The short answer is no!  It is and will be too slow for a long time!  But slow does not necessarily mean useless. We must set realistic expectations and create use cases where it is OK to be slow.  I work a lot with performance enhancements and tuning Dynamics 365.  I understand the underlying platform and architecture, how data is stored and fetched from Azure SQL and computed on.  I see the latency and most importantly see the effect of tuning algorithms. 

For “close-to-real time” scenario’s, AI/Copilot does not even come close to what we see in algorithmic performance.  Let’s say you have an eCommerce site, or a POS.  The user selects the products, and we need to present a price within a few milliseconds.  Algorithms can do that.  AI cannot.  But AI solutions are excellent to build and adjust the business data and rules used for a pricing engine, when done in the background by AI agents.

This means that at the AI can set up and feed Dynamics 365 with the right data/setup to fulfill defined goals and scenarios. In the future I expect we can have AI Agents connected through MCP that monitor current sales data, cost changes, competitor pricing,  and availability.  We can have AI Agents that adjust and review current pricing and come up with recommendations on what to change.  Price adjustments are then a logical business decision based on actual data to optimize revenues.

Today I see that the algorithmic performance is very fast, but the human reaction time/processes to adjust pricing and related data is slow.  In the future pricing will be based on defined goals given to your Dynamics 365 agent.  This agent will then perform analysis, run simulations, ensure approvals, monitor effect and adjust with optimalizations.  But also communicate the price change effects to those responsible.  The agents will enter the data into the forms and tables, making it ready to be used fast algorithmic applications.

We then get the best of two worlds;  Fast algorithmic real time performance, mixed with the slow asynchronous AI that analyze and to all the heavy lifting to find better pricing and ensure better profitability.

So don’t fire your traditional developers yet.  They are still needed to create lightning-fast real-time algorithms.   

(This article was written without the use of any slow AI’s)