ZARA · ACTIVE
00:00:00 UTC

Automated Science

OpenScientist

The frontier of science, automated.

Join the waitlist Meet Zara ↓
01Autonomous hypotheses
02Proprietary tool access
03Self-authored briefs
04Compounding research

The best automated AI scientist. Discovering at the frontier.

Real hypotheses. Real experiments. Real research.

Introducing Zara

Research that moves on its own.

Zara plans, collaborates, runs scientific tools, evaluates results, and improves through reinforcement learning.

Planning the research
Objective
Identify an antibody candidate with stronger target binding and acceptable developability.
  1. 1Review target constraints
  2. 2Generate candidate sequences
  3. 3Evaluate binding and stability
Searching recent antibody design literature
PubMed bioRxiv Protein Data Bank
Affinity maturation of therapeutic antibodies Nature Biotech · retained
Developability prediction from sequence bioRxiv · retained
Running structural docking
Docking Agent
Connecting to structural docking server
Connection established
12 candidate poses submitted
Sequence Agent
produced 24 candidates
Evaluation Agent
requested developability checks
Shared workspace updated
24 candidates evaluated · 5 advanced
CandidateBindingStabilityDevelop.
ZR-17 Selected
ZR-09 Advanced
ZR-22 Advanced
ZR-31 Rejected
Candidate ZR-17 selected
structural docking
Reinforcement learning environment
Antibody Candidate ZR-17
Selected after literature review, sequence generation, structural docking, and multi-objective evaluation.
Candidate sequence Docking results Evaluation summary References Research brief
Policy confidence
Research brief complete Confidence high
Rewarding stronger scientific decisions
Decision Action Reward
  • Valid hypothesis+1.2
  • Tool selection+0.8
  • Failed docking path−0.4
  • Improved candidate ranking+1.6
Evidence Agent
Stronger binding affinity confirmed[1]
Developability within accepted range[2]
Low aggregation risk predicted[3]
Claims linked to sources · 8 citations verified
references.bib created

Capabilities

How Zara works

01

Tool-connected

Servers & lab APIs

02

Hypothesis-driven

Parallel research threads

03

End-to-end

Idea to written brief

04

Compounding

Each result sharpens the next

Case study

Self-authored antibody generative discovery.

Zara scoped, built, and evaluated a de novo antibody generative model. Then she wrote the brief herself.

De novo sequencesAutonomous pipelineZero human drafting
research_brief_antibody_gdm.md, Zara
Autonomous Antibody Generative Discovery
Research Brief · OpenScientist

Get access

Join the waitlist for Zara

Be first in line.

You're in

We'll reach out as we open access.