"Why don't we just pour all our data into Claude and use that as our planning tool?"If you work in supply chain planning, you have probably heard that question in your own company. It sounds reasonable. The general AI assistants are impressive, everyone is already using them, and they will happily answer any question you throw at them. This episode takes that question seriously and answers it properly.Søren Hammer Pedersen hosts this session of the S&OP MasterClass and brings in Benjamin Obling, CPO of Perito IBP at Roima Intelligence. Benjamin spends his working life embedding AI agents into integrated business planning software, which makes him exactly the right person to explain where the general assistants end and dedicated planning AI begins.The surprising starting point is that the two are built on the same raw material. The LLMs behind an embedded planning agent and behind Claude or Copilot can be identical. What separates them is instruction, context, and governance. An embedded agent knows which page you are on, which step of the process you are in, which of your six different forecast versions you actually mean, and which data your role allows you to see. A general assistant knows none of that, and it will still give you a confident answer.By the end of the conversation you will understand where each kind of AI belongs in your planning setup, why verification is the difference between a useful alert and a dangerous one, and why the real test of any AI analysis is a simple question. Did we make a better decision?In this episodeWhy "just pour the data into Claude" keeps coming up, and what that question missesWhat actually separates embedded AI from bolt-on AI when the underlying LLMs are the sameHow role-based governance keeps planners inside the data they are allowed to seeWhen a general AI tool is still the right choice for a plannerWhat changes for planners and executives in the monthly process, and what stays the sameChapters01:10 Welcome and the question every planner hears02:04 Bolt-on and embedded AI defined05:05 Instructions, context and the six forecasts problem09:17 Security, roles and data governance11:10 The cost of iteration and pre-prepared analysis14:18 Sharing best practice through one data model16:38 When to use which tool19:26 Same monthly process, different division of labour21:47 The shadow AI pitfall and proving the answer25:01 How planners actually receive embedded AI27:32 Less is more and the decision testContact and followQuestions, topic ideas, or guest suggestions: podcast@roimaint.comFind more episodes at: https://www.roimaint.com/en/catalog/node/insights-webinars-events-and-podcastsIf you want to know more about what we do in Perito IBP, we are here to help.Production This podcast is brought to you by Roima.This podcast is produced by Montanus.
Every shift on a modern factory floor throws off more signal than any team could ever read. Machines, quality checks, maintenance logs, operator comments, all of it lands in systems of record, and almost all of it goes unread. Managers make do with a handful of dashboards while the answers to their most expensive problems sit in the data, waiting for someone with the time to go looking. Most of the time, nobody does.In this S&OP Masterclass from Roima, host Søren Hammer Pedersen sits down with Rafael Amaral to look at what changes when agentic AI is pointed at that data. Rafael has spent around 20 years in manufacturing technology, split between supply chain planning and manufacturing execution systems. He is CTO and co-founder at TilliT, now part of Roima, where he leads the engineering team behind the TilliT stack, a cloud native MES application.The conversation moves from planning, the subject of the previous episode, onto the shop floor, where the data is densest and the losses are most expensive. Rafael explains Aura, an agentic AI toolset that reads factory data, forms its own hypotheses, writes its own queries, and returns findings a manager can act on. Think of it as a small team of data scientists that never sleeps, correlating things a human would rarely think to compare.You will come away understanding how the analyst bottleneck really works, why the most valuable factory insights are the ones nobody has time to find, and what it looks like to walk in on a Monday morning to a board of evidence-backed recommendations rather than a fresh round of firefighting.In this episodeThe "data is gold, so where is my shovel" problem with systems of record like MES and ERPHow Aura forms its own hypotheses, writes its own queries, and refines its analysis with no human in the loopThe agent swarm explained as a mini factory, a plant manager agent directing maintenance, quality and operations specialistsThe CFO agent that adds your labour and utility costs so findings arrive with the financials already worked outWhy the second wave of AI in supply chain is one professionals cannot afford to sit outChapters03:18 What agentic AI and Aura are04:40 The problem of siloed factory data06:41 A team of data scientists on demand12:43 Correlating machines with operator processes14:38 The agent swarm as a mini factory19:08 What you get on Monday morning20:22 Concrete findings from the floor24:15 The CFO agent and the financials27:32 The second wave of AI in supply chainAbout Rafael Amaral Rafael Amaral is CTO and co-founder at TilliT, now part of Roima, where he heads the engineering team behind the TilliT stack, a cloud native MES application. He has worked in manufacturing technology for around 20 years, with the first half of his career in supply chain planning and the second half in manufacturing execution systems. He has spent that time close to the shop floor, from his first TilliT customer's plant manager to the breweries and wineries he happily admits a soft spot for. His current focus is Aura, the agentic AI toolset that he describes as changing the game for how much value teams can pull out of their own manufacturing data. Contact and followQuestions, topic ideas, or guest suggestions: podcast@roimaint.comFind more episodes and get in touch through the Roima website.Production This podcast is brought to you by Roima.This podcast is produced by Montanus.
The word "agentic" is everywhere in supply chain right now, and rarely defined. In this S&OP MasterClass from Roima, host Søren Hammer Pedersen sits down with Rafael Amaral, CTO and Co-Founder of TilliT, to cut through it.They trace the shift from the first wave of AI, which made our existing forecasts and plans better, to the agentic wave, where systems can take a goal and work towards it on their own.Rafael offers a genuinely usable definition of agentic AI, explains why giving a model tools changes everything, and is candid about why so many projects fail, from the 10,000-bottle order nobody approved to the temptation to dump an entire dataset into the context window.The conversation then turns practical: where agentic AI is already creating value in planning, how optimisation is being democratised beyond PhD-level data science, and how Roima's Aura connects the data silos across IBP, MES, WMS and PLM to surface insights teams have been missing.They close on the question everyone asks: what does this mean for the people doing the work?Essential listening for supply chain managers, Heads of S&OP, IBP managers and anyone weighing where to place their next bet.Key takeawaysThere are two waves of AI in supply chain. The first made existing tasks better; the second, agentic wave takes on goals and works towards them. "What is my OTIF?" is a request; "how do I increase my OTIF?" is a goal the system works towards on its own.A model becomes agentic when you give it tools to query data, run code or take action, and let it navigate them in a loop. More power means more risk, so guardrails and gates are non-negotiable.Projects fail on missing basic assertions and on dumping too much raw data into the model. Success comes from pairing the right use case with the right methodology.Aura gives AI the power of a developer, writing and running its own code across the full dataset, security-built from the ground up. It is like hiring a team of specialists you could never afford, and Roima's digital thread lets it connect insights across IBP, MES, WMS and PLM silos that never spoke to each other.Waiting for 100 per cent accuracy is the wrong test, because humans make mistakes too. The real question is the cost of an error and the gates around it, and so far these tools empower planners and keep them in control.Chapters03:37 Why supply chain professionals need to care now04:53 From the first wave of AI to the agentic wave08:29 What "agentic AI" actually means14:11 The big pitfall: dumping all your data into the model15:18 Where we are now, and the tipping point19:41 Democratising optimisation without an army of PhDs24:30 Real use cases: from chatbots to Aura32:11 The digital thread: connecting the silos33:44 What this means for the humans, and wrap-upGuest and host Guest: Rafael Amaral, CTO and Co-Founder of TilliT (Roima Intelligence). More than twenty years in manufacturing technology across supply chain planning, optimisation and execution systems.Host: Søren Hammer Pedersen, CCO, PERITO IBP at Roima Intelligence. Production This podcast is brought to you by Roima.This podcast is produced by Montanus.
Most companies know they need an S&OP process. Few have one that actually works.The gap between having the meetings and having a functioning integrated business plan is where most organisations quietly struggle, and where the biggest operational and commercial risks accumulate. In this episode of the S&OP MasterClass, Søren Hammer Pedersen sits down with Benjamin Obling – a practitioner with 16 years of focused S&OP experience across industries and organisations – to work through the first two weeks of a monthly S&OP cycle. Not the theory.The actual steps: who does what, when, with which tools, and what goes wrong when any of it is missing. Benjamin brings the perspective of someone who has seen both extremes up close: companies with sophisticated IBP tools and no process, and companies with well-structured processes and nothing to put in them. His argument is that neither works alone. The foundation of a reliable S&OP process is built in weeks one and two, in demand planning and supply planning, the engine room of the S&OP process.Get those two weeks right, and the rest of the cycle becomes a decision-making conversation grounded in data. Get them wrong, and you spend weeks three and four firefighting.This episode is the first in a two-part series. Today covers demand planning and supply planning. Part two will go into the balancing and executive S&OP stages.In This EpisodeWhy having an S&OP process without the right data tools, or the right tools without a process, produces equally poor resultsThe four main steps of a monthly S&OP cycle, and why skipping the first two guarantees chaos in the last twoThe difference between minor and major forecast overrides and why most minor adjustments are a waste of timeWhy inventory planning is a strategic decision disguised as a technical one and the risks of leaving it to a few specialistsHow to run a supply-side MRP simulation before uploading your demand plan to the ERP systemThe two non-negotiables for a high-functioning S&OP process: strong data tools, and clear process ownershipEpisode contents00:07 Recap: the four phases of S&OP08:49 Why balancing week matters — and why most companies skip it15:20 The "loudest voice wins" problem in demand prioritisation30:03 Simulating MRP before it hits the ERP system31:47 Extending visibility up and downstream: suppliers in the picture34:26 Rule of delegation: what to resolve in week 3 vs. escalate to week 436:18 Outputs of the balancing week25:37 How to get C-level to actually show up and engage27:25 Structuring the executive meeting: State of the Union + decisions09:17 Avoiding the post-planning firefighting trap37:22 The fifth step: following up on actual planning behaviour38:56 Summary & closing thoughts39:13 Outro & call to actionProduction This podcast is brought to you by Roima.This podcast is produced by Montanus.
AI is moving fast and it’s changing the way supply chains are run. But a big question remains: how do you tap into its power without losing control or replacing the judgment you and your team bring to the table?In this episode of the S&OP MasterClass, Søren Hammer Pedersen sits down with Benjamin Obling, a supply chain expert with 16 years at Roima.You’ll hear real examples of how companies are using AI in demand forecasting and inventory planning without cutting people out of the process. If you’ve been wondering how to get AI working for you, without creating blind spots or overreliance, you’ll want to listen in. We talk about practical ways to build trust, add smart governance, and use AI where it makes sense while keeping human judgment front and center.In this episode, you'll learn about:How to balance AI automation with human judgment in planning.Key AI-driven improvements in demand forecasting and inventory management.Importance of trust and governance in adopting AI strategies.How to determine what to automate and where humans are needed.Embracing AI while maintaining control through guardrails and alerts.Episode content00:44 The rise of AI in supply chain planning03:39 Current challenges in AI integration for planning04:51 Trust and governance in AI predictions07:03 Finding starting points for AI automation in planning08:45 Leveraging AI across different planning disciplines09:13 Identifying critical areas for human oversight12:19 Balancing automation and human judgment16:56 Knowing when to trust AI predictions19:18 Ensuring transparency in AI-driven processes23:36 Impacts of AI on sales and operations planning29:10 Future outlook for AI in supply chain planning Production This podcast is brought to you by Roima.This podcast is produced by Montanus.