Asta's search now grades its own results and tries again
- Source
- x.com
- Date
Two updates to Asta, our ecosystem of AI agents for science: a one-click handoff from AutoDiscovery to Asta’s data analysis tools, & paper search that evaluates its own results + searches again when they fall short. 🧵
AutoDiscovery explores your datasets & surfaces hypotheses potentially worth further investigation. Meanwhile, DataVoyager is our agent for data-driven discovery & analysis. When a hypothesis surprises you, the natural next move is to interrogate it—now that's one click.
Click “Explore with Asta” on any AutoDiscovery hypothesis, and DataVoyager opens with your datasets + results already loaded—no re-uploading or re-explaining context needed. Ask a follow-up question or leave the field blank & DataVoyager will suggest lines of inquiry.
Find papers is Asta's literature search. Its new default mode, Deep search, interprets your query, evaluates whether the results actually answer it, & keeps searching if they don't.
Context
Ai2's Asta literature search adds a check between retrieving papers and finishing an answer. Its Deep search mode interprets the research question, evaluates whether the results address it, and searches again when they fall short. The July 21, 2026 announcement says the mode also supports follow-up questions and summaries, with longer searches as the tradeoff. The intended improvement is a question-directed search loop that can pursue missing evidence instead of treating the first batch of papers as sufficient.
Checking sign-in…
Loading comments…