`MLJAR Studio`: local AI data analyst that turns chats into notebooks

Natural-language data analysis now ends as a runnable .ipynb, not a dead-end chat. Local execution, auto environment setup, and built-in AutoML make it a practical bridge between Jupyter and cloud copilots, though $199 one-time limits impulse adoption.

[ KEY POINTS ]

Chat-driven analysis is saved as a reproducible .ipynb; generated Python stays inspectable, editable, and rerunnable instead of disappearing in chat history.
The app auto-creates a local Python environment and installs missing packages during the session, cutting setup friction on Mac, Windows, and Linux.
Built-in AutoML covers tabular classification, regression, and multiclass work, so exploratory analysis and baseline modeling sit in one desktop workflow.
Data ingress is broad: CSV, Excel, Stata, Parquet plus PostgreSQL, MySQL, SQL Server, Snowflake, Databricks, and Supabase.
Model choice is flexible with local Ollama, user-supplied OpenAI API keys, or the vendor add-on; zero-egress local mode is the clearest differentiator.

Originalmljar.com/Read original →

// related

#0001
#0001Agents & tools Google AI Forum1 hour ago
`Antigravity 2.0` Install/Update Bug Workaround: Roll Back to `1.23.2`
40radar
AntigravityAI coding agent IDE — integrated with Google tooling
Rollback first, then disable background updates. Until the installer stabilizes, manual updates are the practical path.
- Download 1.23.2 from previous releases instead of installing 2.0; this is a rollback workaround, not a permanent fix.
- After opening Antigravity, go to Editor Settings > Application > Update and disable Windows Background Updates.
- Switch update mode to Manual Mode and restart. This avoids being pushed back into the broken update path.
Source: discuss.ai.google.dev/t/solution-for-antigravity-2-0-insRead original →
FIG-0011:1
40radar
FIG-0011:1
#0002
#0002Agents & tools Google AI Forum2 hours ago
`Antigravity` agent sessions hit template mismatch blocker
40radar
AntigravityAI coding agent IDE — built around Google's stack
Prompt generation fails before a new agent session starts. Treat it as a blocker, not a workflow tweak.
- The failure happens in PlannerGenerator / CascadeManager, before the agent session reaches useful work.
- communication_style calls CascadeConfig.GetSupportsLatexRendering, but the current protobuf type lacks that field.
- Impact is complete session creation failure. If your workflow depends on Antigravity, keep a fallback agent ready.
Source: discuss.ai.google.dev/t/agent-executor-error-communicatiRead original →
FIG-0021:1
40radar
FIG-0021:1
#0003
#0003Agents & tools Google AI Forum2 hours ago
`Antigravity` paid user reports forced update broke workspace
50radar
AntigravityAI coding IDE — Google's agent-based dev environment
A forced IDE update allegedly scattered projects, split threads, and wiped settings. Treat agent IDEs as volatile infra: pin versions and keep rollback paths ready.
- Ultra costs $200/month, yet the reported update shipped with no email, warning, migration path, or visible rollback.
- Breakage hit core state: projects, conversations, settings, and config. Back up .vscode-style config and agent history separately.
- Users are reportedly reverting to 1.23.2 and disabling auto-update. Version pinning becomes part of production workflow.
Source: discuss.ai.google.dev/t/200-month-subscriber-and-this-upRead original →
FIG-0031:1
50radar
FIG-0031:1

`MLJAR Studio`: local AI data analyst that turns chats into notebooks

// related

`Antigravity 2.0` Install/Update Bug Workaround: Roll Back to `1.23.2`

`Antigravity` agent sessions hit template mismatch blocker

`Antigravity` paid user reports forced update broke workspace