Overview
Platform Overview
Wawe AI is a single research environment layered over 190+ computational biology and chemistry tools. You ask questions in plain English; an autonomous agent plans a multi-step workflow, selects the right tools, runs them, self-corrects on errors, and returns a synthesis-quality answer with a full trace of what it did.
How a query flows
Every workspace is conversational and backed by the same orchestration core. A single research question moves through four stages:
- Plan — the agent decomposes your request into a sequence of tool calls.
- Act — it invokes tools (structure prediction, docking, database lookups) with the arguments it inferred.
- Reflect — results are checked; on failure the agent retries or picks a different tool.
- Synthesize — findings are assembled into a grounded answer you can export.
Answers are evidence-grounded. Every claim traces back to a tool call, a database record, or a live computation — visible in the run trace, never invented.
The four Spaces
Tools are grouped into purpose-built workspaces that map onto the drug discovery pipeline. Each Space feeds the next — a target found in Biology becomes a design brief in Chemistry, a candidate validated at the Work Station, and a run tracked in the Grid.
Biology Space
Genomics, targets & protein structure
Chemistry Space
Molecular design, ADMET & synthesis
Work Station
Docking, MD & binding validation
Wawe Grid
Lab operations & experiment tracking
What the agent can reach
The tool catalog spans 25 categories. A non-exhaustive sample of what the agent orchestrates on your behalf:
- Structure prediction — ESMFold and AlphaFold for protein folding from sequence.
- Docking & dynamics — AutoDock Vina, DiffDock, and GROMACS for binding and MD simulation.
- Cheminformatics — SMILES parsing, ADMET prediction, and retrosynthesis planning.
- Knowledge bases — UniProt, RCSB PDB, NCBI, ChEMBL, PubChem, and OpenTargets.
Sessions, memory & traces
Every conversation streams results as they compute and is saved automatically. Session memory lets the agent carry context across turns — refer to "that compound" or "the target from earlier" and it knows what you mean. Each run keeps a tool trace you can open to see exactly which tools ran, with what inputs, and in what order.
Access & deployment
Wawe runs on enterprise Azure infrastructure with authentication and role-based access for research institutions. Access is granted per-request during the pilot — see Getting Started to register.