July 1, 2025

Meet ContextLens — Memory That Connects the Dots

Meet ContextLens — Memory That Connects the Dots

Meet ContextLens — Memory That Connects the Dots

We’re thrilled to announce the launch of ContextLens — the new memory engine that brings intelligent, connected reasoning to Datoshi.

It’s more than an upgrade. ContextLens replaces our previous memory system with something fundamentally better: a semantic graph that understands your data universe, keeps track of what matters, and helps Datoshi reason more like a human—or better yet, like a teammate.

What Is ContextLens?

ContextLens is a semantic graph of everything you work with in Datoshi — and how it all fits together.

Whenever you upload a dataset, connect a data source, read a document, or run a query, ContextLens turns that into a node, and maps its relationships to other nodes in your workspace. These connections — between documents, tables, datasets, SQL, and even notes — become part of a living knowledge graph.

It’s like a visual brain for your AI assistant — one that actually understands how things are related, not just what you uploaded.

You can explore all of this yourself in the new ContextLens page in your workspace panel — a fully interactive Visual Graph Explorer where you can browse and filter every node and connection.

How Does It Work?

  1. Everything becomes a “node”
    ContextLens sees each element — datasets, documents, database tables, notes, and queries — as a distinct object with meaning.

  2. It connects the dots
    If a dataset powers a query, and that query supports a note, ContextLens links them — so your assistant knows how your data world fits together.

  3. It retrieves with understanding
    When you ask a question, ContextLens finds relevant nodes and intelligently expands to include related ones — giving the assistant deeper context to respond meaningfully.

  4. You can explore the full graph
    Head to the ContextLens section to see every node, trace connections, and filter what you see — perfect for navigating complex data relationships.

Example Use Cases

  • You’re analyzing customer retention and wonder:
    “What’s our churn trend over time?”
    ➤ ContextLens recalls your data, remembers the definition of churn you used last week, and gives you an updated answer.

  • You come back after a few days and ask:
    “What did I note about VIP customers again?”
    ➤ If you’ve annotated that node before, ContextLens brings your own words back into the conversation.

  • You upload a PDF forecast and ask:
    “Compare this document’s projection with actuals from the database.”
    ➤ The AI connects your uploaded PDF to your live data sources, surfacing only what matters.

  • You revisit a chart and ask:
    “What query did I use for that revenue metric?”
    ➤ Jump into the graph and trace the lineage back to the SQL — fully visible, fully linked.

  • You’re working across tabs and ask:
    “Pull all the KPIs we’ve tracked in this workspace.”
    ➤ ContextLens searches across nodes — datasets, queries, notes — and returns the key metrics you’ve saved.

  • You start a new conversation and ask:
    “Show me trends from the feedback document I uploaded last week.”
    ➤ The assistant retrieves the PDF, identifies the relevant content, and pairs it with supporting data for deeper insight.

Why It Matters

  • Smarter insights: More relevant, connected answers from your assistant — even across tabs and sessions.

  • Less repetition: No need to remind the AI what your data means — it already knows.

  • Cross-source reasoning: Your documents, databases, and dashboards no longer live in silos.

  • Clarity & control: See exactly what’s stored, how it’s connected, and when something was added — all from the graph.

  • Teamwide memory: Teammates benefit from shared knowledge without starting from scratch.

Explore It Yourself

Check out the new ContextLens page in your user panel. It’s where the entire memory graph lives — complete with filters, search, and visual navigation of your nodes.

This is your data universe, made visible and searchable — so you and Datoshi can work with full context.