Mintlify just raised $45 million at a $500 million valuation to solve one problem: making company knowledge readable by AI agents. The same problem exists for you as an individual. Your bookmarks, saved posts, and notes are personal documentation that AI agents cannot access. The tools to fix this now exist. Most people have not connected them yet.
On 14 April 2026, Mintlify closed a $45 million Series B. The investors were serious: Andreessen Horowitz, Salesforce Ventures, Bain Capital Ventures. The valuation: $500 million.
Mintlify makes developer documentation tools. But this round was not really about documentation. It was about something bigger.
Han Wang, Mintlify’s co-founder, put it plainly: “This round accelerates our mission to become the knowledge layer that makes products understandable, usable and discoverable by AI agents.”
Not readable by humans. Readable by AI agents.
That is the shift. And it has a direct parallel in your personal life that almost nobody is talking about.
The enterprise problem, stated simply
Companies produce enormous amounts of knowledge. Docs, wikis, runbooks, API references, internal guides, decision logs. Most of it is technically accessible to humans but practically invisible to AI agents.
An AI agent can browse the web. It can read a PDF you paste in. But it cannot natively search your company’s internal Confluence, query your product docs, or pull facts from a proprietary knowledge base. Not without infrastructure designed specifically to expose that knowledge.
Mintlify’s $500 million bet is that building this infrastructure is a product category worth a serious amount of money. That making knowledge AI-readable matters more than making it human-readable, because AI is now the primary way people interact with software.
They are almost certainly right.
But here is what the enterprise funding round misses: the same problem exists for individuals. Quietly, at a smaller scale, with no VC money and no dedicated infrastructure team. Just you, with years of curated bookmarks that your AI agent has never once been able to see.
What you have been building without realising it
Think about the last six months of things you saved.
On X, you bookmarked threads about pricing strategy, AI tools, product launches from companies you follow. On Reddit, you saved posts about technical problems, career discussions, niche communities in your field. On LinkedIn, you saved articles from people whose thinking you find valuable.
That collection is not random. You filtered it deliberately. Every save was a decision: this is worth keeping.
That is documentation. Personal documentation. A curated knowledge base built around exactly the things you care about.
The difference between a company’s internal wiki and your bookmark collection is scale, not kind. Both are knowledge repositories. Both exist because someone decided certain information was worth preserving. Both are now useless to AI agents because neither is exposed through the right infrastructure.
Sébastien Dubois, writing in March 2026, identified this gap clearly. His framing: structure is what makes knowledge machine-readable. The information exists. The AI is capable. The missing piece is structured access.
Your bookmarks are unstructured. Scattered across three platforms. Locked behind closed APIs. Siloed in a way that serves the platforms, not you.
Why the platforms are not going to fix this
There is a common assumption that X, Reddit, or LinkedIn will eventually sort out their own bookmark problems. They won’t. At least, not in the way that matters here.
X added keyword bookmark search in July 2024. It is unreliable. Users report exact keyword matches returning no results. The display limit sits around 800 bookmarks, with no warning when you hit it. New saves fail silently.
Reddit’s saved posts have had no search for over eight years. The limit is 1,000 posts. When you hit it, the oldest disappear without any notification. There is no export.
LinkedIn has no search for saved posts at all. Posts disappear when an author deletes them or when a connection drops.
These platforms built bookmarks as a secondary convenience feature, not as personal knowledge infrastructure. They have no incentive to make your saved content accessible to AI tools outside their own ecosystems. Your data is useful to them precisely because it stays locked inside.
So the infrastructure has to come from somewhere else.
The knowledge you are leaving on the table
Here is what the gap costs you in practice.
You spend three months following a topic closely. Pricing strategy. AI tooling. A specific market you are researching. You save forty threads, posts, and articles representing hours of deliberate reading and curation.
When you ask Claude a question on that topic, it has no idea any of that exists. It gives you a technically correct but completely generic answer drawn from its training data. Nothing from your months of careful accumulation reaches it.
The answer you get from Claude without your context is the same answer anyone would get. But you are not just anyone on this topic. You have been building specific, filtered knowledge about it.
That gap between a generic answer and a specific one is where real value lives. And right now, most people are leaving all of it on the table.
The enterprise-to-consumer gap
Mintlify is building the knowledge layer for companies. Nozomio is building context infrastructure for enterprise AI agents. A wave of well-funded startups is attacking the problem of making structured company knowledge available to AI in real time.
Almost none of them are building for individuals.
That gap is where most people currently sit. Your personal knowledge exists. Your AI agent cannot access it. There is no Mintlify for your bookmarks.
This is partly a tooling problem and partly an awareness problem. The tools to fix it now exist. But almost no one knows how to connect them.
The protocol that makes this possible is MCP.
What MCP actually changes
The Model Context Protocol (MCP) is an open standard created by Anthropic that lets AI tools connect to external data sources. Instead of copy-pasting information into every conversation, you connect a source once and the AI can pull from it dynamically.
The MCP specification defines a standard interface for AI tools to query external servers and receive structured results. Claude Desktop, Claude Code, Cursor, and Windsurf all support it. Over 10,000 MCP servers exist as of early 2026.
The relevant part for personal documentation: MCP lets you expose your own knowledge to Claude. Not as a one-time paste. Not as an attachment. As a live, searchable data source that Claude can query during any conversation, without you lifting a finger.
Connect your Notion workspace via MCP and Claude can search your notes mid-conversation. Connect your Google Drive and Claude can retrieve documents when they are relevant. Connect your bookmarks and Claude can search everything you have saved across X, Reddit, and LinkedIn.
That last one is the missing piece. Until recently, exposing social bookmarks via MCP required building your own server. Most people are not going to do that.
Making your bookmarks AI-readable
ContextBolt is the consumer answer to Mintlify’s enterprise problem.
Install the Chrome extension. Visit your X bookmarks page once and your collection starts building automatically. Every save from X/Twitter, Reddit, and LinkedIn is captured in the background. AI assigns a main topic and specific tags to every bookmark. The whole collection becomes semantically searchable: search by meaning, not keywords.
The Pro tier at £4/month adds the MCP endpoint. Four tools: search your bookmarks by meaning, browse all your topic clusters, retrieve bookmarks from a specific topic, get your most recent saves.
Add it to your Claude Desktop or Claude Code config and your bookmark collection becomes live AI context. Not a read-it-later archive you never revisit. Not a write-only drawer of things you once found interesting. A personal knowledge layer Claude can query when it is relevant.
That is what Mintlify is building for companies at $500 million valuation. This is the version for individuals at £4 a month.
What this looks like in practice
The practical version is simpler than the framing suggests.
You ask Claude a question about a topic you have been following closely. Instead of a generic answer, Claude runs a semantic search across your bookmarks and surfaces the eight most relevant saves from your collection. The answer is informed by content you specifically selected as valuable over months of deliberate reading.
That is the difference between a general-purpose AI and one that knows your specific slice of the world.
You do not have to maintain anything after the initial setup. ContextBolt captures every new save automatically. The collection grows every time you bookmark something. The value compounds over time.
Context engineering is the broader skill that underpins this. A personal documentation layer is its most practical application for everyday users. You are not engineering prompts. You are engineering what Claude knows before you ask the question.
The uncomfortable take on all of this
Mintlify’s $500 million proves something that most AI coverage has not said directly: personalised knowledge is not a nice-to-have for AI agents. It is foundational infrastructure.
Companies are spending serious money to build this for their internal knowledge. Individuals get nothing, because the economics of VC do not support building for individual users at £4 a month.
But the underlying problem is identical. An AI agent without access to your specific knowledge gives you generic answers. It does not matter how capable the model is. A more powerful Claude with no access to your data is still just giving you the same answer it gives everyone else.
The models are getting better fast, and they will keep getting better. But the differentiation will increasingly be what data each person has connected. The people who build a personal documentation layer now will get compounding returns from it. Those who rely on generic AI will get generic AI.
Your bookmark collection is high-signal personal data you have been accumulating for years. Most of it is better curated than the average web page on the same topics, because you filtered it yourself.
The personal AI context stack you build over the next six months will matter more than which AI subscription you pick. Start with what you have already built.
ContextBolt captures your X/Twitter, Reddit, and LinkedIn bookmarks automatically. Free tier includes 150 bookmarks with AI tagging, topic clustering, and semantic search. Pro (£4/month) adds the MCP endpoint so Claude can search your entire collection mid-conversation.