
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
EF Core 11 preview 2 is adding practical query features — The host opened with new support for MinBy/MaxBy, SQL Server full-text improvements, experimental vector functions, and SQL Server 2025 JSON_CONTAINS, which can use JSON indexes when compatibility level is set to 170.
Devart is pitching dotConnect as the “don’t write the plumbing” layer for .NET data access — Yuri described dotConnect as an ADO.NET provider family for databases and cloud apps like Oracle, MySQL, Salesforce, Dynamics 365, QuickBooks Online, and Zoho CRM, with a “Universal” license for teams juggling multiple backends.
Entity Developer’s main sell is visual ORM modeling that writes the boilerplate for you — In the demo, Yuri generated an EF Core model from a MySQL sample database, edited properties and associations in a designer, and showed those changes immediately reflected in generated C# classes.
The workflow is explicitly database-first or model-first, then sync in either direction — The tool lets you start from an existing schema or an empty model, then update the model from the database, update the database from the model, or generate SQL scripts for review before execution.
Devart is leaning hard on enterprise-friendly positioning: compatibility, support, and longevity — Yuri emphasized support for .NET Framework including 4.8.1, .NET Core, modern .NET, ASP.NET, Blazor, and MAUI, while framing paid licensing around security audits, performance testing, and responsive technical support.
AI isn’t in dotConnect yet, but Devart says something is coming in Q3 — Asked about vector search and AI, Yuri said current work is more around an AI assistant for Devart’s database studios, with a teaser that a new AI-related product announcement is likely in the third quarter.
The host kicked things off with a brisk rundown of Entity Framework Core 11 preview 2: MinBy and MaxBy now translate through LINQ, SQL Server full-text search got new functions plus table-valued function support, and SQL Server 2025’s JSON_CONTAINS is now used for better JSON querying when compatibility level is 170. He also flagged ongoing work toward a unified API for vector search and vector indexes, rather than every provider inventing its own one-off shape.
Before the main topic, the host plugged Microsoft Build in San Francisco on June 2–3, with names like Satya Nadella and Scott Hanselman. Then he brought in Yuri from Devart, who introduced himself as part of the company’s sales team and recalled meeting the host in Poland at a .NET conference — a small human moment that made the whole thing feel more like community than product pitch.
The host admitted he mostly knew Devart from the old Entity Framework 6 era, and Yuri filled in the bigger picture: Devart has been around since 1997 and now runs two major product lines — database tools and data connectivity. He listed ADO.NET providers, ODBC drivers, Python connectors, SSIS components, Excel add-ins, and even Delphi support, which prompted a fun detour into nostalgia about Pascal, Delphi, Visual Basic, and debugger quality 15-plus years ago.
Yuri framed dotConnect as a family of advanced ADO.NET providers for secure, high-performance access to major databases and cloud services, from Oracle and PostgreSQL to Salesforce and QuickBooks Online. His core pitch was simple: instead of writing and maintaining all the connection code yourself, you use Devart’s connectors, get direct connectivity, API and SQL support, Visual Studio integration, and a support team behind it.
A big chunk of the presentation was about breadth: .NET Framework, .NET Core, modern .NET, ASP.NET, Blazor, and MAUI are all supported, along with Entity Framework, EF Core, Dapper, NHibernate, and Devart’s own LinqConnect. Yuri kept coming back to three words — security, performance, support — positioning dotConnect as something built for enterprise teams that need audits, stability, and someone to call when production gets weird.
In Visual Studio, Yuri added a Devart EF Core model, chose database-first, connected to a MySQL sample database, picked a subset of film-related tables, and generated classes using the standard EF Core template. The host asked whether templates could be customized for company-specific rules, and Yuri said yes, with help from Devart’s technical team.
Once the model was generated, Yuri added a new country property, changed a type, created an association manually, and showed how saving the model immediately updated the generated classes. Then he walked through the reverse path too: update the database from the model or generate a SQL script first, so teams can review changes in a tool like dbForge Studio for MySQL before execution.
The chat hit the practical stuff: yes, .NET Framework 4.8.1 is supported; yes, mobile scenarios are possible through MAUI; no, dotConnect doesn’t have a community edition beyond a 30-day trial. Yuri also confirmed Devart is originally from Ukraine, now spread across places like the Czech Republic, Slovakia, Spain, and Ukraine, and teased that while dotConnect itself doesn’t yet support AI-specific features, Devart has an AI-related launch coming in Q3.
Share
Keep Reading
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.

Playbook
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.