Your personal knowledge graph, built from real work.

Over a career you accumulate people, companies, conversations, decisions, commitments, and hard-won context. That web is uniquely yours. Researchers have described a personal knowledge graph as structured information about entities personally related to you, their attributes, and the relations between them: a spiderweb with you at the center.

In practice it is simpler. A personal knowledge graph is the connected memory of your work life: who you know, where they work, what you discussed, what you decided, what still needs doing, and how those pieces relate. It is not a pile of files. It is the map that makes the files useful. When the map is missing, even great notes become hard to reuse under time pressure.

Most people keep that map in memory, email archives, chat history, a CRM they never update, and three note apps from three different jobs. The edges go fuzzy. The details fade. Changing companies resets half the board. Grafite exists to capture the graph as you work, then keep it searchable and askable.

A knowledge base is a filing cabinet: notes, docs, bookmarks, transcripts. Searchable if you remember the right keyword. Mostly static. The items do not really know about each other.

A knowledge graph maps relationships. It is not only that you know Sarah. It is that Sarah works at Acme, you have met twelve times this year, you mainly discuss rollout risk, she introduced you to their CTO in March, and last call she mentioned a vendor evaluation in Q4. That is context you can act on, not a contact card you forget to update.

People connect to companies. Companies connect to meetings. Meetings connect to decisions and tasks. Tasks connect back to people. The more you capture, the more the web compounds. Month two is more useful than month one because new work links to old work.

Before a client call, you open the person or company view and see the last three conversations, open tasks, and the awkward pricing moment you annotated last month. You are not reconstructing history from memory five minutes before join.

After a busy week, you ask what still needs follow-up with a specific account. The answer comes from meetings and tasks together, not from a separate todo app that never got updated.

When a topic returns months later, you can find the original decision and the reasoning around it. That is the opposite of "I think we talked about this somewhere." It is the practical payoff of connections, not just storage.

When you change roles, your graph still exists. Some notes you shared with the old team stay shared. Your private reflections, relationship history, and career-spanning context stay with you. That portability is a core reason we insist on personal ownership language.

Documents matter. Email matters. Still, a surprising amount of the context that makes you effective lives in meetings and informal voice notes: the hesitation before a yes, the side constraint nobody wrote down, the introduction that later becomes a deal, the commitment that never made it into the tracker.

If you capture those moments and link them to people and tasks, you get a graph spine that other systems rarely create on their own. A CRM has contacts but weak conversation memory. A doc tool has pages but weak relationship history. A meeting bot has transcripts but often weak ownership and weak surrounding product.

Grafite's bet is that the lightest useful personal knowledge graph starts where work actually happens, then grows outward. Meeting capture, editor, people, tasks, search, and Ask Grafi are not separate product lines. They are the minimum set that makes the graph real.