Agentic AI Use cases

Every assembly, plant, stock position and open order a change touches, before the board.

An engineering change order, an ECO, is the record that alters a design, its bill of materials (BOM) and everything built from it.

WHAT JUNO’S ROLES WOULD DO WHEN A CHANGE IS RAISED01A change is raisedthe change request entersimpact analysis in the PLMsystem02Where-used,everywhereassemblies and plantsacross PLM, SAP S/4HANAand Oracle Fusion03Stock and ordersstock by plant, openpurchase and sales orders,planned orders04Draft the recordaffected items, aneffectivity date afteruse-up, order changes05People approvethe analyst, the buyer,the change board, thestewardedited dates and rejected drafts feed the next rule version
One change, read across five systems in place and drafted as one record for four named approvers. Designed, yet to be built.
7.0 days
median engineering change order, request to production, across 4,075 companies (APQC, checked 23 September 2026)
dated source
10%
of new product development cost goes to change orders, median across 325 companies (APQC, checked 23 September 2026)
dated source
90%
of engineering leaders delay launches because of late design changes (CoLab, 26 July 2024)
dated source
14
assemblies at three plants touched by one showcase change, with six open order lines worth $86,000
showcase estate
A moment in the day

Where this use case starts

PS
Tuesday 09:10 · controller board 7741-X · a change request enters impact analysis

At 09:10 the product engineer moves a change request on controller board 7741-X into impact analysis: a voltage regulator is replaced by a pin-compatible part from Acme Components GmbH. Juno reads the where-used across SAP S/4HANA and the Oracle Fusion division, where the same board carries a different item number, and finds 14 assemblies at three plants.

It also finds 2,150 units of the old regulator in stock at plant 1210, six open purchase order lines with Acme Components GmbH worth $86,000, and two sales orders whose delivery dates fall inside the use-up. The draft proposes an effectivity date after use-up, seven weeks out, and two order changes for the buyer.

P. Sharma, master-data steward, approves the affected-item list at 10:20 after moving one assembly the draft had placed at the wrong plant. D. Okafor approves the two order changes. The change board on Thursday sees one page, with the query on show and each figure traced to its system. showcase estate

Why this is a top use case

What the market is doing, with the evidence

The PLM and ERP vendors now ship change agents, and each works inside its own system. Siemens introduced a Teamcenter AI BOM agent that finds impacted assemblies, evaluates alternatives and executes updates across occurrences, with a person confirming each update (Siemens, 23 June 2026) 1; Teamcenter 2606 adds recommendations for change workflow templates and task assignment (Siemens, 12 June 2026) 2.

SAP’s Joule Change Record Management Agent analyses problem reports and change requests, recommends next steps and initiates changes, planned for the second quarter of 2026; SAP claims up to 20% less time creating change requests, a vendor figure (SAP, 7 October 2025; SAP, January 2026) 3, 4. Oracle’s Component Replacement Assistant reads part numbers from change notices, runs where-used across every impacted structure and adds use-up analysis to the change request (Oracle, 30 July 2026) 6. PTC’s Arena AI Engine compares files across revisions (PTC, 9 December 2025) 10, and Dassault Systèmes’ Aura companion orchestrates requirements, projects and changes (Dassault Systèmes, 11 February 2026) 20.

The pain is measured across thousands of companies. The median engineering change order takes 7.0 days from request to production across 4,075 companies, and change orders take a median 10% of new product development cost across 325 companies (APQC, checked 23 September 2026) 28, 29. 90% of engineering leaders delay launches because of late design changes, and 10% launch every product on time (CoLab, 26 July 2024) 30; 70% of chip respins trace to specification changes (Semiconductor Engineering, 27 March 2025) 31.

The gap the vendors’ agents leave is the one Juno is built for. Each agent sees one system, and a change touches PLM, two ERPs, the planning suite, the supplier portal and the quality system at once. 400 senior manufacturing professionals ranked PLM the top system for agents to reach and named change histories among the data agents need (ASSEMBLY, 25 August 2026) 15. Gartner expects PLM AI to evolve into agent-driven capabilities that automate cross-functional workflows, as quoted by one vendor (PTC Arena, 11 August 2026) 18.

By hand today

How it is done today

The engineer raises a change request in the PLM system. A change analyst runs where-used in PLM and again in each ERP, because bills of materials differ by plant, and pastes the results into a spreadsheet. A planner checks stock, open purchase orders, planned orders and work in progress for each affected part at each plant.

A buyer e-mails suppliers and contract manufacturers about open orders and tooling. Quality checks the qualification status of any alternate. Finance estimates scrap and rework. The change board meets weekly and approves the change with an effectivity date, and the master-data steward updates the ERP change master, BOM validity and routing. Eight people, five systems and a median of 7.0 days per change (APQC, checked 23 September 2026) 28.

Today

Eight people, five systems, one spreadsheet

  • Where-used in PLM, then in SAP S/4HANA, then in Oracle Fusion, pasted into a sheet
  • Stock, open orders and work in progress checked plant by plant
  • Suppliers and contract manufacturers e-mailed about open orders and tooling
  • Qualification, scrap and rework estimated by quality and finance
  • The weekly board approves; the steward updates the ERP by hand
7.0 daysmedian, request to production (APQC, checked 23 September 2026)
With Juno

One draft, four approvals, confirmed write-back

  • Where-used across PLM and both ERPs in one query, read in place
  • Stock, open purchase and sales orders and planned orders in the same draft
  • Effectivity date after use-up, order changes and supplier notices drafted
  • The analyst, the buyer, the board and the steward approve by name
  • Write-back to PLM and the ERP with the acknowledgement kept
one pagebefore the board, every figure traced (showcase estate)
Stage by stage

How Juno enables it

Change request, affected item, bill of materials, where-used, effectivity, stock, purchase order line and sales order line are objects the supply chain ontology already holds, so this use case starts from the map the rest of the site describes. The six stages below say what would be registered, asked, taught and written back, and who would approve each draft.

01
Register
PLM change and BOM tables; the BOM, change master, stock, purchasing and sales tables in SAP S/4HANA and Oracle Fusion; the planning suite’s planned orders; the supplier portal’s confirmations; the quality system’s qualification records. All read in place. Objects: change request, affected item, where-used, effectivity date, stock by plant, open purchase and sales order lines, planned and production orders, supplier, approved manufacturer part, routing, qualification status, product change notice.
02
Map
Change request, affected item, bill of materials, where-used, effectivity, stock, purchase order line and sales order line are ontology objects already. The estate’s PLM columns would be mapped to them with a confidence and a reason. The link between the SAP item number and the Oracle item number for the same board is found and confirmed, so where-used runs once across both.
03
Ask
“What does this change touch across both ERPs, and what is the open-order value?” and “Which assemblies at plant 1210 still carry the old regulator after the proposed effectivity date?” The answer names the systems it read, shows the query and applies the asking engineer’s access rights.
04
Teach
The company’s change classes, effectivity codes and use-up terms: “use-up” means building out the stock of the old part before the new one takes effect. A product change notice, a PCN, is the notice the company sends its own customers; its template words would be taught so the draft reads as the company writes.
05
Act
Trigger: a change request enters impact analysis, a PCN arrives, or a daily scan finds open changes with impact data missing. The policy: “When a change enters impact analysis, list every affected assembly, plant, stock position and open order; draft an effectivity date after use-up where stock covers demand; draft order changes under $25,000 for the buyer.” It would be released at A2, the change analyst approving each draft. Write-back: the PLM system returns the revision, the ERP the change number.
06
Learn
Every edited effectivity date, moved assembly and rejected order change would be recorded with the rule version and the facts at the time. The measure that improves is the share of changes with complete impact data before the board, with hours of impact analysis per change beside it; the next rule version carries what the analysts corrected.
TRIGGERDRAFTDECIDETHE BOARDWRITE BACKJunoChange analystBuyer · D.OkaforChange boardSteward · P.SharmaWhere-used,three systems14 assemblies ·three plantsDraft the recorditems · effectivity· order changesConfirm set anddateedits where thedraft is offApprove orderchangestwo lines · above$25,000Approve thechangeThursday changeboardApprove masterdatachange master · BOM· routingWrite back,confirmedPLM revision · ERPchange numbershowcase estate · designed, yet to be built · a named person approves every draft at A2
Who does what on the Tuesday change, lane by lane: Juno drafts and writes back; four named people decide. Showcase estate; the roles are designed, yet to be built.
Who does what

The person and the agent

The agent A2 Supervised
  • Finds every assembly, plant, stock position and open purchase and sales order the change touches, across PLM and both ERPs
  • Drafts the affected-item list, an effectivity date after use-up and the order changes, each with its basis
  • Writes the approved record to PLM and the ERP and keeps each acknowledgement
  • Records every edit and rejection for the next rule version
a drafted change record with its basis; then the approvals, by name
The person
  • The change analyst confirms the affected-item set and the date, editing where the draft is off
  • D. Okafor, buyer, approves each order change above the value threshold
  • The change board approves the change and its effectivity date each week; P. Sharma approves the master-data update after it
  • The quality engineer decides whether an alternate is qualified; an auditor samples the trace

The market runs this at draft-and-approve: the PLM and ERP vendors’ change agents analyse, recommend and initiate, and a person confirms each BOM or record update (2025 to 2026) 1, 3, 6. The blueprint designs the write-back roles here at A2 as well, with the change analyst as the named approver: ECO tracking and propagation, product change notice issuance and spec baseline control at A2, and the two impact assessment roles at A3, because an assessment reads and a propagation writes.

What you would configure

The settings on this agent

Agent settings · Engineering change impact across BOM, stock and orders
ElementSettingSet byChanges
ObjectiveEvery change request reaches the change board with its full impact across PLM, both ERPs, the planning suite and the supplier portal, and a drafted effectivity dateThe engineering change manager, in a sentenceNew version
Data scopePLM change and BOM tables; SAP S/4HANA and Oracle Fusion BOM, change master, stock, purchasing and sales; planned orders; supplier confirmations; qualification recordsA. Novak, data leadNew version
RulesThe impact policy and its planning table: list what the change touches; effectivity after use-up where stock covers demand; draft order changes under $25,000The change manager with the Planner AgentNew version, simulated first
TriggerWhen a change request enters impact analysis; when a PCN arrives; a daily scan at 07:00 for open changes with impact data missingThe change managerAt once
Actions permittedAttach the impact report · write the affected-item list and effectivity proposal · draft a purchase order change · set a use-up flag · notify a supplier · draft a PCNThe change manager, ownerNew version
LimitsOrder changes above $25,000 go to the buyer with the reason · 40 drafts a day · confidence floor 0.85 on a where-used match · a routing change always waits for the stewardThe change manager and J. Harrison, financeNew version
EscalationA change with an alternate, to the quality engineer · a change touching a product in ramp, to the programme manager at once · a failed write-back, to the ownerThe change managerAt once
LevelA2 Supervised with the change analyst as the named approver; D. Okafor approves order changes; the change board decidesThe engineering directorUp on evidence, down in one step
FeedbackEvery edited date, moved assembly and rejected draft is recorded and usedAlways onFixed

Every row is a setting or a versioned rule; the level and the limits are the ones a manager signs.

Built from the blueprint

The roles this use case would draw on

Lima’s agent blueprint is a design of 960 agent roles across five end-to-end chains. It is the map of what Juno can be configured to do, and the first agents are built from it one at a time, the way the PO cancellation agent was.

Blueprint · designed, yet to be builtThese roles are designed on the map. Each is configured from the same parts.
Conceive 2 roles
Requirement change impact A3
Spec baseline & freeze control A2
Design 3 roles
Design change impact assessment A3
ECO tracking & propagation A2
Physical ECO implementation A3
Launch 1 role
Product change notice issuance A2
Tapeout and Silicon 1 role
Mask revision & ECO control A2

Additional capability in Idea-to-Market lists the rest of the map for this chain.

How it is measured

The measures that show it working

Measure
Change order cycle time, baseline 7.0 days median (APQC)
Change order cost share of new product development cost, baseline 10% median (APQC)
Hours of impact analysis per change
Share of changes with complete impact data before the change board
Open-order and stock exposure per change, in currency
Launches delayed by late design changes, baseline 90% of companies report delays (CoLab, 2024)
Time to create a change request, vendor claim of up to 20% less (SAP, 2026)

Baseline the share of changes with complete impact data before the change board first. It is a count the board secretary can take from the last quarter’s minutes, it is the number the board feels, and it moves in the first month. Change order cycle time follows, against APQC’s 7.0-day median 28; the cost share moves over a year of programmes, so it is read per programme.

Terms on this page
Engineering change order (ECO)The record that alters a released design and everything built from it, with an effectivity date.
Where-usedThe list of every assembly and plant whose bill of materials contains a given part.
Effectivity dateThe date from which the changed bill of materials applies to new builds.
Use-upBuilding out the stock of the outgoing part before the new one takes effect, so the stock is used.
Product change notice (PCN)The notice a manufacturer sends its customers before it changes or ends a part.
Questions
Why does the where-used run in three places today?

The PLM system holds the engineering bill of materials, and each ERP holds its own manufacturing bill per plant with its own item numbers. Juno maps all three to one bill-of-materials object, so one query reads them together.

Who approves what?

The change analyst confirms the affected-item list and the date. D. Okafor approves order changes above $25,000. The change board approves the change. P. Sharma approves the master-data update. Each approval is recorded by name.

What would be written back, and how is it confirmed?

The impact report and the affected-item list to PLM, which returns the revision; order changes and the use-up flag to the ERP, which returns the change number and the material status; the supplier notice through the portal, which records the acknowledgement.

Next in this chain
The question this leaves

This change was the company’s own decision. What happens when the change comes from outside: a supplier ends a part that sits on nine active bills of materials?

Answered next

Component end-of-life watch with alternates

Every end-of-life notice caught, the exposure and the last-time buy sized, and a qualified alternate proposed while the last order date is still open.

Sources

Dated sources used on this page

  1. [1] Siemens Digital Industries Software (Teamcenter blog), How AI BOM agents transform engineering change workflows, 23 June 2026 (checked 23 September 2026) · https://blogs.sw.siemens.com/teamcenter/ai-bom-agent-plm-engineering-change/
  2. [2] Siemens Digital Industries Software (Teamcenter blog), Introducing Teamcenter 2606, 12 June 2026 (checked 23 September 2026) · https://blogs.sw.siemens.com/teamcenter/teamcenter-2606/
  3. [3] SAP News Center, Innovative Updates for Supply Chain Management (SAP Connect), 7 October 2025 (checked 23 September 2026) · https://news.sap.com/2025/10/sap-connect-innovative-updates-supply-chain-management/
  4. [4] SAP, Joule Agents: Change Record Management Agent (solution brief), January 2026 (checked 23 September 2026) · https://www.sap.com/assetdetail/2026/01/d8db813c-367f-0010-bca6-c68f7e60039b.html
  5. [6] Oracle Help Center, Oracle Fusion Cloud Product Lifecycle Management 26A What's New: Improved Component Replacement Assistant, 30 July 2026 (checked 23 September 2026) · https://docs.oracle.com/en/cloud/saas/readiness/scm/26a/plm26a/26A-plm-wn-f42907.htm
  6. [10] PTC, PTC Launches Arena AI Engine to Accelerate Intelligent Automation Across PLM and QMS Workflows, 9 December 2025 (checked 23 September 2026) · https://www.ptc.com/en/news/2025/ptc-launches-arena-ai-engine
  7. [15] ASSEMBLY Magazine (Propel Software survey of 400 senior manufacturing professionals), Manufacturers Rank PLM as Top Data Source for AI Agents, 25 August 2026 (checked 23 September 2026) · https://www.assemblymag.com/articles/100340-manufacturers-rank-plm-as-top-data-source-for-ai-agents
  8. [18] PTC Arena (blog quoting the Gartner Magic Quadrant for PLM Software in Discrete Manufacturing Industries, 9 June 2026), PTC (Arena) Named a Visionary in the 2026 Gartner Magic Quadrant for PLM in Discrete Manufacturing Industries, 11 August 2026 (checked 23 September 2026) · https://www.arenasolutions.com/blog/arena-2026-gartner-magic-quadrant-plm-visionary/
  9. [20] Dassault Systèmes, Dassault Systèmes Unveils a New Way of Working for Industry with AI-Powered Virtual Companions, 11 February 2026 (checked 23 September 2026) · https://www.3ds.com/newsroom/press-releases/dassault-systemes-unveils-new-way-working-industry-ai-powered-virtual-companions
  10. [28] APQC, Engineering change order (ECO) cycle time in days (Open Standards Benchmarking measure; median 7.0 days, n=4,075), undated, checked 23 September 2026 (checked 23 September 2026) · https://www.apqc.org/what-we-do/benchmarking/open-standards-benchmarking/measures/engineering-change-order-eco-cycle
  11. [29] APQC, Engineering change order (ECO) costs as a percentage of the total new product development cost (median 10.0%, n=325), undated, checked 23 September 2026 (checked 23 September 2026) · https://www.apqc.org/what-we-do/benchmarking/open-standards-benchmarking/measures/engineering-change-order-eco-costs
  12. [30] CoLab Software (survey of 250 engineering leaders by Global Surveyz), 90% of companies delay product launches, 26 July 2024 (checked 23 September 2026) · https://www.colabsoftware.com/research/90-of-companies-delay-product-launches
  13. [31] Semiconductor Engineering (reporting the 2024 Siemens EDA and Wilson Research Group Functional Verification Study), First-Time Silicon Success Plummets, 27 March 2025 (checked 23 September 2026) · https://semiengineering.com/first-time-silicon-success-plummets/

See Juno answer and act on your own data

Bring two systems and one decision your planners make every week. In the first week we register the data, confirm the mapping with your team, answer your questions in ChatLima and run the first agent at the level you choose.