Webinar Recap: From Decision Intelligence Concept to Actionable Roadmap

Summary

In our Future.Now webinar, “From Decision Intelligence Concept to Actionable Roadmap,” co-hosted with Deloitte, we explored how organizations can move from an ambitious decision intelligence vision to a practical roadmap they can act on. Supply chains once ran on planning and optimization, and that worked well in a stable world. The last several years have brought a level of volatility, complexity, and ambiguity that planning alone can no longer absorb. AI promises to help, yet most enterprises still struggle to turn that promise into value, with only a small share seeing real returns and the large majority of pilots never advancing past the pilot stage.

This first episode of a three-part series laid out how to close that gap through a clear, decision-centric approach. We walked through three connected steps: shaping a vision anchored in better decisions rather than data alone, translating that vision into prioritized use cases, and structuring those use cases into a phased roadmap built to scale. A live demonstration showed how the Aera platform supports each step, and a customer example illustrated what the results can look like when the approach is followed with discipline.

Key Takeaways

• A decision intelligence vision must stay decision-centric, not data-centric. Many organizations set out to become data-driven or AI-native, assuming that seeing a problem is the same as acting on it. That assumption often proves false. A stronger vision focuses on the decisions that create value and treats data and AI as enablers of those decisions, rather than as the goal in themselves.
• Decisions fall into two types, and both can be supported. Structured decisions have fixed inputs, stable logic, and a bounded set of outcomes, which makes them well suited to digitization and automation. Situational decisions are shaped by novel circumstances and competing priorities, where human judgment still leads. The right approach depends on the specific decision, not on the limits of the technology.
• Human involvement is a dial, not a switch. Even among structured decisions, the level of oversight varies. Some run fully autonomously with a human out of the loop, some surface for approval and are governed by exception, and others keep a person involved in every decision where risk or novelty runs high. Matching that level to the decision is what makes automation both safe and useful.
• Turning vision into use cases follows a pragmatic, repeatable path. The work moves through orientation, ideation, and prioritization: understanding where value leaks today, identifying the decisions that could address those gaps, and weighing each option by estimated value against estimated complexity. Prioritization considers direct and indirect value alongside effort, data readiness, change management, and dependencies on other programs already underway.
• Scaling depends on how a roadmap is structured and paced. A program can deepen a single use case, extend it to new products or markets, or broaden into new decisions across the network. It can also advance sequentially or run several use cases in parallel to compress time-to-value. The starting point, whether a proof of concept or a committed roadmap, shapes how much momentum and buy-in the program can sustain.
• A self-funding roadmap protects momentum. Value captured in an early release can be safeguarded and reinvested to fund the next phase, so each wave is de-risked by the one before it. Waiting for perfect data tends to stall programs indefinitely; the more effective path is to think decision-backward and treat data as a step in the journey rather than a barrier to starting.
• The payoff is a business strategy, not a technology project. One customer brought three skills live within three months, saw the program pay for itself in the following three, and generated $2.7 million in value in eleven weeks. The same customer later cited $50 million in projected supply chain cost savings and $100 million in inventory reduction, the kind of outcome that can be communicated directly to shareholders.

Speakers

	Duncan Micklem, Client Partner, Aera Technology Duncan has more than 20 years of experience at the intersection of industrial operations and enterprise software, working with global organizations to turn AI investments into measurable outcomes. His background spans leadership roles at C3 AI, Yokogawa, and KBC, with a focus on closing the gap between strategy and execution and enabling more scalable, data-driven decision-making.
	Kevin Overdulve, Partner, Supply Chain & Network Operations Practice Leader, Deloitte Kevin leads the supply chain practice at Deloitte Belgium and drives the firm’s global decision intelligence agenda. He also serves as the Global Aera Technology Alliance lead, helping organizations transform supply chain decision-making through AI.
	Jeroen Nysen, Director, Supply Chain Analytics & Intelligence, Deloitte Jeroen leads the supply chain analytics and intelligence offering at Deloitte Belgium, with deep expertise in decision intelligence and control towers. He helps global organizations evolve from reactive, siloed operations toward connected, predictive, and AI-enabled supply chains.

Full Recording

Access the full webinar recording here.

Q&A

During the presentation, attendees submitted questions. Below are several of the key questions and answers.

Q: How is decision intelligence different from decision automation? And for building an actionable roadmap, have you seen examples where companies wanted to go slow by first focusing on automation before moving into the intelligence area?

A: Decision intelligence (DI) is a broad organizing principle for improving how decisions get made across the enterprise value chain. By enabling better, faster, and more frequent decisions, organizations unlock measurable business value. This improvement operates across two complementary dimensions:

1. Functional dimension: A close look at how decisions are made today, to find and remove inefficiencies such as too many people involved in a decision, overlooked value drivers, or too much exception handling, followed by a deliberate redesign of the process.
2. Technical dimension: Putting digital technologies to work, including data analytics, optimization algorithms, automation, and artificial intelligence, to further strengthen the quality and speed of decisions.

Decision automation describes how far technology-enabled automation is built into decision-making processes, directly supporting decision augmentation objectives such as faster decision cycles. This capability operates across three maturity stages:

1. Human in the loop: DI systems provide analytical insights and scenario simulations, while human stakeholders keep the final say over how to interpret them and what to decide.
2. Human on the loop: DI systems deliver actionable recommendations with full contextual analysis and impact projections, so human stakeholders can approve, modify, or reject them.
3. Human out of the loop: DI systems operate autonomously within predefined parameters, making and executing decisions independently while remaining fully auditable and governed.

Relationship to decision augmentation. Decision automation is a key enabler of decision augmentation, which itself is a primary objective within the broader decision intelligence framework.

Distinction from process automation. Decision automation should not be confused with robotic process automation (RPA), which focuses on automating routine workflows that require little or no real decision-making. Organizations looking to strengthen decision-making should adopt a decision intelligence lens rather than a purely process-automation approach.

Implementation approach. Organizations should start by getting clear on decision intelligence principles and turning that into a coherent organizational vision and roadmap. When piloting initial use cases, many organizations benefit from starting with simpler, more modest augmentation and building on it over time. This works best when decision intelligence initiatives are treated as evolving products rather than one-off projects, which allows for steady improvement across successive releases and ongoing refinement based on real experience.

Q: Decision intelligence use cases may also take control away from the operations team. What kind of change management actions do you follow (with a focus on the operations team) to improve acceptance and adoption of DI use cases?

A: Successful decision intelligence adoption hinges on early stakeholder alignment, transparent communication, peer-driven advocacy, and data-driven course correction. By positioning operations teams as partners rather than subjects of change, and by demonstrating tangible value to individuals, organizations can turn potential resistance into sustained adoption and competitive advantage. The four components below explain how.

1. Early stakeholder alignment: establishing a foundation of value-driven design.
   The cornerstone of successful DI adoption is establishing a value-driven mindset from the outset. Rather than imposing a solution, the design process must:
   • Diagnose existing pain points in the how decisions are made today, whether manual bottlenecks, long cycle times, fragmented stakeholder involvement, or unmanageable complexity, that create measurable value leakage.
   • Engage operations teams early as core design partners, not passive recipients. Their hands-on experience with process friction is invaluable for finding root causes and validating solutions.
   • Build consensus on objectives by bringing in functional, technical, and operational perspectives from day one, so everyone is aligned on both the value opportunity and the implementation approach.

   This early involvement transforms operations teams from skeptics into advocates, as they see their input directly shape the solution.

2. Transparent communication: reframing the narrative from job threat to value elevation.
   A critical psychological barrier in DI adoption is the fear of automation-driven job displacement. Effective change management requires deliberate reframing:
   • Shift the narrative from “automation replaces manual work” to “automation frees capacity for higher-value activities.”
   • Articulate the true objective, by reducing time spent on routine, low-discretion decisions so operations teams can focus on exception handling, strategic analysis, and process optimization.
   • Communicate tangible benefits to individual contributors, including career development opportunities, less repetitive work, and greater autonomy in decision-making.

   This messaging must be consistent, transparent, and reinforced by leadership throughout the implementation cycle.

3. Peer-driven advocacy: rolling out training and enablement, and empowering power users.
   Successful adoption requires structured enablement and peer-driven advocacy:
   • Develop comprehensive training materials tailored to different user personas (operators, supervisors, analysts) so everyone understands the DI system’s logic, inputs, and decision outputs.
   • Identify and empower power users, individuals deeply involved in design, development, and testing, to serve as change agents and ambassadors within the broader operations team. Their credibility and hands-on knowledge accelerate peer adoption.
   • Provide ongoing support during and after implementation to answer questions, troubleshoot issues, and reinforce best practices.
4. Data-driven course correction: continuously monitoring for improvement opportunities.
   After launch, ongoing monitoring makes targeted improvements possible:
   • Leverage monitoring dashboards (e.g., Aera’s Control Room) to track:
     • User adoption rates and engagement patterns
     • Root causes of system rejections or workarounds
     • Performance of DI use cases against baseline
   • Conduct root cause analysis on adoption friction, whether technical (system usability), organizational (unclear benefits), or behavioral (resistance to change), and address gaps with precision.
   • Iterate based on feedback, refining training, system design, or messaging to remove barriers and reinforce success.

Q: Does your decision model use causal AI architecture?

A: Aera does not use a causal AI architecture as its foundation. The platform runs on the Decision Data Model plus a hybrid, multi-engine stack: deterministic logic, optimization, machine learning, and agentic reasoning working together, rather than a single causal-inference paradigm. What Aera does support today are graph and knowledge techniques, including causal graphs, through its native knowledge graph and graph reasoning. Full causal reasoning is a named core capability on the roadmap.