Go back to Market Research
ClearOps supplies software for machinery manufacturers and their dealer networks. It connects dealer management systems, plans spare-parts inventories and replenishment, locates parts across dealers, and coordinates workshop and technician work. A separate Celonis-based planning application uses operational data to investigate and explain replenishment proposals, with approved decisions written back to enterprise systems. 1 2 4 8
Overall supply chain score: 4.8/10.
| Dimension | Score |
|---|---|
| Supply chain depth | 4.8/10 |
| Decision and optimization substance | 5.4/10 |
| Product and architecture integrity | 4.8/10 |
| Technical transparency | 4.6/10 |
| Vendor seriousness | 4.4/10 |
ClearOps addresses difficult operational work: maintaining dealer connections, reconciling part identities, and producing replenishment decisions across fragmented service networks. Its inventory methodology also has more substance than a generic AI claim. The weakness is the economic objective: configurable service targets and inventory policies remain central, while the cost of machine downtime, uncertainty about future supply, and trade-offs across the whole service system are much less explicit. 3 5 14
The agentic presentation raises a separate issue. Explaining an existing calculation may reduce investigation time, but it does not establish better numerical decisions or remove the recurring work of maintaining policies and integrations. ClearOps has documented conventional automation that should form the baseline for evaluating its newer agents. Their incremental economic and labor benefits remain unmeasured publicly. 7 9
ClearOps versus Lokad
ClearOps starts with an organizational boundary: the manufacturer, its independent dealers, their systems, and the service work they perform. Lokad’s inventory optimization approach starts with the numerical decisions and their economic consequences under uncertain demand and supply. ClearOps has specific expertise in getting usable data through dealer networks; the comparison becomes more demanding when deciding what to do with that data. 3
Consider a cheap part whose absence stops an expensive machine. A cluster fill-rate target can encourage stocking frequently requested parts while undervaluing a rarely needed component with severe downtime consequences. Conversely, raising every service target can immobilize capital in items with little economic justification. The relevant objective should value the consequences of each shortage, repair delay, substitution, transfer, and additional unit of stock. ClearOps’ published inventory formulation exposes cost inputs and service constraints, but not this complete economic calculation. 5
The perspective developed in Introduction to Supply Chain also treats software as a continuing process of improving decisions. For a dealer network, that includes the labor of maintaining part mappings, changing rules, validating results, and recovering failed transactions. Faster proposal review is useful only as part of that broader productivity account. The customer evidence below establishes integration work more clearly than a measured reduction in the total cost of that operation.
History: connectivity, optimization, and the AI repositioning
Genpact traces the product idea to an AGCO consulting project in 2014 and describes a separate ClearOps business emerging in 2020. Barkawi’s portfolio dates the SaaS offering to 2016. These accounts describe different stages of the same commercial lineage: consulting work, a deployed application, and a standalone company. 17 22
| Period | Public direction | Evidence and implication |
|---|---|---|
| 2022 | IoT-enabled connectivity and service execution. | A company announcement reported five million parts planned daily and 30,000 orders processed weekly. These were activity figures, not measured decision quality. |
| September 2022 | ClearOps and PTC Servigistics partnership. | The announcement assigned downstream connectivity to ClearOps and multi-echelon optimization to Servigistics. |
| 2025 | Genpact and Celonis partnerships. | Genpact presented a transition toward a “Big Data Hyperscaler”; Celonis added process intelligence and a digital-twin narrative. |
| November 2025 | Inventory Planning 2.0 with TUM. | A disclosed change to the replenishment algorithm, with a defined service-constrained objective. |
| May 2026 | €8.6 million Series A. | The first institutional round, led by Hitachi Ventures with Schoeller Group and Barkawi Group, funded expansion and further AI development. |
| 2026 | Agentic planning and ClearWhy AI. | The visible new mechanism is explanation and investigation around planning outputs, with ERP execution subject to approvals. |
There is a consistent aftermarket focus, and the algorithm work represents a material product change. The surrounding language nevertheless expands faster than the evidence. Hyperscaler, digital twin, and agentic platform describe very different technical propositions. A network of interfaces does not become a hyperscaler because it processes many records. A representation of past workflows does not demonstrate accurate forecasts of future failures, repairs, or supply. 17 16
The 2022 PTC announcement already used AI and machine-learning language, particularly for Servigistics. Those capabilities belong to the named partner and product generation. ClearOps’ current native planning and Celonis application require their own assessment; the earlier partnership cannot establish how their present algorithms work. The reviewed record contains substantial integration continuity, but no public technical account tying the different planning engines into one maintained architecture. 15 5 8
The numerical decision process
Forecasting and replenishment policies
Retail Inventory Management ingests part master data, supersessions, stocks, and orders. It classifies items, evaluates more than 20 forecasting models per SKU and location, and sets safety stock, reorder points, and order quantities. Approved purchase orders can pass into dealer or manufacturer systems through files or APIs. A parts locator and promotion management operate alongside this planning flow. 2
The model count says little about forecast quality. The useful tests concern intermittent demand at the actual decision horizon, calibration of uncertainty, treatment of unobserved demand during stockouts, and how forecasting errors change replenishment economics. Choosing the model with the lowest historical error can still yield poor stock decisions when shortage and excess costs are asymmetric.
Inventory Planning 2.0, announced in November 2025, was developed with TUM’s Stefan Minner and Patrick Helm. Its published description is specific: a periodic-review policy jointly chooses reorder points and quantities within product segments, minimizes costs subject to item and cluster fill-rate requirements, and supports storage and budget constraints. Its intermittent-demand model makes the probability of demand depend on elapsed time since the previous order. 5
This is substantive inventory optimization. Joint segment decisions can allocate stock more effectively than independent fixed buffers. The formulation still makes the service target a controlling input. A weighted fill rate values fulfilled units according to its weighting scheme; it does not automatically value machine downtime, contractual penalties, customer retention, or competing repair alternatives. Segment boundaries can also prevent a scarce resource from being allocated to its best use elsewhere in the network.
The published method leaves the probability distribution of future replenishment lead times and delivered quantities unspecified. A fixed lead time is especially consequential for slow-moving parts: the risk is often a rare demand occurring during an unusually long replenishment delay. Forecasting demand more carefully cannot compensate for treating that delay as known. ClearOps’ September 2026 planner essay criticizes static supplier lead-time assumptions, but does not disclose the supply distribution used by its products. 5 10
The time-since-last-order mechanism also needs empirical validation. An increasing chance of demand is a modeling assumption, not a universal property of intermittent service parts. Installed-base changes, maintenance schedules, and equipment retirement can change the relationship. The public description does not expose the calibration procedure or its sensitivity to those changes. 5
Automatic publication already existed
A July 2025 support article documents Auto Publish: a scheduled replenishment job sends created orders to the Dealer Management section without manual review. Enabling it requires a location-specific job instance that replaces the standard replenishment job. The article identifies an internal Python function, replenishment_planning.PublishCreatedOrders. Its documented scope is publication into Dealer Management, rather than proof of every downstream purchase transaction completing unattended. 7
This is useful automation with a concrete configuration mechanism. It also sets a baseline: a new agent should demonstrate additional work removed, better decisions, or safer handling of exceptional conditions. Renaming a planner as an orchestrator does not measure any of those improvements.
The agentic claim: explanation around an existing algorithm
The current Celonis application connects a planning agent to the Celonis Context Model through MCP. It investigates exceptions, supplies explanations, and writes approved decisions to ERP within stated limits, with an audit trail. It builds on the customer’s Celonis environment rather than providing an independent replacement for that platform. 8
MCP is a red flag here because the advertised advance centers on giving an agent access to the same context and actions used by planners. Connecting tools is a modest engineering step compared with designing an economically sound decision policy. The public material does not establish that this connection changes the optimization objective, models supply uncertainty, or removes the need for individual judgment at scale.
ClearOps’ September 2026 account describes ClearWhy AI as a working prototype built for a Celonis/AWS hackathon. It states the division clearly: the planning algorithm calculates the decision and the AI explains it. Examples concern the drivers of a reorder point and an economic order quantity. The proposed benefit is less manual investigation and fewer defensive overrides; it is not a demonstrated improvement in the underlying algorithm. 9
The hackathon organizer identifies AWS Bedrock Agents and reports potential reductions of 25% in working capital and 20% in operating costs. These are prospective claims for an explanation prototype, without a named deployment, comparison group, or cost account. Winning a development competition does not establish either economic result. 24
Explainability has a further limitation. A fluent account of why a service target produced a particular quantity can make the result easier to approve while leaving the target economically wrong. ClearOps argues that trust was the bottleneck once forecasting was sufficiently good. That diagnosis requires evidence: were rejected recommendations actually better, under what loss function, and after which operational costs? The published account supplies no such evaluation. 9
The later planner essay explicitly recognizes that data access alone is insufficient, and discusses objectives, permissions, constraints, feedback, and changes in supply. These are appropriate concerns. Its example of moving stock after a delayed shipment is a proposed agent workflow, not a measured customer result. The gap is between that design ambition and a repeatable decision system whose quality and labor requirements can be tested. 10
Customer results and their economic meaning
The customer cases describe how dealer data, parts planning, and service execution connect in practice.
| Customer | Documented work | Evidential limit |
|---|---|---|
| AGCO | More than 2,000 dealer outlets and 50 DMS integrations; customized replenishment and service processes; continuing operations and data maintenance. Reported results include 20% better fill rate, 15% higher parts sales, and 9% less non-moving inventory. | The case omits metric definitions, comparison periods, and total delivery and operating costs. |
| Terex | Telematics inform service dates and work-order proposals; planned services feed parts planning; EDI and web-service interfaces carry orders. | Reported operating improvements lack a quantified net economic result. |
| Jungheinrich | Automated master-data updates and multi-supplier replenishment across concurrent sea and air delivery streams. | The case explains the integration problem, but not the uncertainty model or measured financial return. |
| Lippert | Matching supplier part numbers for the same physical component, synchronizing master data, and deriving service bills of materials from work-order history. | The matching and configuration mechanisms are described without accuracy or net savings measurements. |
These cases support a real engineering proposition: resolving part identity, service context, and integration failures can improve operations before any sophisticated optimization is introduced. They also complicate attribution. Higher manufacturer parts sales can result from better availability, a broader rollout, or dealers shifting purchases toward the manufacturer. It is not automatically equivalent to lower total service cost or higher profit for every participant.
Likewise, a fill-rate improvement can come from a larger inventory investment. ClearOps’ September 2026 algorithm evaluation reports more than 10% higher cluster fill rates, 27% better attainment of cluster targets, and over 15% fewer order changes. These are useful comparative indicators, but the disclosure lacks the sample, weighting, inventory cost change, and evaluation protocol needed to establish an economic improvement. A reduction in overrides also needs a quality test: fewer corrections are desirable only when the unmodified decisions are better. 6
Software materials and operational boundaries
The public evidence identifies several layers with different responsibilities.
| Layer | Evidence | Ownership and maintenance consequence |
|---|---|---|
| Application development | A 2024 employer vacancy describes Java/Kotlin backends and React micro-frontends written in TypeScript. | Server logic and multiple client components need coordinated interfaces, testing, and upgrades. |
| Replenishment jobs | The support article exposes a Python scheduling function. | Jobs and location configuration form another maintained execution boundary. |
| Cloud operations | ClearOps names AWS hosting in four regions. A current DevOps engineer describes work with Kubernetes, Terraform, Ansible, and CI/CD. | Cloud configuration, orchestration, and deployment automation require continuing operational expertise. |
| Celonis application | Planning and agent features use Celonis context and, for ClearWhy, AWS Bedrock. | The combined service depends on both the planning implementation and the host platform. |
| Earlier PTC arrangement | Servigistics supplies the optimization capability in the 2022 partnership. | The responsibility for a decision depends on which planning product is deployed. |
This is already enough technical mass to make interface ownership and release compatibility material costs. The available evidence does not identify a common database topology or explain how the native, Celonis, and Servigistics planning arrangements share models and configuration. The useful architecture question is which components a particular customer must operate and pay for, and how a changed part definition or planning policy propagates between them.
ClearOps’ connectivity product describes a data hub, an upload portal for unintegrated dealers, connector templates, and a query router. It explicitly recognizes that dealer connectivity is continuing work: new dealers, changed systems, relationship management, and local adoption. Its current figures are more than 100 integrated ERP/DMS systems and 2,500 connected locations. Interface count establishes coverage; it does not remove the maintenance obligation. 3
Security mechanisms
The Trust Center identifies automated monitoring, intrusion and file-integrity checks, and CI/CD scans intended to block high-severity issues before release. It also specifies AWS residency options. These are operational mechanisms, but they do not expose the preventive boundaries that matter for an agent allowed to write orders: tenant isolation, narrowly scoped credentials, permitted actions, egress restrictions, and recovery from compromised execution. 21
The Celonis description promises limits and audit trails. It does not explain where those limits are enforced independently of the agent or what prevents a compromised tool response from producing an unauthorized action. Certification and a record of past writes do not answer that design question. 8
Configuration maintenance, labor productivity, and accountability
ClearOps’ integration layer and inventory policy layer create two continuing workloads. Dealer schemas, part substitutions, and service histories must stay consistent. Planning segments, targets, ordering costs, and exception rules must stay economically appropriate. Automating the calculation does not automate the decisions about what the calculation should mean. 3 5
The AGCO case explicitly includes 24/7 software operations, data maintenance, and support. Genpact describes a combined software and service relationship built over many years. These arrangements can absorb work that customers would otherwise perform, but that work belongs in total ownership cost alongside licenses, dealer onboarding, client staff, and any additional platform subscription. The public cases do not provide that account. 11 17
Workshop and Technician Management illustrates a different productivity improvement: a shared scheduling board, mobile work-order execution with offline use, and speech-to-text documentation. The ERP retains the core record, while ClearOps manages service assignments and execution details. These features can reduce clerical work for technicians. They do not establish that the schedule minimizes downtime cost or coordinates parts and labor under uncertainty. 4
For planning, the relevant post-2022 expectation is much higher than quicker review of the same proposals. The operating model should handle ordinary decisions at scale and make the logic cheaper to improve. ClearOps’ public evidence does not show versioned policy changes, historical replay of decisions, economic regression tests, or rollback of a harmful planning revision. The labor spent maintaining targets and investigating exceptions can therefore remain a substantial part of the operating model. 2 7 10
Accountability should distinguish a data integration failure, a poor economic objective, a miscalibrated forecast, and a client’s departure from an agreed policy. An explanation that persuades a planner to approve an order does not establish that the order was economically sound. ClearOps publishes customer benefits and prospective agent benefits, but no performance acceptance conditions or contractual remedy tied to net economic returns. 11 9 24
Supply chain score
Each subscore ranges from 0 to 10. The overall score is the unweighted average of all 25 subscores.
Supply chain depth
| Criterion | Score |
|---|---|
| Economic framing | 5/10 |
| Decision end-state | 6/10 |
| Conceptual sharpness on supply chain | 6/10 |
| Freedom from obsolete doctrinal centerpieces | 3/10 |
| Robustness against KPI theater | 4/10 |
Average: 4.8/10.
Economic framing — 5/10. Costs matter in the formulation, but the service target does not automatically price the consequences of machine downtime or competing uses of capital. The economic objective remains incomplete. 5
Decision end-state — 6/10. The operating design reaches replenishment and service execution. Confidence is limited by the lack of an end-to-end account of unattended decisions, exceptions, and recovery. 7 12
Conceptual sharpness on supply chain — 6/10. Part identity, installed equipment, and irregular demand receive specific treatment. Joint uncertainty in future demand and supplier receipts remains insufficiently explained. 5 14
Freedom from obsolete doctrinal centerpieces — 3/10. Segments and service targets still determine inventory policy. Moving the review of their outputs into an agent workflow leaves their economic limitations intact. 5 8
Robustness against KPI theater — 4/10. Published improvements lack the definitions and cost changes needed to connect them reliably to profit. Fewer overrides and better target attainment are intermediate measures. 6 11
Decision and optimization substance
| Criterion | Score |
|---|---|
| Probabilistic modeling depth | 5/10 |
| Distinctive optimization or ML substance | 6/10 |
| Real-world constraint handling | 6/10 |
| Decision automation and labor productivity | 5/10 |
| Resilience under real operational complexity | 5/10 |
Average: 5.4/10.
Probabilistic modeling depth — 5/10. The demand mechanism is more specific than generic AI forecasting. Its calibration, dependence structure, and combination with uncertain supply are not inspectable. 5
Distinctive optimization or ML substance — 6/10. The disclosed inventory method provides algorithmic substance. The newer explanation layer should be evaluated separately from improvements to the numerical policy. 5 9
Real-world constraint handling — 6/10. The planning scope goes beyond independent buffers and accommodates operational restrictions. Segment boundaries and the treatment of network-wide economic competition remain limitations. 5 13
Decision automation and labor productivity — 5/10. There is concrete automation, but total labor across configuration, exception handling, and maintenance is unmeasured. The incremental benefit of agents is especially uncertain. 7 9 11
Resilience under real operational complexity — 5/10. Named deployments demonstrate substantial integration requirements. The cases provide little measurement of decision quality during disruption, recovery from partial execution, or how a poor order leads to a tested revision of the planning logic. Explanation alone leaves that improvement work unresolved. 11 12 13
Product and architecture integrity
| Criterion | Score |
|---|---|
| Architectural coherence | 5/10 |
| System-boundary clarity | 6/10 |
| Security seriousness | 4/10 |
| Software parsimony and ownership cost | 4/10 |
| Compatibility with programmatic and agent-assisted operations | 5/10 |
Average: 4.8/10.
Architectural coherence — 5/10. The aftermarket functions fit a coherent operating need. Multiple planning arrangements leave unclear which data, policies, and runtime components are shared. 2 8 15
System-boundary clarity — 6/10. The division between dealer systems, ClearOps, and host platforms is broadly understandable. Transaction ownership, failed writes, and policy synchronization need more precise contracts. 3 4 8
Security seriousness — 4/10. Monitoring and release checks address detected problems. Public descriptions do not establish the isolation and authority boundaries that prevent an agent or compromised dependency from reaching unrelated data and actions. 21 8
Software parsimony and ownership cost — 4/10. Integration coverage creates a continuing operational obligation. The combination of application components, cloud services, and optional partner platforms lacks a published whole-life cost account. 3 19 20
Compatibility with programmatic and agent-assisted operations — 5/10. Programmatic execution is supported, but connectivity alone does not establish safe autonomous decisions or an efficient process for revising their logic. 7 8
Technical transparency
| Criterion | Score |
|---|---|
| Public technical documentation | 5/10 |
| Inspectability without vendor mediation | 4/10 |
| Portability and lock-in visibility | 3/10 |
| Implementation-method transparency | 6/10 |
| Evidence density behind technical claims | 5/10 |
Average: 4.6/10.
Public technical documentation — 5/10. Method explanations and an operational support article expose useful mechanisms. Model validation, APIs, and runtime boundaries receive much less detail. 5 7
Inspectability without vendor mediation — 4/10. The public material supports a functional assessment. Reproducing the inventory evaluation or testing the agent requires access unavailable in the published record. 6 9
Portability and lock-in visibility — 3/10. Retaining existing enterprise systems reduces replacement scope. Portability of planning configuration, historical decisions, and the logic embedded in partner platforms remains unspecified. 3 8
Implementation-method transparency — 6/10. Customer cases and connectivity documentation make the ongoing rollout and data work clear. Staffing, recurring cost, and acceptance criteria remain much less explicit. 3 11 17
Evidence density behind technical claims — 5/10. The inventory and integration claims have concrete support. The stronger claims of autonomous productivity and economic returns rely on prospective benefits and incomplete evaluations. 5 14 24
Vendor seriousness
| Criterion | Score |
|---|---|
| Technical seriousness of public communication | 4/10 |
| Technology roadmap integrity | 5/10 |
| Conceptual sharpness | 5/10 |
| Economic accountability | 4/10 |
| Defensibility in an agentic-software world | 4/10 |
Average: 4.4/10.
Technical seriousness of public communication — 4/10. Concrete technical descriptions coexist with broad AI, hyperscaler, and autonomy claims. The latter outrun the disclosed measurements of numerical quality and operating effort. 17 9 24
Technology roadmap integrity — 5/10. The aftermarket focus and inventory method show continuity and material development. The relationship between successive partner architectures and the newer agentic promise remains weakly documented. 15 5 8
Conceptual sharpness — 5/10. The domain problems are well identified. Treating explanation and trust as the principal bottleneck risks neglecting a wrong objective or inadequately modeled uncertainty. 14 9
Economic accountability — 4/10. Reported operating benefits are insufficient to reconstruct realized net returns. The allocation of responsibility for an economically poor but correctly executed policy is not specified. 11 17 24
Defensibility in an agentic-software world — 4/10. Dealer connectivity and part knowledge can retain value. An agent that explains inherited policies offers a weaker advantage unless it also reduces operating and improvement labor at scale. 3 9 14
Sources
Sources checked on November 4, 2026.
[1] ClearOps platform overview
ClearOps platform overview. Current product perimeter.
[2] Retail Inventory Management
Retail Inventory Management. Native dealer inventory planning and execution flow.
OEM–Dealer Connectivity. Integration mechanisms, scale, and continuing connectivity work.
[4] Workshop and Technician Management
Workshop and Technician Management. Workshop scheduling, mobile execution, and ERP boundary.
[5] Inventory Planning 2.0 developed with TUM
Inventory Planning 2.0 developed with TUM. Published inventory method and research collaboration.
[6] Inventory planning evaluation and CNA nomination
Inventory planning evaluation and CNA nomination. Reported algorithm evaluation and its metrics.
[7] Auto Publish for Replenishment Planning
Auto Publish for Replenishment Planning. Public operational support article; scope is automatic publication into Dealer Management.
[8] Celonis Spare Parts Planning
Celonis Spare Parts Planning. Current Celonis application, MCP connection, and approved ERP execution.
[9] ClearWhy AI explainability prototype
ClearWhy AI explainability prototype. September 2026 explanation prototype and intended productivity mechanism.
[10] From Operator to Orchestrator
From Operator to Orchestrator. Proposed future division of planning work between agents and people.
AGCO customer case. AGCO rollout, reported outcomes, and ongoing support.
Terex customer case. Terex service and replenishment implementation.
[13] Jungheinrich customer case
Jungheinrich customer case. Jungheinrich master-data and multi-supplier integration.
Lippert customer case. Lippert part identity and service configuration work.
[15] PTC and ClearOps service partnership
PTC and ClearOps service partnership. 2022 partner announcement; historical roles of ClearOps and Servigistics.
[16] ClearOps and Celonis partnership announcement
ClearOps and Celonis partnership announcement. 2025 partnership and process-intelligence claims.
[17] Genpact partnership and product history
Genpact partnership and product history. Partner account of origins, product development, and service relationship.
[18] ClearOps Series A financing
ClearOps Series A financing. May 2026 financing announcement.
[19] Full-Stack Developer employer vacancy
Full-Stack Developer employer vacancy. Employer advertisement dated April 9, 2024, retained in the public search index; describes application-development duties.
[20] ClearOps DevOps engineer professional site
ClearOps DevOps engineer professional site. First-person description of current DevOps duties.
[21] Trust Center and security mechanisms
Trust Center and security mechanisms. Current hosting and security-control descriptions.
[22] Barkawi company portfolio
Barkawi company portfolio. Parent portfolio description and SaaS history.
[23] ClearOps company announcement
ClearOps company announcement. Company-issued March 2022 announcement.
[24] Celonis and AWS hackathon results
Celonis and AWS hackathon results. Organizer account of the 2026 competition and prospective prototype benefits.