Move 37

Chris Caplice: “Reinforcement Learning (RL), can serve as a discovery engine. A great example of this is the 2016 five-game match of the strategy game Go between Google DeepMind’s AlphaGo and Lee Sedol, an 18-time title winner. The 37th move in the 2nd game made by AlphaGo was so unique and unusual that commentators initially thought it was a mistake. AlphaGo ended up winning the game and the entire match 4 games to 1. Thus, the story of “Move 37” was born – an insight or action that is unprecedented and not based on historical moves.

The question I ask in the article is, where is the Move 37 in freight transportation? How can we apply RL and other AI techniques to generate new ideas, not just faster implementations of existing methods?Link

Posted in Artificial Intelligence, Innovation | Leave a comment

Will LLMs ever solve optimization problems on their own?

Adam DeJans Jr. answers: “Eventually, yes.” “However, I suspect optimization solvers themselves will remain specialized numerical engines for a very long time.

Continue reading
Posted in Artificial Intelligence, Optimization | Leave a comment

Practical Applications of Constraint Programming

Mark Wallace shares a retrospective look at practical applications of constraint programming (CP), starting from the 1990s, when CP gained the most commercial significance. “Constraint Programming has undoubtedly been successful. It has become a standard tool for tackling complex government and industrial applications. Constraints have made a variety of applications, from scheduling to sustainability, more accessible and thereby allowed more applications to be tackled and solved. In practice, CP is typically used as a component of larger systems, and even CP models may be mapped down to hybrid algorithms including MIP, SAT and heuristic methods.

The current obsession with LLMs has taken the focus away from formal modelling and algorithmic problem solving. However, amongst the computer-literate it is recognised that LLMs provide a user interface, but the underlying techniques of modelling and solving, using CP and other technologies, are essential to their practical benefit. In short CP will continue to be a key problem solving tool for years to come!Link

Posted in Artificial Intelligence, Constraint Programming, Optimization | Leave a comment

Google Open Knowledge Format (OKF)

Google has introduced Open Knowledge Format (OKF), a new open specification designed to standardize the way content is packaged and shared with AI agents. Google Cloud Tech’s X post describes it as a “vendor-neutral” standard, built to be readable by both humans and machines without requiring new tools or software to implement. OKF formalizes the “LLM-wiki pattern” into a portable, interoperable format that provides a standardized way to represent enterprise knowledge for modern AI systems. Read more

Posted in Agents, Artificial Intelligence, Standards | Leave a comment

Claude Code Revelation

Link

Posted in Algortithms, LLM | Leave a comment

Are CVs dead?

If they are not dead, they are seriously weakened. CVs are becoming a formality rather than the primary signal of talent. Skills, portfolios, and scenario-based assessments are taking over — especially in technical and knowledge-work fields like decision intelligence.

Nobody cares what your CV says anymore; what matters is what you leave on the internet. Companies are scanning the digital fortresses, such as GitHub, where candidates actively build. Saying I can do this is over. Candidates who say, I did this, here’s the link, here’s the code, here are the results, are blowing past those with 500+. HR teams at elite companies are using custom algorithms to scan real projects and community contributions on these platforms.” Link

Posted in Trends | Leave a comment

Why Rules Still Matter in an AI World


Pierre Berlandier
published the results of “a candid, side-by-side experimental comparison” of two approaches to decision management: rule-based and LLM-based. “We used a simple decision scenario as a test case: determining whether a traveler qualifies for access to the fictitious Big Blue airport lounges. We defined a set of simplified lounge access rules, implemented them both as formal business rules and as native, natural language instructions within an agentic platform, and evaluated their behavior.” Read the results here. A complete solution with IBM ODM is included.

Posted in Decision Intelligence, LLM, Rule Engines and BRMS | Leave a comment

Making Better Decisions in Supply-Chain Planning

Watch the interview with Jeff Metersky, senior vice president of strategy and innovation at GAINS, who cuts through the hyperbole and talks realistically about the benefits of agentic AI and improved supply chain planning. Agentic artificial intelligence may be top of mind for many in supply chain these days, but the notion that the technology is going to drive an autonomous supply chain is overrated, Metersky says. Link

“I don’t think that that’s the right destination for everything,” he says. “I don’t think that that’s where we want to land at the end of the day.” Certainly, there are decisions that can be automated, but others will always require human intervention and advice.  “AI is a tool in our tool bag, but it’s not the only tool.”

Continue reading
Posted in Agents, Artificial Intelligence, Supply Chain | Leave a comment

DecisionCAMP-2026 Program is now Live!

This year, we received an overwhelming number of submissions — so many, in fact, that we had to add an extra day to the event! Join us August 25–28 for four days packed with the latest in Decision Intelligence, Business Rules, Optimization, and AI-driven decision-making, brought to you by some of the brightest minds in the field.

Check out the full program here: https://decisioncamp2026.wordpress.com/program/

We’re thrilled to announce Philippe Kahn as our keynote speaker, along with an incredible lineup of presenters who will share real-world insights and cutting-edge approaches to decision automation. Whether you build decision services or apply them to drive better business outcomes, DecisionCAMP-2026 is the place to be. See you there! http://decisioncamp.org/

Posted in Decision Intelligence, DecisionCAMP, Events | Leave a comment

How to govern work shared between humans and AI

AI agents are entering operational workflows. Trisotech delivered a webinar about their experience of governing work shared between humans and AI agents. They provide practical BPMN patterns for task assignment, human review, supervision, and handoffs between human and AI performers. Link

Posted in Agents, Artificial Intelligence, BPM, Business Processes | Leave a comment