Skip to content

Lacuna

Lacuna is Tiptree’s living library for machine learning research. It organizes papers, authors, institutions, venues, concepts, and research directions into a source-traceable research graph.

Althea can use that context directly.

When you’re viewing a paper, author, or research direction in Lacuna, Althea can start with that page’s context already attached.

That means you can ask:

“What is the core claim of this paper?”

“Compare this method to the closest work in the same direction.”

“What experiments would you run next?”

“Turn this into a short related-work memo.”

You don’t need to paste the title, abstract, author list, or surrounding research direction. Lacuna already knows where you are. Althea uses that context as the starting point.

The exact context depends on the page, but it can include:

  • paper metadata and summary
  • extracted concepts
  • figures and captions
  • neighboring papers
  • author and venue context
  • research-direction context
  • source links back to Lacuna and the original paper

Althea carries over the useful bits, rather than dragging the whole page into the chat.

This is useful when your question starts from a specific artifact:

  • “Explain this paper in the context of diffusion models.”
  • “Find the weakest assumption in this method.”
  • “What would I need to reproduce this result?”
  • “Which neighboring papers should I read first?”
  • “Is this paper enough evidence for the claim I’m making?”

For broader questions, use Literature Search. Lacuna context is a sharp starting point; Literature Search is the wider sweep.

Lacuna is built around provenance. When Althea answers from Lacuna context, it should preserve that habit: cite the paper, point back to the relevant Lacuna page when possible, and keep source trails visible.

This is especially important for scientific work. A good answer is not just plausible. You should be able to inspect where it came from.

Lacuna and Literature Search work well together.

Lacuna gives Althea a structured starting point. Literature Search expands outward from that point, retrieves more material, and synthesizes a report.

A typical flow:

  1. Open a paper or research direction in Lacuna.
  2. Ask Althea a focused question from that page.
  3. If the answer needs a wider sweep, ask for a literature search from the same starting point.
  4. Use the resulting report as a working artifact, with citations and source trails intact.

Lacuna also has an MCP surface for agents and local tools.

Use this when you want Lacuna research context outside the web app, for example from an agent runtime or local coding environment. The MCP package and install command depend on the release channel you’re using, so check the current release notes or your Tiptree workspace setup before wiring it in.

Once connected, Lacuna behaves like a research context provider: agents can search papers, inspect author or direction context, and pull source-grounded summaries into their own workflows.