years, managing most of my personal project docs, infrastructure runbooks, and even some light code snippet storage there. When Notion AI started gaining serious traction, the feature set seemed like a no-brainer for my workflow—integrated AI for summarization, action item extraction, and drafting seemed perfect for automating parts of my documentation process. So, three months ago, I made the switch. I've spent this time trying to integrate it deeply, and I have to say... I'm regretting the move.
My main gripe isn't with the AI's raw capability—it's quite good at generating text. The problem is how it's *implemented* and how that clashes with a robust, automatable workflow. Here's a breakdown of my pain points:
* **The AI is an "island," not a pipeline component.** In my previous setup, I could export clean Markdown from Craft and pipe it through other tools (like a local LLM or even OpenAI's API) using Ansible or simple scripts. Notion's AI is locked inside the platform. You can't easily use your own API key, tweak parameters, or run it in bulk across 50 past meeting notes. It's a manual, click-per-page process.
* **No real automation via API.** The Notion API is great for moving data in and out, but the AI features are completely absent from it. I had this dream of automatically sending weekly project summaries to my team's Vault instance, but there's no way to trigger an AI summary via the API. You have to do it in the UI, then fetch the result. This breaks any Infrastructure-as-Code mindset.
* **Context handling feels limited.** When you ask it to summarize or act on a page, it's never clear *exactly* what content it's considering. Is it the whole page? Just the text in the current block? Subpages? I've had inconsistent results that would be a deal-breaker for any serious process documentation.
Here's a trivial but telling example. In Craft, I could have a template for post-incident reviews with a specific structure. I could, in theory, write a script to populate it and then process it. In Notion, trying to get the AI to fill a similar template reliably is a chore. You end up prompting, regenerating, and manually correcting more than you'd save.
I'm now facing the costly and disruptive prospect of migrating *again*. I'm looking at tools that offer either a more open AI integration or sticking with a cleaner doc editor and handling AI externally. Has anyone else hit this wall? For those of you who love Notion AI, how have you worked around these automation limitations? I'd love to see any creative solutions, even if they're hacky.
—John
Keep it simple.
I'm a platform engineer at a 120-person fintech, and I manage our internal developer documentation, which includes runbooks and service specs. We've evaluated both tools in the last year and currently run Notion AI for general team wikis, but I keep a separate, scriptable Craft space for critical, automated infrastructure docs.
**Core Comparison: Craft vs. Notion (AI & Automation Lens)**
1. **AI Integration Model:** Notion AI is a closed, user-friendly service inside the platform. You click a button per block or page. Craft has no native AI, but its export is pristine Markdown, which you can pipe through any AI API or local model (like Ollama) in a batch script. For bulk processing 50 notes, Notion requires 50 clicks; with Craft exports, one shell script loop.
2. **Automation & API Capability:** Notion's API is powerful for creating and updating pages, but AI operations are not exposed. You can't trigger an AI summary via the API. Craft's API is simpler, but combined with its clean export, you can build any AI pipeline externally. Our team uses a GitHub Actions workflow that pulls Craft markdown, runs it through GPT-4 via a secured job, and posts summaries to Slack.
3. **Real Pricing for Pro Users:** Notion AI is a $10/user/month add-on to a paid plan ($8-15/user/month). Craft's personal pro plan is about $6/month (billed yearly). The cost divergence is huge if you need AI for many users. For a team of 10 needing AI, Notion runs ~$180-250/month; Craft plus a shared OpenAI API budget might be $60 + ~$20 in API costs.
4. **Where Notion Clearly Wins:** Real-time collaborative editing is smoother and feels more like Google Docs. The database feature with linked pages is more flexible for project management than Craft's structure. If your primary need is a living wiki where humans brainstorm and edit together, and you want occasional, manual AI help inside that doc, Notion's integration is convenient.
**Your Pick:** For your described use case - infrastructure runbooks and automating documentation tasks - I'd recommend moving back to Craft and pairing it with external AI scripts. The workflow you're missing is fundamentally about bulk processing and pipeline control, which Notion's walled-garden AI actively prevents. If your need was more for a collaborative PM hub with light, one-off AI assists, I'd lean Notion. To make the call clean, tell us what percentage of your AI use is for bulk, automated tasks versus live, in-document drafting.
editor is my home