Afleveringen
-
What happens when the “Claude Code moment” reaches hardware engineering, supply chain, and the factory floor?
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro speaks with two French industrial AI founders building directly into that shift:
Matthias Berahya-Lazarus, CEO & co-founder of Cognyx
Thibaut Wilhelm, CEO & founder of OplitCognyx is building an AI engineering platform for hardware — what Matthias frames as “Claude Code for industrial products.”
Oplit is building an AI supply chain platform for industrial companies — using agentic supply chains to optimize factory performance.
This conversation goes well beyond generic copilots.
We talk about why AI may expose that many companies never had a real digital thread, why supply chain is such a strong playground for agents, why engineering AI needs executable knowledge rather than another chatbot, and why the future factory will be far more software-heavy, automated, and AI-native than most people expect.
Key themes:
• AI shifting the bottleneck from execution to product thinking
• Supply chain as code
• The move from planners doing repetitive scheduling to managers defining objectives
• Why advanced factories may stop planning in Excel by 2030
• Why engineering needs composable, executable knowledge
• Why digital maturity determines whether industrial AI works or fails
• Why reindustrialization will not look like old factories coming back
• Why the most interesting industrial AI companies are solving deterministic, messy, high-value operational problemsIf you care about PLM, digital thread, engineering software, manufacturing AI, MES, supply chain planning, industrial data, or the future of European reindustrialization, this is a conversation worth hearing.
#IndustrialAI #AgenticAI #SupplyChainAI #EngineeringAI #DigitalThread #PLM #ManufacturingAI #FactoryOfTheFuture #Reindustrialization #HardwareEngineering #Cognyx #Oplit #ThreadMoat
-
Text-to-CAD. Autonomous simulation. Agentic workflows. AI copilots for PLM. Engineering teams “10x faster.”
Most of it still sounds like science fiction.
But what happens when you put two founders building real AI-native engineering software in the same conversation?
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro speaks with Pradyut, co-founder of Bild, and Martin Bielicki, co-founder and CEO of Bench, about what AI is actually changing in engineering software, CAD data, simulation, product development, and manufacturing workflows.
Bild is building CAD data management that connects engineering to manufacturing. Bench is building an AI orchestration layer across CAD, simulation, PLM, and beyond.
The discussion cuts through the hype:
AI is already changing how startups code. QA and validation are becoming the bottleneck. Prompting matters less than context. Frontier model cost is becoming a real burn-rate issue. And engineering AI will not move as fast as software AI because CAD, simulation, manufacturing, sourcing, and PLM are different technical worlds.
The real unlock is not “AI replacing engineers.”
Timeline
00:00 – Introduction: Bild, Bench, and AI across engineering
00:29 – Pradyut introduces Bild
00:54 – Martin introduces Bench
01:16 – The OpenAI moment
01:55 – Bench was created because of the LLM breakthrough
02:25 – Bild’s early exposure to DALL·E
03:33 – How AI changed startup coding
04:01 – Cursor, Claude Code, Slack, Graphite, and same-day delivery
05:45 – Multi-agent development
06:37 – Why “looking good” is becoming commoditized
07:28 – Why old software stacks limit AI innovation
08:21 – Prompting vs context
09:53 – From prompts to loops
10:40 – Frontier LLM costs
11:20 – Token costs as the new AWS-style shift
12:15 – AI spend caps and productivity measurement
13:45 – Cheaper models and model routing
14:39 – Right model, right task
15:39 – AI and engineering org structure
16:19 – QA, validation, and human-in-the-loop checks
17:58 – How AI may reorganize hardware teams
18:50 – Multi-agent coding conflicts
21:16 – Where AI lives inside the product stack
21:42 – Bench: AI for context, planning, and judgment
22:43 – Bild: opt-in AI for CAD data and IP boundaries
24:19 – When will engineering have its OpenAI moment?
25:04 – Why engineering AI evolves use case by use case
26:35 – Faster adoption in consumer products?
27:04 – From text-to-CAD to DFM and manufacturability
29:09 – The coming CDFAM AI demo wave
30:05 – Advice for young engineers
30:43 – Don’t compete with agents. Build differentiated skills.
33:05 – Creativity roles and AI in physical sciences
35:06 – Why top engineers become more valuable
35:57 – Digital transformation reality check
37:21 – Prints, redlines, and physical sign-offs are still alive
39:31 – Can startups move faster than legacy vendors?
40:10 – Big OEMs asking for AI engineering visions
41:15 – Buying AI vs buying value
42:50 – Why transformation programs route back to incumbents
43:37 – Build vs Windchill
45:30 – Startup visibility vs legacy vendors
47:56 – Capability checkboxes vs real user experience
49:13 – AI agents for CAD and CAE workflows
49:33 – End-to-end orchestration and organizational readiness
51:49 – Keeping skilled engineers in the loop
52:26 – Trust but verify for hardware AI
53:39 – Where to meet Bild and Bench
55:31 – Closing remarksFeaturing Pradyut of Bild and Martin Bielicki of Bench. Hosted by Michael Finocchiaro.
#AI #EngineeringSoftware #CAD #PLM #Simulation #Manufacturing #DigitalThread #IndustrialAI #HardwareEngineering #Startups
-
Zijn er afleveringen die ontbreken?
-
BOM Wars, Part 2: Why Engineering, Manufacturing, ERP, MES, Service, and AI Still Can’t Agree
The BOM debate is back. And somehow, it got even more dangerous.
In this Future of PLM panel, Michael Finocchiaro brings together Christine Longwell, Gus Quade, Brion Carroll, Pat Hillberg, David Schultz, and Oleg Shilovitsky for a high-energy debate on one of the most persistent fractures in product lifecycle management:
Who really owns the Bill of Materials?
Engineering says the EBOM defines the product.
Manufacturing says the MBOM defines what can actually be built.
ERP says the operational BOM is what matters.
MES wants execution context.
Service wants the as-maintained truth.
And AI? AI is useless unless all of this data is normalized, contextualized, and connected.This episode goes deep into EBOM vs MBOM, recipes vs discrete manufacturing, fashion vs aerospace, service BOMs, circular economy, Conway’s Law, ISA-95, product memory, data governance, and why every “single source of truth” eventually collides with organizational reality.
The conclusion?
The BOM is not just a list of parts.
It is a battleground between systems, silos, budgets, ownership, and the future of industrial AI.Timeline
00:00 – Introduction: BOM Wars, Part 2
00:54 – Christine Longwell and Gus Quade join the panel
02:20 – Autodesk’s MaintainX acquisition and service implications
03:14 – Jörg Fischer’s provocation: “The BOM doesn’t exist”
04:30 – ERP BOM vs MES vs MBOM: where does manufacturing truth live?
05:31 – Engineering defines the product, manufacturing defines the action
08:04 – Why BOM logic changes by industry
09:50 – Fashion, fabric, tech packs, suppliers, and PLM
12:29 – Should EBOM, MBOM, as-built, as-shipped, and as-serviced live in one system?
14:55 – Product memory and algorithmic BOM transformation
16:00 – Conway’s Law: why BOM structures mirror organizations
19:21 – Can product memory connect engineering and manufacturing logic?
21:02 – Is the MBOM a separate object or just a different view?
22:33 – Digital thread, service BOMs, and lifecycle responsibility
24:44 – TWA 800, aircraft traceability, and why as-built data matters
27:30 – Why silos exist because organizations exist in silos
28:29 – AI orchestration across PLM, MES, ERP, and service
30:36 – Service BOM, software BOM, spare parts, and terminology chaos
32:31 – Why AI needs normalized data before it can add value
33:58 – Conway’s Law and the limits of database-driven transformation
37:55 – EBOM vs MBOM through the CAD and manufacturing lens
39:31 – When does a product become “real”?
41:47 – CIOs, data governance, and who should own cross-silo truth
44:38 – Autodesk’s data model approach and the unified product record
45:06 – PTC Orbit, Jetstream, and the race toward digital thread platforms
48:36 – Master data models, ontologies, and common exchange standards
50:32 – Closing question: what BOM belief will be wrong in five years?
51:27 – Oleg: data misalignment and unit-of-measure disasters
54:44 – Gus: CAD-optimized vs ERP-optimized structures will stop being binary
57:37 – David: the danger of returning to point-to-point integrations
59:14 – Brion: role-based UX on top of shared product ontology
61:16 – Pat: digital thread will eventually collapse EBOM/MBOM boundaries
62:50 – Christine: BOM is product definition, not just a parts list
63:57 – Do we need BOM Wars Part 3?Featuring:
Christine Longwell
Gus Quade
Brion Carroll
Pat Hillberg
David Schultz
Oleg Shilovitsky
Hosted by Michael Finocchiaro#PLM #BOM #EBOM #MBOM #DigitalThread #Manufacturing #ERP #MES #EngineeringSoftware #AI #ProductLifecycleManagement #FutureOfPLM
-
🚨 PLM just had a major market signal.
For the first time since 2008, Gartner has published a Magic Quadrant for PLM — and Aras is now positioned alongside the traditional PLM giants.
In this breaking-news episode of AI Across the Product Lifecycle, I sit down with Josh Epstein, CMO of Aras, to unpack what this means for the PLM market, why digital thread has become central to enterprise software strategy, and why “governed engineering AI” may be the real battleground for the next generation of product development platforms.
We discuss why PLM has become too important for analysts to ignore, how Aras positions itself differently from Siemens, Dassault Systèmes, and PTC, and why AI in engineering cannot just be another copilot bolted onto messy enterprise data.
The key question:
Can AI transform engineering without a governed, explainable digital thread underneath it?Josh also goes deep on Aras Innovator Edge AI, Thread RAG, product memory, context graphs, lifecycle-aware AI agents, and what the engineer’s workday could look like when PLM starts decomposing into governed micro-experiences and agent-driven workflows.
If you care about PLM, digital thread, engineering AI, enterprise software, or the future of product development, this one matters.
⏱ Timeline
00:00 — Breaking news: Gartner brings back the PLM Magic Quadrant
00:36 — Why did Gartner wait so long after 2008?
02:06 — Has the PLM market fundamentally changed?
04:28 — Why PLM is more complex than ERP or CRM
06:05 — Aras vs. the “Big Three” PLM incumbents
06:37 — Why CAD-agnostic PLM may now be an advantage
07:21 — Governed engineering AI vs. generic AI hype
09:37 — Trust, governance, observability, and explainability
11:34 — Why AI needs the digital thread to be actionable
12:45 — PLM data complexity: versions, effectivity, access, context
15:05 — How to market AI in 2026 without overpromising
15:53 — Aras Innovator Edge AI, Thread RAG, and workflow agents
17:08 — Product memory, context graphs, and decision traces
19:08 — Does Gartner validation change the sales conversation?
21:17 — Is PLM still the right category name?
23:50 — Cognitive digital thread vs. product memory
26:00 — What does an engineer’s day look like in three years?
27:00 — Adaptive PLM, micro-experiences, and agent-driven work
29:30 — Why PLM AI cannot just be dumped into a data lake
31:30 — The physical-world constraint: “close enough” is not enough
32:00 — Has PLM had its OpenAI moment yet?🎯 Subscribe for more conversations on AI, PLM, CAD, manufacturing software, digital thread, and the next generation of engineering platforms.
💬 Comment THREAD if you want more deep dives on PLM, governed AI, and the engineering software startups reshaping this market.
#PLM #DigitalThread #EngineeringAI #Aras #Gartner #MagicQuadrant #ProductLifecycleManagement #AI #EnterpriseAI #Manufacturing #CAD #PDM #ProductDevelopment #IndustrialAI #AgenticAI #AIEngineering #DemystifyingPLM #ThreadMoat
-
What happens when some of the most respected voices in PLM gather in a Spanish vineyard to discuss AI, digital transformation, trust, community, and the future of engineering?
In this special Share PLM Summit 2026 edition of The Future of PLM Podcast, host Michael Finocchiaro is joined by Jos Voskuil, Oleg Shilovitsky, Rob Ferrone, Patrick Hillberg, Nina Dar, and Maria Morris for a candid, unscripted discussion about the ideas that emerged from one of the industry’s most unique events.
The conversation explores why the human side of PLM remains the hardest part of transformation, whether AI will fundamentally reshape consulting and knowledge work, how organizations build trust during digital change, and why community may be becoming more important than technology itself.
From AI adoption and organizational change to conference design and the future of professional expertise, this episode offers practical insights and thought-provoking perspectives from some of the industry’s most experienced practitioners.
Topics Covered• The evolution of Share PLM Summit and its human-centered approach
Timeline
• AI’s impact on engineering, consulting, and PLM careers
• Why trust may be the real ROI of conferences
• Lessons from successful and unsuccessful PLM transformations
• Human adoption versus technical implementation
• Digital transformation beyond software deployment
• The future of work in an AI-driven world
• Community, collaboration, and knowledge sharing00:00 Welcome & introductions
01:20 Why Share PLM Summit feels different
03:30 Breaking away from traditional PLM conferences
05:45 Why attendees travel across continents to attend
07:35 PLM as a people-centered discipline
09:50 AI, digital overload, and human connection
12:40 Measuring conference ROI beyond leads and sales
15:10 Most impactful presentations from the summit
20:05 Data, AI, and the Gentelligence perspective
22:10 Helena Haapio’s keynote and the future of work
24:50 Will AI replace consulting and expertise?
30:05 AI, critical thinking, and engineering risk
31:10 Sponsors, trust, and community building
36:00 Workshops, learning, and audience engagement
42:20 Sustainability and digital product passports
48:20 The Share Nest initiative
51:55 The future of conferences and professional development
58:20 Trust as the new business currency
01:01:00 Community, networking, and collaboration
01:03:40 The value of disagreement and debate
01:05:00 One word that defines Share PLM Summit 2026
01:07:00 Closing thoughts#PLM #AI #DigitalTransformation #Engineering #Manufacturing #Industry40 #DigitalThread #DigitalTwin #ProductLifecycleManagement #IndustrialAI #FutureOfPLM #SharePLM #EngineeringLeadership #SystemsEngineering #Innovation #TechnologyLeadership
-
AI in manufacturing does not fail because the demo is bad.
It fails when the answer cannot be trusted.
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro speaks with Fay Goldstein, Co-Founder and CEO of Bardin AI, and Scott Lionello, Co-Founder and CPO of Omnae Technologies, about where industrial AI is really going: beyond chatbots, beyond copilots, and into the operational workflows that actually run manufacturing businesses.
Bardin AI is building an application engineer for industrial automation sales and support teams, helping them answer complex technical questions without escalating everything to senior engineers. Omnae is building supply chain collaboration software that allows AI agents to operate safely across real suppliers, buyers, orders, invoices, and messy enterprise data.
The conversation goes straight into the hard parts of industrial AI:
trust, auditability, determinism, human-in-the-loop workflows, knowledge graphs, API costs, token burn, procurement risk, sales engineering bottlenecks, and why “just add a chatbot” is not enough when mistakes touch contracts, general ledgers, supply commitments, or customer trust.
Fay and Scott also discuss how AI is changing startup operations and software development, why young professionals need to show AI fluency rather than fear AI replacement, and why mid-market manufacturers may adopt practical AI faster than large enterprises waiting for top-down transformation programs.
The big takeaway: the next wave of industrial AI will not be about flashy demos. It will be about operational relief.
Fewer escalations.
Faster quoting.
Cleaner supplier collaboration.
Better support workflows.
Safer automation.
More trust in the decisions AI helps make.This is a grounded, founder-level conversation about how AI is moving into the less glamorous but highly valuable parts of the product lifecycle: sales, support, procurement, supply chain, and the industrial back office.
Topics covered: industrial AI, agentic AI, supply chain AI, procurement, pre-sales engineering, industrial automation, knowledge graphs, AI trust, human-in-the-loop workflows, manufacturing software, digital transformation, enterprise AI, startup innovation, and the future of AI across the product lifecycle.
-
What happens when AI moves beyond chatbots and starts reshaping the actual tools engineers, architects, and designers use every day?
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro speaks with Chloë Guidi of Qonic and Moritz Rietschel of Raven about the AI-native future of CAD, BIM, and AEC workflows.
Qonic is building a modern, cloud-based BIM platform from scratch, including its own solid modeling kernel, with a mission to make BIM lighter, faster, more accessible, and more data-rich. Raven is building AI-first workflows for complex design environments like Rhino, Grasshopper, Revit, Tekla, and Archicad, helping users navigate fragmented toolchains with less friction.
The conversation cuts through the hype and focuses on what is actually changing:
AI-assisted software development.
AI-native design workflows.
Smarter BIM quality checks.
More accessible CAD and AEC tools.
The economics of LLM-powered software.
The difference between “software built with AI” and “software that only makes sense because AI exists.”Chloë and Moritz also discuss whether engineering and BIM are heading toward their own “OpenAI moment,” why open standards and data quality matter, and what young engineers should do as AI changes the skills required to stay relevant.
This is a practical, founder-level look at how AI is moving into the real workflows of design, modeling, validation, and engineering decision-making.
Topics covered: AI in CAD, AI in BIM, AEC software, digital twins, Rhino, Grasshopper, Revit, engineering workflows, AI coding, MCP, open standards, startup innovation, and the future of AI-native engineering tools.
-
What happens when AI, virtual reality, and spatial computing move beyond demos and start reshaping real engineering work?
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro speaks with Jay Wright, Co-Founder and CEO of Campfire, and Oluwaseyi “Shay” Sosanya, Co-Founder and CEO of Gravity Sketch, about the future of immersive engineering workflows.
This is not a “metaverse” conversation. It is about what spatial tools can actually do for product development, design reviews, manufacturing validation, training, collaboration, and digital transformation.
Jay explains why AI is becoming a first-class user inside Campfire, acting almost like another participant in a 3D workspace. Shay breaks down why Gravity Sketch keeps humans at the center of the design process while using AI to remove friction, speed iteration, and help teams communicate better.
The conversation covers the hard parts too: why LLMs still struggle with geometry, why industrial companies remain cautious about cloud and AI adoption, why employees are already using AI tools outside official policy, and why the next breakthrough in engineering may not be AI replacing CAD, but AI controlling and accelerating the tools engineers already use.
For anyone working in CAD, PLM, industrial AI, digital thread, manufacturing, design, or engineering software, this is a sharp look at where spatial computing is actually useful and where the hype still needs to become workflow value.
Featuring:
Jay Wright, Co-Founder & CEO, Campfire
Oluwaseyi “Shay” Sosanya, Co-Founder & CEO, Gravity Sketch
Host: Michael Finocchiaro, AI Across the Product LifecycleTranscript source:
Timeline00:00 Welcome and guest introductions
03:05 Jay Wright on being bullish about AI after ChatGPT
04:33 Shay Sosanya on cautious optimism and the speed of AI progress
07:06 Why 3D geometry is harder for AI than language
08:42 AI capabilities are moving faster than expected
10:07 How Gravity Sketch adopted AI in software development
12:27 Campfire’s AI-assisted development workflow
13:32 AI agents in meetings, code, and product workflows
16:11 Using AI with existing 3D assets, BOMs, documents, and legacy data
18:26 Campfire’s spatial workflows for engineering, training, and sales
20:02 Where AI sits in the software stack
20:28 Campfire’s spatial agent as a first-class user
21:46 Gravity Sketch’s human-first approach to AI in spatial design
23:36 Foundation models, 3D generation, and geometry engines
25:29 AI cost, IP protection, customer data, and bring-your-own-LLM models
28:00 Has engineering had its ChatGPT moment yet?
29:05 Why physical product development will see staged AI adoption
31:17 The engineering-to-manufacturing gap
32:13 Simulating manufacturing workflows before production
34:12 AI connectors, Blender, Fusion 360, and tool control
35:18 Advice for young engineers worried about AI
39:41 Making real products, not just AI-generated concepts
40:00 Digital maturity in industrial companies
41:21 Why many manufacturers remain at low digital maturity
42:31 Headsets, cloud, InfoSec, and adoption barriers
43:39 Employees are already using AI and immersive tools informally
46:57 Can agile startups move industrial customers faster than incumbents?
48:17 Campfire on solving workflows rather than selling AI novelty
50:29 Gravity Sketch on value, workflow depth, and avoiding AI hype
53:09 Where to see Campfire and Gravity Sketch next
56:12 Closing thoughts -
Riverside Event Title
Physics Has a ChatGPT Moment: AI, Simulation, and the Future of EngineeringWhat happens when AI stops guessing and starts solving physics?
In this episode of AI Across The Product Lifecycle, I’m joined by Hardik Kabaria, co-founder and CFO of Vinci, and Andy Fine of the Fine Physics Consortium, for a sharp discussion on one of the biggest shifts in engineering software: AI-native physics simulation.
Vinci is building a physics intelligence layer: a foundation model for physics designed to answer real engineering questions around heat transfer, thermo-mechanical deformation, high-fidelity simulation, and manufacturing-resolution analysis. Hardik says Vinci is already deployed with tier-one hardware companies and can run simulations from hundreds of millions to over a trillion degrees of freedom.
This is not vague AI hype.
We dig into what makes AI simulation credible, why deterministic physics matters, how engineers can validate results, and why thermal problems are becoming mission-critical across semiconductors, electronics, batteries, EVs, data centers, robotics, and advanced manufacturing.
If your product generates heat, deforms under load, consumes power, or depends on simulation to avoid expensive failures, this conversation matters.
Timeline
00:00 — Introduction: Vinci, Fine Physics Consortium, and the “OpenAI moment” for simulation
01:11 — What is physics intelligence?
02:18 — Why physics is universal and governed by differential equations
03:08 — Physics-based AI vs. surrogate models
04:01 — What makes a physics foundation model credible?
06:51 — Why business value beats white papers
08:33 — Where Vinci fits in the engineering workflow
10:16 — Heat transfer, fluid dynamics, and choosing the right wedge use case
11:14 — Vinci’s focus: semiconductor and electronics thermal problems
13:23 — Thermo-mechanical deformation and why materials warp
14:49 — Multi-physics simulation as a long-standing engineering holy grail
16:06 — Yield, reliability, and manufacturing risk in electronics
17:04 — ROI: faster design loops and thousands of analyses per day
19:23 — Uncertainty, validation, and trust in AI simulation
20:08 — Training on 45TB of physics simulation data
21:47 — Residual norms and transparency at inference time
24:42 — 300 million to 1.2 trillion degrees of freedom
25:51 — GPU requirements and why Vinci is built for modern hardware
27:09 — Quantum computing, GPUs, and future scalability
30:22 — Wedge use cases: chips, boards, servers, batteries, defense, robotics, steel plants
31:45 — Who buys AI-native simulation software?
33:50 — Why thermal engineers are Vinci’s first target users
35:06 — Power, cooling, throttling, and data center energy constraints
36:25 — What throttling means in chips, EVs, and thermal runaway scenarios
39:58 — Deployment, IP protection, Docker containers, cloud, and on-prem
41:27 — How to convince skeptical engineers
43:00 — Path to adoption: start with the customer’s real benchmark
44:16 — What engineering leaders should do next
45:47 — The physics brick in the AI factory of the future
46:03 — Final debate: can there ever be one general foundation model for all physics?Join us for a practical, skeptical, deeply technical conversation about what AI can actually do for simulation, hardware design, and the next generation of engineering software.
#AI #Simulation #EngineeringSoftware #PhysicsAI #DigitalThread #Semiconductors #ThermalEngineering #CAE #ProductDevelopment #AIAcrossTheProductLifecycle #TheFutureOfPLM #BetterCallFino
-
Riverside Event Title
Product Memory: The Missing Layer Between PLM, Digital Thread, and AI AgentsRiverside Event Description
Everyone talks about the single source of truth.Then the real product decision happens in a meeting, spreadsheet, email, Teams chat, supplier exchange, or inside someone’s head.
In this episode of The Future of PLM, I’m joined by Oleg Shilovitsky of OpenBOM, Rob McAveney CTO of Aras, Brion Carroll of Digital Solution Group, David Segal of TCS, and Jonathan Scott of Razorleaf for a sharp discussion on one of the most important emerging ideas in PLM and enterprise AI: Product Memory.
The core question:
If digital thread connects the data, what captures the reasoning?PLM manages parts, BOMs, changes, documents, requirements, and workflows. But it often misses the “why” behind decisions: assumptions, rejected options, supplier constraints, manufacturing context, cost tradeoffs, effectivity logic, and informal reasoning.
This discussion explores whether Product Memory becomes the next layer above PLM, ERP, MES, QMS, ALM, supplier systems, documents, and collaboration tools: a contextual, semantic, AI-ready memory of how product decisions are made across the enterprise.
We cover:
Can Product Memory avoid becoming another inconsistent data layer?
What should be captured, and what should be filtered out?
Why does eBOM-to-mBOM still break so many digital threads?
How do semantics and ontology determine whether AI can trust product context?
Can AI agents safely recommend or execute PLM changes?
How do we capture human decision-making without scaring the humans?Timeline
00:16 — Introduction: single source of truth, broken digital threads, and Product Memory
03:02 — Oleg defines Product Memory beyond single source of truth and digital thread
06:28 — Rob on dependency graphs and hidden context in unstructured documents
08:36 — Brion on Product Memory as an “orb” fed by siloed enterprise systems
11:39 — Jonathan on semantics: why “part” means different things across functions
13:46 — David on Product Memory from an enterprise architecture perspective
18:21 — Avoiding inconsistent data across PLM, ERP, PIM, e-commerce, and supply chain
22:09 — Why engineering-to-manufacturing translation is so hard
25:00 — Why engineering release is not the finish line
30:05 — Missing memory: decisions in people’s heads, spreadsheets, and informal actions
33:57 — Why skipping change steps can slow the enterprise down
35:57 — AI agents, requirements ingestion, and asking “why” like a three-year-old
39:48 — Why AI agents must document their own reasoning
42:49 — Product Memory flywheel: capture, review, flow, and distribution
45:35 — Industrial AI, physical AI, agentic AI, and real-time product memory
48:21 — Semantic consistency, meta layers, and vetting data before Product Memory
52:15 — Dependency graphs, imperfect data, and improving ontology over time
55:12 — Human maturity: is the organization ready?
56:56 — Where companies should start looking for missing Product Memory
1:03:58 — Rob’s call to action: start capturing decision traces now
1:05:03 — Closing: eBOM, mBOM, ISA-95, and semantic translationThis is not a theoretical PLM buzzword session. It is a practical debate about architecture, governance, trust, and human maturity before AI agents can operate safely inside the product lifecycle.
#PLM #ProductMemory #DigitalThread #AI #AgenticAI #EngineeringSoftware #EnterpriseArchitecture #BOM #MBOM #EBOM #Manufacturing #TheFutureOfPLM #BetterCallFino
-
What happens when AI hits both sides of the engineering equation: design and sourcing?
In this episode of AI Across the Product Lifecycle, Michael Finocchiaro sits down with Adar Hey, CEO and co-founder of Jiga, and Or Israel, CEO and co-founder of Bananaz, for a grounded discussion on where AI is actually creating value in engineering right now. Bananaz is building an AI layer on top of CAD to automate manual engineering work, while Jiga is rethinking custom part sourcing with software, supplier intelligence, and AI-enabled operations.
This is not a hype piece. The conversation gets into the real tradeoffs: where LLMs help, where deterministic workflows still matter, how engineering startups are using AI internally to ship faster, how customers think about ROI, and why security, traceability, and IP protection still make or break adoption. It also explores a bigger question: when will engineering have its true “OpenAI moment”? Adar argues adoption in physical industries takes time even when the technology is ready, while Or says the shift is already underway and could become unmistakable in 2026 to early 2027.
One of the strongest parts of the episode is the discussion around digital maturity. Both founders place many target customers around a 2 to 3 out of 5: digital enough to understand the value, but far from autonomous or agentic. From there, the discussion turns practical: how do you introduce change without breaking habits, and how do you prove business impact across engineering, manufacturing, and supply chain?
If you care about CAD copilots, sourcing automation, engineering productivity, AI in industrial software, startup execution, and the future of digital engineering, this episode is worth your time.
Timeline
00:14 — Intro: Adar Hey of Jiga and Or Israel of Bananaz
00:40 — What Bananaz does: AI layer on top of CAD
01:24 — What Jiga does: sourcing custom parts more efficiently
02:20 — Were they bullish or skeptical on AI in 2022?
06:01 — How AI changed the way they build software
10:50 — Token costs, burn rate, and ROI of AI tools
14:22 — Where AI sits in the product stack
18:00 — Off-the-shelf LLMs vs open-source models
20:15 — Bring-your-own-model vs vendor-managed AI
22:22 — Security, SOC 2, and protecting customer IP
26:01 — Are they more bullish now than in 2022?
27:28 — Who owns IP when designs are partially AI-generated?
31:52 — Advice for younger engineers worried about AI replacing jobs
35:49 — When will engineering get its “OpenAI moment”?
40:09 — Digital maturity of current customers
42:29 — Do tools like Jiga and Bananaz move the maturity needle?
47:30 — Closing thoughts and where to meet the founders nextHashtags
#AI #EngineeringAI #CAD #PLM #DigitalThread #Manufacturing #SupplyChain #IndustrialAI #EngineeringSoftware #AgenticAI #Jiga #Bananaz #AIAcrossTheProductLifecycle #BetterCallFino
-
What happens when engineering teams suddenly have the equivalent of 10,000 new AI coworkers?
In this Day 2 Threaded Miami session, Daan Goossens of CoLab lays out a sharp version of the problem: AI can massively increase productivity, but if companies do not have the right collaboration, context, and decision-making structure in place, they will not scale output. They will scale chaos. That framing runs through the entire talk and makes this one of the clearest strategic discussions from the event.
Daan explains that CoLab’s vision for the future is not about replacing engineers. It is about helping humans do more with more, pairing human judgment with AI agents and the right engineering context. His point is that the engineer remains accountable, especially in safety-critical industries, but AI can dramatically expand what teams are capable of if it is embedded responsibly.
He reduces the path to scaled engineering productivity down to three ingredients: human collaboration and decision-making, strong AI agents, and relevant context and data. Miss one of those and the whole thing breaks. That is why CoLab is building around engineering collaboration first, especially design review, where teams already need to bring together multiple stakeholders, surface issues, and make decisions quickly without drowning in screenshots, PowerPoints, email chains, and Teams messages.
Daan also gives a concrete look at where CoLab is going next. He shows how their design engagement system and AI reviewer, Otto, are evolving into a broader engineering operating system strategy. One especially strong example is the SimScale partnership proof of concept, where a user can request a static load analysis on a crane truss, let the AI build and review the simulation plan, run the simulation, and then bring the results back into CoLab for collaborative review alongside the rest of the design context. The broader message is clear: engineers should not have to keep jumping between disconnected tools just to understand the impact of a design decision.
This is less a product demo than a thesis on where engineering software is headed. CoLab is betting that the future belongs to platforms that can combine collaboration, AI, and engineering context in one place, while partnering with the best core tools rather than trying to rebuild everything themselves.
If you care about AI in engineering, design review, simulation workflows, or the emerging idea of an “engineering OS,” this episode is worth your time.
#ThreadedMiami #CoLab #EngineeringAI #DesignReview #Simulation #DigitalThread #IndustrialAI #ProductDevelopment #EngineeringOS #ManufacturingTech #AI
-
What if the missing piece in industrial AI is not another copilot, but the underlying knowledge and orchestration layer that physical product companies still do not have?
In this Day 2 Threaded Miami session, Lucy Hoag of Violet Labs makes that case from a hardware-first perspective. Her argument is that the physical world is getting dramatically more complex, multidisciplinary, and AI-enabled, while the way companies manage product data is still stuck in a much older, mechanically centered paradigm. The result is familiar: disconnected systems, fragmented context, and no reliable foundation for AI to reason over or act on.
Lucy frames Violet as the knowledge and orchestration layer for the physical world. The company starts by pulling together data across the lifecycle through a large and growing set of no-code integrations spanning requirements, CAD, PLM, MES, ERP, simulation, and domain-specific aerospace tools. That data is normalized into a shared ontology so that parts, items, inventory, requirements, and related objects can be understood consistently across systems instead of remaining trapped in tool-specific silos.
What makes the talk timely is her emphasis on AI readiness. Lucy is clear that generative AI and agentic workflows are exciting, but they do not work reliably if the underlying data is disconnected. Violet’s answer is not just sync for sync’s sake. It is to create the infrastructure that lets companies build reports, automate workflows, trace decisions, and ultimately expose governed, permissioned engineering data to agents through things like MCP. In her framing, AI is not the starting point. Connected context is.
She also gets into the less glamorous but more important details: observability, auditability, approval logic, hybrid sync models, webhook support, and the messy reality of older engineering tools that do not behave like modern SaaS apps. That gives the presentation more credibility than a generic “single source of truth” pitch. Violet is trying to solve the boring infrastructure work that has to exist before agentic AI becomes operationally trustworthy.
Another strong part of the talk is how broad the ambition is without pretending everything needs to live inside Violet’s own UI. Lucy points to chat interfaces, multi-source reports, BOM comparison, clear-to-build views, and MCP-driven custom apps as different ways users can consume the same underlying data foundation. That suggests Violet sees the future less as one monolithic interface and more as a connected data layer that other applications, agents, and teams can build on top of.
This is a useful episode for anyone in PLM, systems engineering, aerospace, hardware startups, manufacturing IT, or industrial AI. It is a thoughtful argument that before the industry gets carried away with agents doing everything, it still has to solve the much older problem of fragmented engineering knowledge.
#ThreadedMiami #VioletLabs #DigitalThread #PLM #SystemsEngineering #IndustrialAI #HardwareEngineering #ManufacturingTech #AgenticAI #EngineeringData #AI
-
What if the real weak point in the digital thread is not the software stack, but the documents people keep creating to survive around it?
In this Day 2 Threaded Miami session, Addy First of Quarter20 makes that argument directly. She points out that while everyone talks about connected systems and AI in manufacturing, a huge amount of real work still happens outside those systems in PowerPoints, Word docs, PDFs, screenshots, and spreadsheets. Those documents become the human-facing interface to engineering, manufacturing, service, quality, and supply chain work — and the moment they are created, they often detach from the source of truth.
That is the core problem Quarter20 is going after. Addy argues that companies can either force every worker, technician, and partner to operate directly inside enterprise systems, or they can fix the documentation problem itself. Her view is that the second path is far more realistic. Documentation is not going away, so the real opportunity is to make it dynamic, connected, and continuously updated rather than static and stale.
She lays out four reasons this matters. Engineering intent often fails to reach execution cleanly. Changes do not propagate reliably once documents are manually created. Work happening through documents creates no useful traceability. And without structured, trustworthy documentation, applying reliable AI becomes much harder. Her punchline is strong: as long as humans are doing work, humans will need documents, and that means documentation has to be brought back inside the digital thread instead of treated as an afterthought.
Quarter20’s answer is a human-facing collaboration layer that sits between systems and teams. The platform ties documents to source data, uses tagged content that can update automatically, and helps teams create, revise, comment on, and reissue documents across the product lifecycle. The point is not just faster authoring. It is preserving context, propagating change, and making downstream teams less dependent on stale text and manual search.
Addy also gives concrete results from early deployments. Customers are reportedly spending about 70% less time creating documents and 95% less time updating them. She also points to gains in first-pass yield and reduced downtime in field service scenarios where technicians were previously being sent outdated PDFs that did not match the machine in front of them.
This is a useful episode for anyone in PLM, manufacturing, service, quality, or industrial AI. It goes after a problem a lot of companies quietly live with every day: the digital thread looks connected on slides, but in reality, documents are still where a lot of the truth gets lost.
#ThreadedMiami #Quarter20 #DigitalThread #ManufacturingAI #PLM #Documentation #FieldService #WorkInstructions #IndustrialSoftware #EngineeringOps #AI
-
What if the real problem in electromagnetic simulation is not user workflow, but the solver itself?
In this Day 2 Threaded Miami session, Masha Petrova of Nullspace makes exactly that argument. Her claim is blunt: the core electromagnetic solvers most of industry still relies on were built decades ago for very different computing environments, and they are no longer good enough for the scale and complexity of modern RF, radar, satellite, automotive, and defense problems.
Masha explains why this matters now. Electromagnetic systems are everywhere, but the hard part is not simulating an antenna in isolation. The hard part is simulating that antenna on the real platform it lives on: a satellite, a drone, a car roof, a larger system where the surrounding body fundamentally changes performance. That is where legacy tools start to break down, forcing engineers to decompose problems, approximate more aggressively, or move into physical prototyping earlier than they want.
Nullspace was built to attack that gap directly. Masha positions the company as a modern full-wave 3D electromagnetic solver designed for electrically large problems, with a proprietary matrix-compression approach and multi-CPU, multi-GPU acceleration. Her pitch is not that Nullspace is a lightweight wrapper around old tools. It is that the underlying numerical engine has been rebuilt for current hardware and current problem scale.
The talk gets especially strong when she grounds it in concrete examples. She walks through antenna-on-platform use cases like CubeSats and automotive shark-fin antennas, showing why the real-world body changes the electromagnetic behavior enough that isolated component simulation is not sufficient. She also highlights phased-array antenna problems, where Nullspace reportedly scales to larger cases with lower memory consumption than incumbent tools in a benchmark run by a defense customer.
Another timely angle is AI readiness. Masha notes that Nullspace is built around a Python API, which makes it easier to integrate into the new generation of AI-driven engineering workflows. Her point is practical: if AI tools are going to help automate simulation setup and execution, they need tools underneath them that already speak the language of modern automation.
This is a useful episode for anyone in RF, electromagnetics, aerospace, automotive, defense, or engineering software. It is not a generic “AI for simulation” story. It is a direct challenge to the assumption that the old solver layer is good enough.
#ThreadedMiami #Nullspace #Electromagnetics #Simulation #RF #Radar #Aerospace #DefenseTech #EngineeringSoftware #CAE #AI
-
What if the real problem with engineering simulation is not physics, but the fact that the tools are still too complex, too fragmented, and too dependent on a handful of experts?
In this Day 1 Threaded Miami session, John Zinn of CognaSIM lays out that case clearly. He argues that simulation teams have become a bottleneck inside engineering organizations because the software is hard to use, hard to standardize, and hard to review across teams. Companies want to democratize simulation and move it earlier into design, but in practice they are still stuck with siloed experts, repetitive setup work, and inconsistent workflows across Ansys, Altair, Siemens, and other toolchains.
CognaSIM’s answer is a universal agentic AI layer for simulation. John describes a system that sits across different simulation tools and gives engineers a common workflow and interface, while also turning simulation instructions into executable plans. Instead of manually hunting through menus, searching YouTube for features, and rebuilding the same setup over and over, engineers can describe the simulation they want, let the system generate a plan, review it, and then have the tool build the simulation automatically.
A big part of the talk is about wasted engineering time. John points out that teams often spend hours not on analysis itself, but on setup, repetition, and tool friction. His case study with an electrified off-road vehicle startup makes that concrete: a battery-pack analysis that took 91 minutes by hand was dramatically compressed by CognaSIM, with one especially painful task, creating bolted connections, reduced from about 20 minutes to less than a minute. The larger point is that much of simulation work is still expensive human labor spent on low-value setup rather than actual engineering judgment.
He also pushes toward a more ambitious vision: simulation-driven design. Instead of running isolated analyses and handing off static results, John wants AI to help connect structural, thermal, modal, crash, and other simulations into a loop that can drive design changes, rerun workflows, and help engineers evaluate tradeoffs much earlier in development. That is a much bigger claim than “AI assistant for FEA.” It is a claim about making simulation more central, more scalable, and more reusable across the product lifecycle.
Another important angle is governance and consistency. John notes that many companies already have simulation design guides and standards, but enforcing them is slow and manual. CognaSIM aims to encode those rules into workflows so teams can not only move faster but also stay aligned with required methods and review expectations.
This is a strong episode for anyone working in CAE, FEA, engineering design, simulation automation, or industrial AI. It is focused less on AI hype and more on a very real bottleneck: too much valuable engineering time is still being burned just getting simulations set up and understood.
#ThreadedMiami #CognaSIM #Simulation #CAE #FEA #EngineeringAI #Ansys #ProductDevelopment #DigitalEngineering #SimulationDrivenDesign #AI
-
In this Day 2 Threaded Miami session, Rut Lineswala of BQP makes that case head-on. He argues that the weakest link in the digital thread is often the simulation layer itself, where engineers still face huge compute demands, long runtimes, and practical limits that force them to make design decisions with extra safety factors instead of better data. His point is simple: if modeling and simulation remain too slow and too expensive, the rest of the digital thread can only be so good.
Rut brings serious technical credibility to the argument. Drawing on his background in aerospace and high-performance simulation, he describes working on hypersonics-scale workflows that consumed around 200,000 CPU/GPU cores and still took close to a month to return results. That experience led him and his co-founder to a sharper question: instead of just throwing more hardware at the problem, why not redesign the solver architecture itself for the compute platforms we actually have now — and for the ones that are coming next?
That is where BQP’s story gets interesting. Rut argues that many incumbent solvers were built over decades for CPU-heavy environments and have only been awkwardly ported to GPUs, which is why organizations often end up using only a fraction of the GPU capacity they are paying for. His claim is that quantum-inspired solvers can make much better use of modern architectures today, while also preparing companies for the coming shift to actual quantum hardware. In his framing, this is not just about speedups. It is about making engineering organizations “quantum ready” before that transition becomes urgent.
He also grounds the pitch in real industrial outcomes. Rut shares examples ranging from optimization work that delivered lighter aerospace designs without sacrificing structural integrity to physics-AI models compressed enough to run on edge devices for space-related applications. The broader message is that the same underlying technology can cut compute costs, improve design quality, and open up workflows that were previously too slow or too expensive to attempt.
This is one of the more ambitious talks from Day 2 because it is not just pitching another engineering tool. It is arguing that the future of simulation infrastructure itself is up for grabs, and that companies that stay tied to legacy solver assumptions will eventually get boxed in.
If you care about CAE, HPC, quantum computing, engineering simulation, physics AI, or the next compute inflection point in industrial software, this episode is worth your time.
#ThreadedMiami #BQP #Simulation #CAE #QuantumComputing #PhysicsAI #EngineeringSoftware #HPC #DigitalThread #IndustrialTech #AI #AIAcrossTheProductLifecycle
-
What if the biggest bottleneck in manufacturing is not engineering itself, but all the manual work that happens after the design is done and before the business can actually move?
In this Day 1 Threaded Miami session, Mayank Makwana of Reeva makes that case directly. He argues that speed is the only sustainable advantage in manufacturing, but most large companies are still slowed down by the invisible layer of manual reconciliation work sitting between PLM, ERP, CRM, Git, Jira, PIM, spreadsheets, PDFs, emails, and tribal knowledge. The systems may be technically integrated, but the real process still depends on humans chasing updates, interpreting edge cases, and carrying business logic in their heads.
Mayank’s point is that this hidden layer is where launches slip, costs rise, and quality breaks. A change may be approved in PLM, but weeks later the part still has not launched because somebody still has to update a spreadsheet, resolve a field mismatch, or translate engineering intent into something another system can act on. He gives examples that will sound painfully familiar to anyone in discrete manufacturing: part data that only partially maps between systems, firmware changes that trigger manual back-and-forth, and longtime employees whose undocumented judgment quietly keeps operations running until they leave.
Reeva’s pitch is to replace that layer with AI agents. Instead of ripping out core systems, the platform sits between them, reads technical documents and structured data, understands what changed, identifies what downstream systems and workflows are affected, and drafts or executes the required updates. The model is designed to be reviewable, governed, and adaptive, with the system learning from exceptions over time rather than relying on brittle static scripts.
That makes this a useful talk because it goes after a very real category of operational drag. A lot of industrial AI still gets framed around copilots or analytics. Reeva is going after the repetitive coordination work that actually slows launches: the emails, the spreadsheets, the undocumented rules, the schema mismatches, and the “Sarah has been here 22 years and knows how this plant works” problem.
This is a conversation about turning manufacturing workflows from fragile human glue into structured, executable, reviewable automation. If you care about PLM, ERP, engineering change, product launches, or AI agents that do more than summarize documents, this episode is worth hearing.
#ThreadedMiami #Reeva #ManufacturingAI #PLM #ERP #EngineeringChange #DiscreteManufacturing #AIAgents #DigitalThread #WorkflowAutomation #AI #AIAcrossTheProductLifecycle
-
What if factory data worked less like a pile of dashboards and more like Instagram or TikTok — surfacing the right operational insight at the right moment without forcing people to hunt for it?
In this Day 1 Threaded Miami session, Jeff Tao of TDengine makes exactly that argument. He explains why industrial teams are still drowning in dashboards, alerts, and raw time-series data, while the real need is much simpler: operators, engineers, and executives want to know what matters now, why it matters, and what action to take. His thesis is that industrial software needs to move from query to feed, from pull to push, and from raw data to contextualized insight.
Jeff frames the problem through the lens of his own journey as a serial entrepreneur and then goes straight into the operational pain point. Traditional factory and utility data systems still rely heavily on humans building dashboards, configuring rules, and interpreting endless streams of signals. That model is slow, brittle, and hard to scale, especially for smaller companies that cannot afford full-time data analysts or process specialists. His vision is an AI-native industrial data foundation that can detect anomalies, identify patterns, forecast outcomes, and present what is happening as a stream of meaningful operational stories rather than static charts.
A major theme of the talk is contextualization. Jeff argues that raw sensor data is rarely useful on its own. What matters is turning continuous machine signals into business-relevant events with meaning: what happened, when it started, how long it lasted, how it compares to baseline, and what the likely business impact is. That is where he sees the future of time-series infrastructure going, especially in the AI era, where events and context need to become first-class citizens instead of afterthoughts layered on top of storage.
He also outlines the technical stack required to make that vision real: time-series storage, real-time analytics, process analytics for root-cause work, standardized data models, asset and event modeling, contextual semantics, and AI-friendly interfaces that can expose the system to agents and other applications. The pitch is not just “let AI do it.” It is that AI only becomes useful once the underlying industrial data foundation is organized, standardized, and open enough to support meaningful reasoning.
One of the sharper points in the session is who this benefits most. Jeff argues that AI-native data infrastructure can flatten access to insight for smaller and midsize industrial businesses that historically lacked the people and budget to build sophisticated analytics teams. In that sense, the talk is about more than data architecture. It is about democratizing operational intelligence.
This is a useful episode for anyone working in industrial software, manufacturing analytics, time-series data, plant operations, or AI for the factory floor. It is opinionated, practical, and built around a very clear idea: the future of industrial data is not more dashboards. It is better understanding delivered automatically.
#ThreadedMiami #TDengine #IndustrialAI #ManufacturingAnalytics #TimeSeriesData #FactoryData #Industry40 #OperationalIntelligence #DigitalManufacturing #AIforIndustry #AI #AIAcrossTheProductLifecycle
-
What if the real problem in supply chains is not visibility at the top, but the broken collaboration happening lower down the stack where the real work still runs on emails, spreadsheets, and mismatched systems?
In this Day 1 Threaded Miami session, Scott and Dan Linonello of Omnae deliver one of the more memorable presentations of the event: a father-and-son talk built on decades of firsthand supply-chain pain. Their pitch is sharp. Omnae is not just another supplier portal for large enterprises. It is a natively multiplayer collaboration system built to actually work for the small and midsize suppliers who still carry a huge share of real operational complexity.
Scott explains the core problem clearly: even when big companies have EDI, ERP, and supplier collaboration platforms in place, most of the actual execution still happens outside those systems. Small suppliers use different tools, different workflows, and often no real shared system at all. That creates constant mismatches in orders, invoices, revisions, quality issues, and delivery expectations. Omnae’s answer is to create a shared operational layer where enterprises get a digital twin of their supplier activity, while smaller suppliers get an actual usable application rather than a portal they are forced to feed.
The talk gets more interesting when Dan takes the conversation into AI and agentic systems. His argument is blunt: most agentic tooling today lives safely in sales and reporting workflows because those domains can tolerate mistakes. Supply chains cannot. If an agent creates the wrong PO, approves the wrong operational change, or triggers the wrong action upstream, the damage is immediate and expensive. That is why Omnae treats agents as participants that can propose actions, but not execute them unilaterally. In their model, operational, legal, and financial commitments still require acceptance on the other side before a state change occurs.
That makes this more than a software demo. It becomes a strong point of view on how AI should be used in operational supply chains: not as unchecked automation theater, but as a structured system for surfacing signals, proposing decisions, and escalating intelligently when confidence is not enough. Dan also explains how Omnae can pull signal from emails, WhatsApp, Slack, and direct system-to-system communication, turning messy unstructured supplier communication into something companies can actually analyze and act on.
The origin story gives the whole session weight. Omnae was built from the Linonellos’ own manufacturing and sourcing experience, then hardened inside their own business before being turned outward as a product. One story in particular lands hard: a Boeing supply-chain issue that left 18 Dreamliners stuck on the tarmac over a tiny part and a revision mismatch. That experience helped shape Omnae’s multi-tier communication and permission model, designed to surface issues earlier and route them to the right level before they become major operational failures.
This is a useful episode for anyone working in supply chain tech, manufacturing operations, procurement, industrial AI, or digital thread strategy. It is practical, skeptical of hype, and grounded in the ugly reality of how engineered products actually get made.
#ThreadedMiami #SupplyChain #Procurement #Manufacturing #IndustrialAI #DigitalThread #SupplierCollaboration #Omnae #ERP #Operations #AI #AIAcrossTheProductLifecycle
- Laat meer zien
