Microsoft MVP. Microsoft MCT. CTO @ EID Lodestar spa. #MSDyn365BC & #Azure Solution Architect. #AI Engineer. International speaker & Author. Runner & Cyclist.

Italia
Joined October 2009
Updated all my iPads and my Macs to version 27. Only on iPhone the latest iOS 27 update is not appearing, despite refreshing, network change etc. Still on 26.6.2 and no other updates available. @Apple has some servers and/or load balancing problems… 😕
1
1
508
Finally it’s appeared 🤩
89
Do you remember this? I have it and using a foldable smartphone is not so easy and smart like a traditional one. #iPhoneDuo is amazing but it must be tested carefully, maybe it’s not a good fit for your every day workflow 😅 #AppleEvent
The new Surface Duo. There’s a new way to get things done. Available today: msft.it/6012TtMgr #DoOneBetter
359
If you have a job queue entry in Business Central that talks to an external system (pushing an export, calling a payment gateway, posting to a shipping API) there's a failure mode that's easy to miss until it bites you in production: the AL transaction can roll back after the external call has already succeeded. #msdyn365bc demiliani.com/2026/09/01/why…
1
3
369
In local #AI for Business Central (but not only😜), we have a new leader: 𝗤𝘄𝗲𝗻𝟯.𝟴-𝗙𝗹𝗮𝘀𝗵-𝗡𝗲𝘅𝘁. Qwen3.8-Flash-Next is a new open-weight, 125B parameter MoE multimodal model from @Alibaba_Qwen. Built on the architecture that will underpin future Qwen4, it supports a 262K context window and advanced reasoning. The model outperforms Claude-4.6-Opus (Max) on many benchmarks and can run fully local. For #AL development, it's very good. But where it wins is on agentic tasks (great choice for private #AI projects). P.S. 37.5 tok/sec fully locally is not bad 😉 #msdyn365bc
1
2
300
Quick local #LLM test on my local hardware: 𝗤𝘄𝗲𝗻𝟯.𝟴-𝟮𝟳𝗕 vs 𝗤𝘄𝗲𝗻𝟯-𝗖𝗼𝗱𝗲𝗿-𝗡𝗲𝘅𝘁-𝟴𝗯𝗶𝘁 writing #AL code, no rules, no skills, no tools — just raw model output. 𝗤𝘄𝗲𝗻𝟯.𝟴-𝟮𝟳𝗕 is a dense 27B model, very good on agentic tasks and on writing code. 𝗤𝘄𝗲𝗻𝟯-𝗖𝗼𝗱𝗲𝗿-𝗡𝗲𝘅𝘁 is an 80B MoE model with only 3B active parameters, purpose-built for coding/agentic tasks — which shows: cleaner, more idiomatic AL and nearly 4x the speed (40.4 vs 10.7 tok/s). For AL-specific coding, the coder-tuned architecture of 𝗤𝘄𝗲𝗻𝟯-𝗖𝗼𝗱𝗲𝗿-𝗡𝗲𝘅𝘁 still wins. #msdyn365bc #Qwen #LocalAI #PrivateAI
1
1
258
Everyone is talking about token/sec when testing local #AI models. Important aspect? Yes, but not the main aspect to check. What about accuracy? Why not doing comparison between models on the same prompt? Who cares about token/sec…
1
188
How can I move my Business Central-related Azure resources to a new Azure region? All the italian partners and customers are waiting to have Dynamics 365 Business Central deployed on Italy North datacenters. But what about the Azure and Power Platform resources used in a Business Central project? Should I move them to the new Azure region too or not? And if so, what are the needed steps? demiliani.com/2026/08/10/how… #msdyn365bc #azure #azurefunctions #logicapps #azuresql #powerplatform #AI #serverless @Azure @AzureFunctions @logicappsio
1
165
Visual Studio Code Agent hook for auditing your AL coding sessions. Some week ago, I received a strange request from auditors: can you track every file touched (read/write) by your #AI coding agents? We should have a clear auditing log of every file a coding #agent session touches. How to do that? We cannot use skills or something similar, because these are triggered by the agent itself. But fortunately there's something new in the horizon for achieve this... demiliani.com/2026/07/30/vis… #msdyn365bc #AI #LLM #GithubCopilot #vscode
1
169
Dynamics 365 Business Central: the mistery around the “Parallel Session Management” codeunit. If you've spent time in the BC base app, you've probably discovered that there's a codeunit object (ID = 490) called Parallel Session Management and in very short words it's the object behind the classic on-prem pattern for fanning work out across multiple NST sessions. What the codeunit does? Why it's not exposed on SaaS? demiliani.com/2026/07/13/dyn… #msdyn365bc
3
327
It was February 2024 (time flies🥴) when Microsoft did a 3-day #BCAIHackathon initiative and me with @DmitryKatson and Jeremy launched a native Semantic Search implementation in Business Central. On that prototype we demonstrated that having a backend-native semantic search (so using embeddings at the SQL level) can improve the performances of searches for semantically terms a lot. And now we have it! 🥂 demiliani.com/2026/06/26/dyn… #msdyn365bc #semanticsearch #AI
2
1
9
309
Do you know that every time you start a session, #GitHub #Copilot can build under the hood a searchable history of everything you work on and then you can ask natural language questions about past sessions or get personalized tips around how to optimize your Copilot usage and save costs? demiliani.com/2026/06/23/git… #AI #msdyn365bc @GitHubCopilot
3
352
⚠️Friendly reminder: On 26 August 2026, Microsoft will retire the Azure OpenAI Assistants API. After that date, the Azure OpenAI Assistants API won't work‼️To avoid disruptions, transition to #Foundry #Agent Service by that date. If your workloads are purely model-inferencing and don't require agentic capabilities, you can use the #Azure #OpenAI Responses API instead. Foundry Agent Service provides consistency and OpenAI compatibility, reducing fragmentation, eliminating overlapping preview surfaces, and providing a clear long-term investment path. Classic agents, which are similar to the Assistants API, are not impacted by this retirement and will continue to work.🚨 #AI
1
360