MOOSE AI

Battery analysis, in plain language.

MOOSE AI is the built in analysis assistant in the Micantis platform. It understands your battery test data, specifications, and workflows. Type a question, get a structured answer with charts, key findings, and specific data points.

Try MOOSE Live Schedule a demo
MOOSE AI

The plain language assistant for your battery data.

Type a question, get a structured answer with charts, key findings, and references. MOOSE knows your data, your specs, and the workflows your team runs.

  • Ask in plain language. Charts and findings come back inline.
  • Schedule recurring reports delivered to your inbox.
  • Clean data in domain terms ("remove the DCIR pulses").
  • Pin and share findings with colleagues, no login required.
  • Search your reference docs with citations to the page.
Try MOOSE Live
MOOSE AI × Python

Your code, MOOSE's reach. Running in our cloud.

Write, run, and schedule Python scripts directly in Micantis. Scripts execute in secure cloud containers with full data access. MOOSE is built into the editor to write and debug.

  • MOOSE writes the code when you describe what you want to analyse.
  • Cloud execution. No local Python or environment setup.
  • Save and share scripts as templates across your team.
  • Schedule recurring Python jobs with email alerts on failure.
  • Ask MOOSE to debug scripts that hit errors.
View on PyPI

Use Cases

MOOSE AI in real workflows.

Automated Data Cleaning

Using AI for fixing bad cycles.

Original Original cycle data with a bad first cycle that drops to roughly 3,000 mAh
Cleaned Cleaned cycle data after MOOSE removes the bad first cycle

Real cycler files come with bad first cycles, restarts, and outliers that wreck downstream analysis. Describe the issue and MOOSE produces a cleaned dataset, side by side with the original — raw data is never modified, every cleaning step is logged.

  • Domain language — "remove the bad first cycle"
  • Side by side original vs. cleaned for verification
  • Non destructive — raw files stay untouched
  • Audit trail for every cleaning operation

Open Architecture

MOOSE is one way to use these tools. Build your own with MCP.

MOOSE is the assistant we built so you don't have to. Under the hood, it calls a set of named tools that are also exposed as Model Context Protocol (MCP) servers. Same tools, available to any agent harness — Anthropic's Claude Managed Agents, AWS Bedrock, Google Vertex AI, or your own DIY runtime.

  • micantis.data_query — live lab data
  • micantis.spec_library — your acceptance criteria
  • micantis.method_library — canonical analyses (capacity fade, EIS fitting, dQ/dV peak tracking)
  • micantis.test_plans — protocol generation
  • micantis.report_templates — your output formats

Bring your own LLM (Claude, Gemini, GPT, or a local model on your hardware). Bring your own playbook with your team's conventions. The tools stay the same.

Co-Scientist Is a Pattern Deploying the Pattern
Open architecture: engineer + LLM agent + playbook on top, Micantis tools (data query, spec library, method library, test plans, report templates) in the middle reached by MCP, live data substrate underneath

See MOOSE AI working with your data.

30 minute walkthrough. Ask any question about your battery test data.

Try MOOSE Live Schedule a demo