Visual data science and cognitive cartography
Visual data science / cognitive cartography
Mantis is a visual data science and AI platform for building, exploring, and collaborating on complex data workflows.
Product highlights
EXPLOREVisual-first analysis
Explore relationships, inspect evidence, and follow context across a map.
Explore the platform
SHARECollaborative workflows
Share discoveries through journeys, vignettes, and reusable workflows.
Explore the platformA new canvas
for curiosity.
SpacesMap and inspect complex data.ComposerAgents grounded in map context.MultimodalText, PDFs, tables, media, and scientific data.WorkflowsBuild reusable, node-based data workflows.Models & servicesChoose OpenAI or private vLLM services per space.Developer toolsMCP, CLI, extensions, notebooks, and APIs.
Semantic spaces.
Explore mappingTurn raw documents, tables, media, repositories, and scientific data into visual spaces where structure and relationships become visible.
Multi-source data ingestionAutomated embedding and clusteringInteractive maps and categoriesBags, trees, and floorsLasso and path exploration
Composer and agents.
Meet the agentsUse Composer and MantisGPT with selections, clusters, documents, and map context kept visible throughout the analysis.
Composer and MantisGPTNarration and summarizationEvidence-linked responsesReusable agent toolsMap-grounded context
Multimodal spaces.
Explore multimodal spacesCreate unified spaces from text, PDFs, tables, images, audio, video, coordinates, molecules, and biological data.
Images, audio, and videoCSV and XLSXH5AD and molecular dataGitHub and Hugging FaceAcademic and web search
Visual workflows.
Build a workflowCompose node-based data workflows, connect reusable steps, schedule runs, and carry mapped evidence into repeatable analysis.
Visual workflow builderCustom Python stepsJourneys and waypointsAgent orchestrationScheduled and reusable runs
Models and services.
See model optionsChoose OpenAI or private, server-hosted vLLM services for chat, embeddings, and instruction-following work in each space.
OpenAI chat and embeddingsPrivate vLLM embeddingsPrivate vLLM instruct servicesChat, embedding, and instruct choicesModel selection per space
Developer tools.
Open developer docsExtend Mantis through MCP, the command-line interface, custom panels, notebooks, and the developer API.
Mantis MCP toolsCLI for Codex, Claude Code, and OpenCodeCustom panels and extensionsInteractive notebooksDeveloper API
Talks, demos, and field notes
Mantis in the field.
Watch the original cognitive-cartography talks, platform introductions, and research keynotes.
View the full playlist
The future of data science is visual.
Mantis combines visual exploration, AI-assisted analysis, and reusable workflows in one collaborative platform.
Why MantisVisual-first exploration.
Move from raw data to navigable maps where concepts, neighborhoods, and relationships remain inspectable.
Automated synthesis.
Build spaces through repeatable ingestion, embedding, clustering, and labeling steps.
Agents in context.
Ground Composer and agent work in selections, clusters, bags, documents, and the state of the map.
Shareable journeys.
Capture and share discoveries through journeys, vignettes, notebooks, and reusable workflows.
Choose OpenAI or private vLLM services for chat, embeddings, and instruction-following work in each Mantis space.
Managed model APIs.
Use OpenAI chat and embedding services for a fast, fully managed path from upload to exploration.
Private vLLM services.
Connect server-hosted vLLM embedding and instruct services when model and data control matter.
Per-space model selection.
Choose chat, embedding, and instruction-following services for the needs of each Mantis space.
See your own data differently.Request Mantis demo access