AI that works with your thinking

Two modes of AI

There are two fundamentally different things you can do with AI in a knowledge tool.

You can use it to replace thinking: summarize this, draft that, answer this question. The AI does the cognitive work, and you approve or reject the output. This is useful and increasingly common.

Or you can use it to amplify thinking: surface this relevant context, connect these ideas, remind me of what I already know. The AI does the retrieval and synthesis work, and you do the actual thinking. This is rarer, and in our view, more valuable.

We built Vita around the second model.

Why replacement has limits

AI summarization and generation tools are genuinely impressive. They save real time on real tasks. But they have a structural limitation when it comes to knowledge work: they don’t know what you know.

When you ask an AI to draft something, it draws on its training data — which is huge, but generic. It doesn’t know about the specific insight you had in a meeting last Tuesday, or the framework you developed over two years of reading, or the customer conversation that changed how you think about the problem.

The result is output that’s competent but not distinctly yours. It’s the average of what many people might say, not the specific thing that only you could say based on what you specifically know.

What amplification looks like

The alternative is AI that works with your existing knowledge rather than around it.

Instead of generating from scratch, it retrieves from your archive. When you’re writing about a topic, it finds the notes and articles you’ve already captured that are relevant. When you’re making a decision, it surfaces the research you did three months ago that you’ve since half-forgotten. When you’re preparing for a conversation, it pulls together everything you know about the person or company you’re meeting.

The output isn’t AI-generated prose. It’s your thinking, better-resourced.

This is the mode where AI genuinely extends what a person is capable of, rather than substituting for it.

The trust problem

There’s another reason we prefer this model: trust.

When AI generates content for you, there’s always a verification burden. You have to check whether what it wrote is accurate, whether it matches your voice, whether it reflects your actual views. For important work, that verification can take almost as long as doing the work yourself.

When AI retrieves from your own notes and sources, the verification question is different. You don’t need to ask “is this true?” — you know it’s true, because it came from you. You only need to ask “is this relevant?”, which is a much lighter cognitive task.

How we think about it at Ancher

We’re not opposed to generative AI. Vita uses it. But we’re deliberate about when and how.

We use AI to understand what you’re working on so we can surface relevant context without requiring you to search. We use it to find connections between captures you might not have noticed. We use it to synthesize your existing notes into a starting point when you’re facing a blank page.

What we don’t do is generate content that gets mixed in with your knowledge as if it came from you. Your archive is yours — a record of what you’ve read, thought, and learned. Keeping it that way is what makes it trustworthy.

The goal is AI that makes you sharper. Not AI that thinks for you.