MS OptiMind: From Problem Description to Solution

Posted on January 25, 2026 by decisionmanagementcommunity

Microsoft Research has just released OptiMind, “a small language model designed to convert business problems described in natural language into the mathematical formulations needed by optimization software. Built on a 20-billion parameter model, OptiMind is compact by today’s standards yet matches the performance of larger, more complex systems. Its modest size means it can run locally on users’ devices, enabling fast iteration while keeping sensitive business data on users’ devices rather than transmitting it to external servers.“

Continue reading →
Posted in Artificial Intelligence, Decision Optimization, solvers | 1 Comment

Gartner To Publish New Magic Quadrant for Decision Intelligence Platforms Next Week

Posted on January 24, 2026 by decisionmanagementcommunity

David Pidsley just announced: “Last summer, I shared that Gartner Says AI Ready For Decision Intelligence Market. Today, I’m pleased to announce that we have a new Magic Quadrant scheduled (3 December 2025), which will be the evolution of our Market Guide for Decision Intelligence Platforms (client login required). This will supersede the Gartner Market Guide of 18 July 2024, which will not be refreshed.” Link

Published on Jan 26, 2026

Posted in Decision Intelligence, Vendors | Leave a comment

Challenge Feb-2026 “Trivial Rule”

Posted on January 24, 2026 by decisionmanagementcommunity

This challenge was inspired by a post from Ron Itelman. You need to use any rules-based platform or GenAI tool to create an AI Agent that implements the following rule:

“Execute a buy order if the current price is lower than yesterday’s price.“

Link

Posted in Challenges, Decision Modeling | Leave a comment

Decidability Problem

Posted on January 23, 2026 by decisionmanagementcommunity

“We often assume that if we have good data and a good model, a decision can be made. That assumption quietly fails more often than we think,” – wrote Ron Itelman. “By decidability, I mean a boundary condition: A decision is decidable if the minimum semantic prerequisites required by the decision rule are grounded.”

“Take something trivial: a trading rule that says ‘Buy if price today is less than price yesterday.’ The logic is simple. The model works. The data exists. But the system can still halt. Why?

When a stock-market trader at a London Stock Exchange desk asks their AI agent, “What were my trades yesterday?” The system resolves ambiguity before it can answer. Which time zone? “Yesterday” from whose perspective: the headquarters in New York, U.S.A. the trader in London, UK or their client at Shenton Way, Singapore? The AI silently decides, then returns a confident answer without exposing those variables to the user.” Link

Posted in Decision Intelligence, Decision Modeling | Leave a comment

“Show me why this loan was approved”

Posted on January 20, 2026 by decisionmanagementcommunity

Link

Posted in Decision Modeling, Explanations | Leave a comment

Decision Timing Mismatch

Posted on January 19, 2026 by decisionmanagementcommunity

Adan DeJans Jr. shared another experience from the trenches of real-world decision-making. “The model assumes decisions are made at a clean point in time, with clean information. But reality doesn’t work that way. Data arrives late, decisions are staggered, overrides happen mid-cycle, and yesterday’s “final” plan quietly becomes today’s suggestion. If your optimization assumes a single moment of truth, it will always feel brittle in production, no matter how optimal the solution looks on paper.” Link More

Posted in Decision Making, Optimization, Scheduling and Resource Allocation, Supply Chain, Uncertainty | Leave a comment

“Hidden Gem” of Decision Optimization

Posted on January 17, 2026 by decisionmanagementcommunity

Jacob Feldman wrote a post about the importance of the Solution Pools called “hidden gems” of optimization solvers. However, they usually require programming expertise, while at decision time, it’s subject matter experts, not programmers, who use already tested and deployed decision models. These business users need to add last-minute constraints on the fly, particularly when there’s no time or technical expertise available to modify the underlying decision model. Jacob asks the question: “How do we unlock this hidden gem and put it in the hands of decision-makers when they need it most?“

“Today, there’s a growing consensus among optimization experts: constraint and linear/MIP solvers deliver greater real-world value when embedded in modern Decision Intelligence Platforms that seamlessly integrate rule engines, machine learning, and optimization. The logical evolution for Decision Intelligence Platforms is to provide graphical Decision Pools—making Solution Pools accessible to business users. The newest OpenRules Decision Playground brings this vision to life.” Link

Posted in Decision Optimization, Human-Machine Interaction, Products | Leave a comment

The ability to ask the right questions is the key to successful decision modeling

Posted on January 16, 2026 by decisionmanagementcommunity

Prof. Warren Powell wrote: “What all of us do, and I think it is without exception, is look at problems through the lens of the modeling frameworks that we have been trained in. We are prototypical hammers looking for nails.

The solution: We need to educate people in how to ask the right questions. These people should *not* be trained in any analytical methodology to avoid the bias that this unavoidably introduces. Instead, they need to learn how to ask the right questions, without any bias toward a solution approach.” Link

Posted in Decision Modeling | Leave a comment

Moving away from “vibe decisioning”

Posted on January 11, 2026 by decisionmanagementcommunity

David Pidsley, a decision intelligence leader at Gartner, posted today this warning: “GenAI tools instantiate flaws across business decision networks with frightening efficiency when requirements are ambiguous. They will cause growing concerns about uneven decision quality, decision debt (inferred decision traces without explicitly decision models), lucky but fragile decision outcomes disguising AI sycophancy as decision logic, and jeopardizing economic viability by adding unspecified additional contextual data to enterprise decisions.

Enterprises will be moving away from experimental “vibe decisioning” toward decision architecture-first platforms with governance and quality controls. That’s another reason for the demand for decision intelligence platforms.” Link

Posted in Architecture, Artificial Intelligence, Decision Intelligence, Trends | Leave a comment

More about Decision Reasoning Traces

Posted on January 10, 2026 by decisionmanagementcommunity

Tony Seale: “Real decisions are never made in a single system. They are made by stitching together signals from CRM, finance, operations, support systems, policy documents, Slack threads – often with human judgement applied at the seams. The most valuable data for enterprise AI is not just what happened, but how and why a decision was reached. This mirrors exactly what foundation model companies discovered when they started building reasoning models. Performance didn’t improve just by scaling data – it improved when they began collecting reasoning traces.” Link

Posted in Decision Intelligence, Decision Making, Decision Tracing, Reasoning | Leave a comment