An open model built to be fine-tuned, an Apple app release flow designed for agents and a Google research tool that now analyzes data.
Thinking Machines releases Inkling, an open model designed for customization
Thinking Machines released Inkling, an open-weight model trained from scratch, with 975 billion parameters and 41 billion active, a context of up to 1 million tokens and reasoning over text, images and audio. The company itself says it is not the strongest model available, but a good base for fine-tuning, which can be done on its Tinker platform. A preview of Inkling-Small, lighter and cheaper, came along.
Companies with very specific data and tasks can start from an open model and fine-tune it to their business, instead of relying only on generic models. The first step is to have quality examples of the tasks you want to improve.
How Dexpert helps: Artificial IntelligenceSources
- Thinking Machines Lab: Inkling: Our Open-Weights Model
How to build and ship Mac and iOS apps without opening Xcode
Scott Willsey shows a flow to build and ship Mac and iOS apps entirely from the terminal, which makes working with agents such as Claude Code easier. Xcode must be installed, but tools such as xcodebuild, notarytool and devicectl run from the command line, and XcodeGen generates projects from a config file. After a few one-time steps, such as creating the certificate and storing the notarization password, signing, notarizing and installing become a single script, with no secrets in the repository.
Automated builds and releases reduce manual errors and let agents and CI pipelines handle the whole lifecycle of Apple apps. Apply the same principle to any mobile project.
How Dexpert helps: Mobile AppsSources
NotebookLM becomes Gemini Notebook and now runs code to analyze data
Google renamed NotebookLM to Gemini Notebook, which stays a standalone product but is integrated into the Gemini app and, soon, AI Mode in Search. Each notebook gets a secure cloud computer to write and run code, enabling data analysis grounded in the user's sources, first for Ultra subscribers and Workspace customers and later for all Pro users. According to Google, more than 30 million people and 600,000 organizations use the tool.
Business teams can run deeper analyses without always depending on the data team. For sensitive data or data tied to internal systems, a custom solution may be the next step.
How Dexpert helps: Digital TransformationSources
Want to put one of these to work in your company?
Dexpert builds custom systems, apps, integrations and AI solutions. Tell us what you need and get an estimate, or take the free digital maturity assessment.