Teams

Direct messages

From Knowledge to Discovery

Mission Control

Supervise agent runs, compute jobs and approvals. Execution status reports what the runtime did; scientific validation is tracked separately and is never inferred from a run finishing.

Idle

Compute

Compute jobs flowing through the simulation gateway — atomistic, electronic-structure, reactor, and continuum solvers — with their queue state, resources, logs, and outputs.

Idle

Select a job

Pick a job on the left to inspect its status timeline, resource request, stdout, parsed result, and artifacts.

Duration —

Started —

Finished —

Exit code —

CPU / GPU —

Memory —

Walltime cap —

Artifacts —

Auto-scroll —

Audit

Who did what, when, and with what result, read from the platform's append-only audit ledger.

Integrity: loading…

Time	Who	Action	Target	Result

Times are shown in your local time zone; hover a time for exact UTC. Viewing and exporting this log is not itself recorded. This console has not been validated for GxP or 21 CFR Part 11 use.

Loading graph…

Drag nodes to rearrange · scroll to zoom · double-click a node to expand its neighbourhood

Answer

Labels, types & props

Node labels

Relationship types

—

Key	Value

Neighbours 0

Design a new schema

Decomposer → Architect → Domain Experts (parallel) → Critic → Refiner. Persists as a GALAXY Schema node when you save.

Domain Slash-separated convention supported. Description (optional)

Depth Max refinement iterations Target critic score

Save to scope

Multi-Agent Schema Designer

Provide a domain and (optionally) context, then run the designer. Each stage's progress streams here in real time. The final schema can be saved directly into your GALAXY scope.

Decomposer — entity ideas
Architect — sketches + relationships
Domain Experts — per-entity specs
Deterministic validation
Critic — score + issues
Refiner — replacements / additions

HPC clusters and other compute back ends. Submit DFT, MD, and simulation jobs here.

Loading…

Appearance

Light or dark theme. Defaults to your OS preference.

Voice

How voice mode reads replies aloud. Saved to your account.

Tenant

—

Mode

—

License: —

Period	Calls	Input tokens	Output tokens	Cache-read tokens	Cost

Research credits: —
Spent (all time): —
Balance: —

Usage is metered per model call and shown at estimated provider cost. Credits are informational during the research phase — nothing is blocked.

brew install ollama && ollama serve &
ollama pull llama3.2:3b   # ~2GB, runs on any 8GB+ Mac
# in .env:
ALCHEMIC_PROVIDER_MODE=live
ALCHEMIC_LOCAL_PROVIDER=ollama
ALCHEMIC_LOCAL_MODEL_URL=http://127.0.0.1:11434
ALCHEMIC_LOCAL_MODEL_NAME=llama3.2:3b

# key from console.anthropic.com
ALCHEMIC_EXTERNAL_PROVIDER=anthropic
ALCHEMIC_EXTERNAL_API_KEY=sk-ant-...
ALCHEMIC_EXTERNAL_MODEL_NAME=claude-opus-4-7
ALCHEMIC_ALLOW_EGRESS=true

From Knowledge to Discovery

Chronicle

Tools

Mission Control

Compute

Detached jobs

Select a job

No recurring schedules

Audit

Channels

SCHEMA Multi-Agent Designer

Multi-Agent Schema Designer

Final schema

From Knowledge to Discovery

Detached jobs

Select a job

No recurring schedules

Channels

Multi-Agent Schema Designer

Name

Vector store .tar.gz or .json

Welcome

Change password

All users

Grant research credits

Your keys

Local — Ollama

External — Anthropic Claude