dbsliceAI

dbsliceAI

AI-assisted analysis for simulation databases.

dbsliceAI connects AI assistants to a bespoke set of tools for querying, analysing, plotting, and rendering simulation results through the Model Context Protocol.

A short edited demo showing a chat session moving from question, to tool calls, to plots, rendered media, and follow-up analysis.

Watch the longer demo

Try the hosted MCP demo

Connect an MCP-capable assistant to a hosted, read-only sample simulation database. The public demo is for exploring the workflow and current tool surface; it does not connect to your private data.

Connect with ChatGPT Connect with Claude

Tools-focused

The demo exposes tool outputs for comparing cases, fitting models, finding trade-offs, rendering geometry, and exploring response surfaces.

Matched-pairs histogram generated by dbsliceAI.

Matched pairs

Observed versus predicted regression scatter plot generated by dbsliceAI.

Regression scatter

Regression feature map showing the compound lean effect on the Yp downstream extract.

Regression feature map

Pareto scatter plot generated by dbsliceAI.

Pareto scatter

Rendered 3D stator surface and exit plane returned by the dbsliceAI render tool.

3D surfaces render

Yp downstream loss image extract rendered from a sample dbsliceAI item.

2D image render

Line plot comparing stator exit loss profiles for three simulation cases.

Line plots

Response surface contour plot generated by dbsliceAI.

Response surface

Report generation with curated references

dbsliceAI can combine tool outputs with curated paper summaries, letting an assistant draft analysis reports that cite both simulation evidence and selected background literature.

ChatGPT session showing dbsliceAI rendered media returned inside an assistant response.
Report generation can combine generated plots, rendered media, dataset context, and curated references.

Interested in using dbsliceAI?

dbsliceAI is developed by Graham Pullan at the Whittle Laboratory, University of Cambridge. If you want to explore dbsliceAI with your own simulation database, get in touch to discuss data preparation, deployment, and collaboration.

ai@dbslice.org

Publications

Harnessing AI for Scalable Analysis of Simulation Data. Presented at ASME Turbo Expo 2026, Milan, Italy, 15-19 June 2026.

Accepted manuscript record in the Cambridge repository. ASME version-of-record DOI forthcoming.