Go back to Market Research
ToolsGroup supplies planning software for manufacturers, distributors, retailers, and spare-parts businesses. Its Decion platform combines demand forecasting, inventory positioning, replenishment, production planning, retail planning, and pricing. The platform brings together the capabilities previously marketed as SO99+, JustEnough, and Evo, with Decy as its conversational planning assistant. (1)
Overall supply chain score: 4.7/10
| Dimension | Score |
|---|---|
| Supply chain depth | 5.2/10 |
| Decision and optimization substance | 5.6/10 |
| Product and architecture integrity | 4.6/10 |
| Technical transparency | 3.8/10 |
| Vendor seriousness | 4.4/10 |
ToolsGroup’s most substantiated capability is service-driven inventory planning: estimate uncertain demand, calculate inventory/service trade-offs, differentiate stocking targets, and generate replenishment proposals. Named customer accounts describe improvements in inventory, availability, and routine execution. The central weakness is the distance between that documented operating model and Decion’s promise of continuously steering the whole supply chain toward economic outcomes. Neither synchronizing pricing and planning applications nor adding an assistant establishes joint economic optimization. (2, 3, 6, 7, 8)
The May 2026 launch describes autonomous execution and probabilistic intelligence across millions of futures. The more concrete evidence shows planner-directed scenarios, service targets, configuration work, and an assistant that helps investigate forecast exceptions. ToolsGroup has automated meaningful work; it has not demonstrated how much additional work its agentic platform eliminates, or how the expanded suite selects economically coherent actions across its component engines. (2, 21, 23)
ToolsGroup versus Lokad
The practical distinction concerns how a business expresses its priorities and changes the resulting decisions.
ToolsGroup’s inventory documentation starts from a desired aggregate service level, then varies targets by item and location while optimizing additional objectives such as inventory investment, margin, or obsolescence. Pricing adds elasticity models and commercial scenarios. Much of the customer’s work consists of configuring these applications and reconciling their recommendations with business policies. (3, 29)
Lokad documents an operating model in which Supply Chain Scientists maintain customer-specific decision logic, express trade-offs in money, and generate prioritized purchases, allocations, or other executable decisions. The economic assumptions are proposed by Lokad and validated with the customer. The continuing engineering work belongs in the cost of the service. See Lokad’s inventory optimization documentation and the broader design perspective in Introduction to Supply Chain.
Consider two spare parts competing for the same purchasing budget. An availability target does not by itself distinguish a missing part that immobilizes a machine from one with an acceptable substitute. The decision system must represent downtime, substitution, supply delays, and the alternative uses of cash. ToolsGroup documents differentiated stock/service trade-offs, but leaves the formulation of such interacting consequences largely undisclosed. Lokad’s stated approach makes those economic consequences the basis of the decision recipe.
The further question is how a harmful recommendation changes the software. Correcting a service target can suppress a symptom while preserving the faulty model. Holimization calls for revising the objective, constraints, data interpretation, or available actions when operational evidence exposes a mistake. This is a more demanding productivity test than making parameter adjustments easier.
What the inventory engine actually optimizes
Service targets organize the computation
SO99+’s inventory datasheet describes probability forecasts, lead times, order frequencies, and minimum order quantities feeding item/location stock-to-service curves. These curves relate inventory investment to achieved service. The system then differentiates targets across the portfolio and network. ToolsGroup’s criticism of crude ABC segmentation is justified: assigning an identical target to commercially different items discards information. Its own published example instead minimizes inventory subject to a required aggregate service level. (3, 28)
This is a substantive inventory mechanism. Its economic limitation lies in what remains outside the service measure. A higher fill rate may favor cheap, frequent order lines over infrequent but costly failures. Adding a margin objective helps only insofar as it captures the consequences of shortages, substitutions, abandoned baskets, downtime, and future purchases. ToolsGroup advertises profit maximization and cost minimization, including in its October 2025 defense of probabilistic planning, but publishes neither the complete objective nor a worked decision showing how those consequences compete. Without that formulation, a margin objective does not establish whether the system sacrifices the right sales when cash or capacity runs short. (5)
Mitsubishi Electric’s spare-parts case illustrates the actual operating model. A five-month implementation established stock/service curves and inventory corridors, with alerts for unusual demand, overstock, postponed receipts, and transfers. Over three years, the customer reports 30% less stock and an increase in service from 87% to 97%. These results support a useful replacement for its previous manual process, which did not properly accommodate seasonality or slow movers. They do not identify the economic optimum or the full cost of maintaining the new process. (8)
Supply uncertainty needs more than a configurable allowance
The March 2023 SO99+ release includes backup suppliers activated according to shortage risk, changing lead-time assumptions, probabilistic bills of materials, lot expiration, equivalent products, and aggregate capacity constraints. These are relevant operational features, and the presence of a probabilistic bill of materials extends the advertised uncertainty treatment beyond finished-goods demand. The release does not disclose their common stochastic model or the method that chooses among competing actions. (4)
ToolsGroup’s October 2025 explanation says planners supply their knowledge of supplier variability, which stock-to-service curves combine with demand uncertainty. The unresolved issue is how this knowledge becomes a forecast: a fixed allowance, a fitted receipt-date distribution, or a conditional model that changes with supplier congestion and outstanding orders. No calibration results or treatment of dependencies is supplied in that explanation. (5)
For example, a delayed container can affect hundreds of items together. Independent buffers may appear adequate while understating the cost of that shared disruption. A risk-adjusted purchasing decision requires probabilities for receipt dates and demand together, followed through to shortages and corrective actions. Decion’s claim to evaluate millions of futures supplies neither that construction nor evidence that its recommendations remain sound when the assumptions change. (2)
Pricing integration: a connection is not a joint decision
ToolsGroup acquired Evo in September 2023, adding pricing and prescriptive analytics to its planning estate. In December 2025 it announced the integration of SO99+ and PriceAI. The specific mechanisms disclosed were automated data exchange, synchronized forecasts and elasticity models, a one-click transition between applications, and pricing scenarios that compare revenue, margin, and volume while checking supply feasibility. (10, 6)
This integration establishes a route for information to travel. It does not establish that price, purchasing, allocation, and markdown decisions are chosen together under a common objective and shared uncertainty. Sequentially making a price compatible with an existing stock plan can preserve a poor stock plan. A joint calculation must also consider changing the purchase, reallocating stock, or accepting a shortage when those alternatives earn more.
The distinction matters most for substitutable and seasonal products. Discounting an overstocked item can cannibalize a profitable substitute; replenishing it may leave stock after the season; reallocating it may be cheaper than discounting. The price-optimization page advertises elasticity, competitive positioning, inventory constraints, and business rules, but offers no worked example resolving this coupled choice. Pricing features and data exchange are documented; the joint optimization claim remains unestablished. (29)
Automation, configuration maintenance, and productivity
Established automation has identifiable mechanisms
Kotsovolos operates over 100 stores and roughly 20,000 SKUs. ToolsGroup ingests SAP and other data daily, generates replenishment proposals, and applies assortment and planogram rules. Nightly shelf checks can substitute an available model and prevent duplicate replenishment. The reported outcomes include 40% fewer shelf stockouts, a replenishment cycle shortened by a day, and 95% fewer replenishment-related store/head-office communications. (7)
Those changes remove specific manual coordination and ordering errors. The case does not report remaining planner, implementer, or support hours. A reduction in messages is therefore evidence of less communication, not a measurement of total labor productivity. The baseline was spreadsheet-heavy; the relevant next comparison is the operating effort of a well-maintained automated decision system.
Decy accelerates the existing exception workflow
ToolsGroup’s 2026 Decy demonstration follows a planner investigating two coffee-machine product lines. Aggregate indicators conceal weak item forecasts. The assistant identifies problems, examines predecessor-product links, recommends changing a supersession factor, and finds a better predecessor. The planner requests analyses and charts, approves changes, and refreshes the forecasts. The industrial newsletter also emphasizes a redesigned Proposal Manager in SO99+ 2026.2, including editing, aggregation, analysis, and export. (21, 22)
This is assisted configuration repair. Its value depends on how many such cases remain and how much work each consumes. The demonstration does not establish automatic detection and durable correction across the whole assortment. Nor does a better fit to observed demand establish that a changed predecessor relationship will improve future replenishment. Reusable diagnostics, evaluation on later outcomes, and controlled deployment would address the underlying maintenance problem.
ToolsGroup’s October 2026 S&OP article retains a similarly human-centered decision boundary: agents prepare scenarios; people select, review, and submit them. A company may choose this governance, but the work remains part of the operating model. An autonomy claim needs a defined scope of unattended decisions and a measured residual workload. (23)
Configuration creates a continuing ownership cost
Service priorities, supplier assumptions, product substitutions, assortment rules, pricing restrictions, and scenario conventions encode business knowledge. As the business changes, someone must revise them, check their interactions, and verify that yesterday’s workaround is not causing today’s bad recommendation. The acquired applications add interfaces and potentially different interpretations of the same products, prices, and stock positions.
The total cost of ownership therefore includes customer planning labor, partner work, vendor support, configuration regression checks, and infrastructure, alongside the subscription. Neither the Decy demonstration nor the cited customer cases quantifies that complete burden. A useful productivity improvement would turn a discovered failure into a tested correction that prevents its recurrence across the portfolio. Faster investigation of the next exception leaves much of the original cost intact.
Technology history and roadmap integrity
ToolsGroup’s history contains both identifiable engineering continuity and promises that outran their evidence.
| Period | Public claim or delivery | What it establishes |
|---|---|---|
| 2014 | SO99+ 7.3.1 embeds machine learning; Lennox uses seasonal clustering. | A named, bounded forecasting application, rather than evidence of general intelligence. (15) |
| February 2017 | Groover reads social sentiment; external-data models supposedly remove the need for statistical assumptions. | Specific proposed inputs, accompanied by an indefensible description of what learning algorithms can do. (16) |
| September 2017 | SO99+ 8.0 moves to Azure and places its learning layer over Microsoft AI; in-memory S&OP is enhanced. | A dated platform migration and application layer, not proof of a new intelligence capability. (17) |
| 2018 | Blockchain is promoted as an infrastructure breakthrough for trusted planning data. | Trend advocacy; the article announces no ToolsGroup blockchain product. (18) |
| 2019–2021 | Collateral promises 50–90% less planner work and promotes fast in-memory production scenarios. | Automation and rapid scenario analysis were already sales propositions years before Decion. (20, 19) |
| 2024–2025 | Decision Hub launches; SO99+ and PriceAI integration follows. | Collaboration and data exchange broaden the suite; neither announcement demonstrates a common economic solver. (14, 6) |
| 2026 | Decion launches as a self-steering platform; Decy assists exception investigation. | A new platform identity and assistant, with the autonomous scope still insufficiently demonstrated. (1, 21) |
The 2014 evidence is particularly useful for interpreting the later AI vocabulary. Lennox was clustering seasonal demand profiles; another cited deployment forecast promotional effects. These are intelligible learning tasks. In 2017, the promotion shifted to machines understanding external data without statistical assumptions. Learning still requires assumptions about which historical relationships persist, which signals precede the decision, and how unseen cases behave. Deep learning does not remove that dependence. (15, 16)
Groover’s announcement proposed using tweets and sentiment for new-product and promotional forecasts. A May 2017 award announcement repeats the feature without identifying a Groover customer, a held-out comparison, or an economic outcome attributable to the social signal. (32) The absence of a demonstrated outcome leaves that promise unresolved; today’s broader demand-sensing label supplies no answer.
The blockchain article illustrates weak technical judgment more directly. It conflates shared, tamper-resistant records with assurance about the truth of supply-chain data. A ledger cannot establish whether a recorded physical event happened. The production-planning brochure has a milder version of the same marketing problem: keeping computations in memory is presented as the reason for exceptional speed, without workload, hardware, or comparison figures. Neither assertion establishes distinctive engineering. (18, 19)
Finally, the current platform FAQ says the underlying engines and roadmap continue under the Decion brand, with no action required from existing customers. That continuity sets a demanding test for the new autonomy narrative: which decisions or maintenance tasks have actually disappeared? The old workload-reduction percentages cannot establish the assistant’s incremental contribution. (1, 20)
Software bill of materials and architectural integrity
Acquisitions explain the application boundaries
Accel-KKR invested in ToolsGroup in 2018. ToolsGroup subsequently bought Mi9’s former JustEnough demand-management business in 2021, Onera in 2022, and Evo in 2023. These transactions added retail planning, inventory visibility and fulfillment, and pricing capabilities. Onera’s contribution can be followed into the 2023 Fulfill.io launch, which describes inventory-aware order routing. (13, 11, 12, 10, 30)
The relevant integration question is how these applications agree on decisions. SO99+/PriceAI synchronization is documented; common branding does not settle ownership of product hierarchies, elasticity estimates, inventory reservations, or conflicting rules. A pricing recommendation calculated from available stock must be revisited when fulfillment commits that stock elsewhere. The cost and reliability of this coordination depend on the interfaces and failure handling, not the name of the platform.
Public engineering evidence shows a substantial stack
ToolsGroup’s software-engineer advertisement for its Colombia team describes the following technologies. Its cloud documentation independently identifies Azure hosting and customer-environment segregation. The job does not map components to the former SO99+, JustEnough, Onera, and Evo products. (25, 26)
| Layer | Disclosed components | Ownership consequence |
|---|---|---|
| Application languages | Java, Python, C#, JavaScript/TypeScript | Several runtimes and toolchains to maintain. |
| Interfaces | React, Vue.js, GraphQL | Multiple UI frameworks and an additional API layer. |
| Data | PostgreSQL, SQL Server, Redis, Kafka | Relational persistence alongside caching and event transport; data ownership and replay need explicit design. |
| Runtime and hosting | Docker, Kubernetes, Azure AKS, App Services, Functions | Container orchestration plus other hosting models, with deployment, monitoring, and recovery work. |
The stack combines relational storage, caching, and event transport across several application families and hosting patterns. Its component roles are understandable; its overall consolidation is not documented. Each boundary creates work in schema changes, upgrades, observability, incident recovery, and reconciliation. Azure management transfers some operating work to Microsoft; it does not erase its cost.
Public product materials describe integration and synchronized data, but leave the cross-product decision contracts undisclosed: which application may change which decision, what must be recomputed afterward, and how conflicts are resolved. This limits assessment of the expanded platform beyond its component features.
Security assurance emphasizes process over containment
ToolsGroup’s trust page emphasizes ISO/IEC 27001 certification, an oversight committee, scanning, and review procedures. This presentation raises a security-theater concern: compliance provides reassurance without establishing which attacks the architecture prevents. Counting controls can reward administrative activity while leaving unnecessary dependencies and excessive privileges intact. (26)
The page describes administrative gateways separating employee workstations from customer environments, customer-specific encryption keys, and segregated databases or schemas. These identify intended boundaries. Their strength depends on the privileges of application services, operators, and shared infrastructure after a compromise. Encryption at rest cannot constrain a service already authorized to decrypt the data. The reviewed documentation does not map those privileges or establish how a compromised component is contained across the application estate. Decy’s authority to change planning data also needs enforceable limits beyond conversational approvals. (26)
The software inventory above makes this consequential: multiple runtimes, databases, and integration services require more code to be secured and more credentials to be constrained. Monitoring adds components whose own access must be justified. A stronger security account would explain eliminated attack paths, restricted service identities, and recovery that survives malicious administrative actions. It would demonstrate those properties under adversarial testing; a statement that penetration tests occur supplies no such result. (25, 26)
The 2026 cloud security brochure specifies recovery time and recovery point objectives of 24 hours. In the stated worst case, service restoration may take a day and up to a day’s data may be lost. That matters to continuous pricing and fulfillment: replayed recommendations must be reconciled with orders and commitments that survived in transactional systems. The brochure establishes recovery targets but does not describe that business-level reconciliation. “Always-on” decision marketing needs to be read with these operating limits. (27)
Predictive validation and economic accountability
A digital twin must be tested as a forecast
The digital-twin page promises simulation of inventory, logistics, disruptions, and operational interventions. That makes it a conditional forecast of a large interacting system. Up-to-date source data is necessary, but does not show that the predicted effects of a price change, supply delay, or reallocation are accurate. The page supplies no evaluation of scenario predictions against later outcomes, no calibration results, and no analysis of how prediction errors change the chosen actions. (24)
The same discipline applies to demand sensing. The demand-planning page promotes external signals and continuous adaptation. A signal earns its place only if it arrives before the relevant decision and improves results against a baseline using the same information and update frequency. More frequent calculation does not make information actionable after the ordering deadline. (31)
Aggregate forecasting measures also need care. Ackermans’ case describes a concrete failure: forecasting clothing at style level hid shortages in particular sizes, leaving the wrong stock in stores. The reported improvement followed visibility at the finer SKU level. This is precisely why a flattering aggregate accuracy figure cannot validate the decisions underneath it. The published 97% figure in Mitsubishi’s case is a service level, not forecast accuracy. (9, 8)
Returns need a defined failure condition
The named customer cases report useful operating results, but do not calculate returns net of licenses, implementation, ongoing configuration, customer labor, and partner effort. Inventory released is also not the same quantity as recurring profit: it can free working capital once while yielding smaller continuing carrying-cost savings. A credible return calculation must keep these effects separate.
ToolsGroup’s October 2026 product-leadership article locates the principal obstacle to agentic S&OP in trust and adoption, saying the technology is ready enough. It also advocates measuring overrides and learning from missed outcomes. That position needs a corresponding test of the product itself: when a customer supplies the agreed data and follows the agreed process, which missed economic result makes the implementation unsuccessful? Without that boundary, adoption can become an explanation for failure that leaves the technology proposition untouched. The article does not establish such a boundary. (23)
The brochure quantifies recovery objectives and references an availability SLA. The performance claims leave economic failure, responsibilities, and remediation undefined. (27)
Detailed supply chain score
The scores below are editorial assessments of documented capabilities and their limitations. Each dimension is the mean of its five criteria; the overall score is the mean of all 25 criteria.
Supply chain depth
| Criterion | Score |
|---|---|
| Economic framing | 4/10 |
| Decision end-state | 7/10 |
| Conceptual sharpness on supply chain | 6/10 |
| Freedom from obsolete doctrinal centerpieces | 4/10 |
| Robustness against KPI theater | 5/10 |
Average: 5.2/10.
Economic framing — 4/10. ToolsGroup names margin and cost objectives, but its disclosed inventory mechanism remains organized around service targets. The treatment of interacting monetary consequences is too opaque to establish system-wide economic optimization. (3, 5)
Decision end-state — 7/10. Daily replenishment, supplier proposals, transfers, and fulfillment routing are operational outputs. The Kotsovolos account links calculations to store execution. Pricing and inventory decisions are not shown to share one economic selection process. (7, 30)
Conceptual sharpness on supply chain — 6/10. Intermittency, multi-echelon inventory, substitutions, and uncertain component use receive explicit treatment. The broader assertions about modeling every future supply-chain state lack equivalent technical detail. (4)
Freedom from obsolete doctrinal centerpieces — 4/10. Differentiated stock targets improve upon ABC classes, but target service, inventory corridors, exception review, and scenario arbitration still structure the documented process. (8, 28)
Robustness against KPI theater — 5/10. Customer cases expose concrete availability and inventory changes. Forecast resolution, residual workload, and net economic attribution remain insufficiently specified for the broader performance promises. Ackermans’ size-level shortages show why aggregate measures are inadequate. (9)
Decision and optimization substance
| Criterion | Score |
|---|---|
| Probabilistic modeling depth | 5/10 |
| Distinctive optimization or ML substance | 5/10 |
| Real-world constraint handling | 7/10 |
| Decision automation and labor productivity | 6/10 |
| Resilience under real operational complexity | 5/10 |
Average: 5.6/10.
Probabilistic modeling depth — 5/10. Demand distributions feed inventory decisions, and supply variability is represented in the advertised stock/service calculation. Joint receipt-date forecasting, dependencies, and calibration are not disclosed sufficiently to validate the broader risk-adjusted optimization claim. (5)
Distinctive optimization or ML substance — 5/10. Seasonal clustering and proprietary stock/service relationships identify actual computational mechanisms. The algorithmic contribution behind current prescriptive and agentic claims is much less inspectable. (15, 3)
Real-world constraint handling — 7/10. Supplier alternatives, order restrictions, lot expiration, capacity, and probabilistic bills of materials address genuine operating constraints. Public documentation does not show how fidelity and runtime behave when these constraints interact at scale. (4)
Decision automation and labor productivity — 6/10. The retail deployment removes specific replenishment and coordination tasks. Remaining labor across customers, ToolsGroup, and partners is unmeasured, and the new assistant’s incremental productivity is unestablished. (7, 21)
Resilience under real operational complexity — 5/10. Deployed inventory processes accommodate operational variation. The documented improvement workflow leans on configuration changes and human analysis; a repeatable process for testing and revising the decision formulation is insufficiently exposed. (8, 22)
Product and architecture integrity
| Criterion | Score |
|---|---|
| Architectural coherence | 4/10 |
| 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: 4.6/10.
Architectural coherence — 4/10. The suite joins identifiable acquired application families. Data synchronization is documented; common decision semantics and failure handling across those families are not. (6, 11, 12)
System-boundary clarity — 6/10. The planning and fulfillment roles relative to ERP and store systems are understandable. Recovery across those boundaries becomes less clear when recommendations and external commitments must be reconciled after an outage. (7, 27)
Security seriousness — 4/10. The documents identify administrative boundaries and recovery targets, but leave component privileges, compromise containment, and agent authorization insufficiently exposed. Certification and procedural claims earn no security credit; the substantial software estate needs a demonstrated account of how attacks are prevented and their damage limited. (26, 27)
Software parsimony and ownership cost — 4/10. Multiple application families and a broad infrastructure stack create maintenance obligations. The published customer outcomes omit the continuing configuration and support effort needed to assess the total cost. (25)
Compatibility with programmatic and agent-assisted operations — 5/10. Automated ingestion, data exchange, and assistant-triggered actions are documented. There is little public evidence of customer control over versioned decision logic, portfolio-wide repair, or reusable testing. (6, 22)
Technical transparency
| Criterion | Score |
|---|---|
| Public technical documentation | 4/10 |
| Inspectability without vendor mediation | 3/10 |
| Portability and lock-in visibility | 3/10 |
| Implementation-method transparency | 5/10 |
| Evidence density behind technical claims | 4/10 |
Average: 3.8/10.
Public technical documentation — 4/10. Datasheets explain stock/service mechanics, and security material describes administrative policies. Model definitions, current decision interfaces, and reproducible numerical examples are much thinner. (3, 26)
Inspectability without vendor mediation — 3/10. Buyers can inspect brochures and demonstrations, but cannot independently reconstruct how the expanded platform evaluates uncertain economic trade-offs. Scenario interfaces do not expose that calculation. (14, 24)
Portability and lock-in visibility — 3/10. Data exchange and proposal export establish some access to inputs and outputs. The reviewed documents do not explain how a customer could export executable forecasting, optimization, and configuration logic for use elsewhere. (6, 22)
Implementation-method transparency — 5/10. Mitsubishi and Kotsovolos describe data preparation, integration, rules, and operational changes. The ongoing maintenance workload and dependencies on implementers remain largely unquantified. (8, 7)
Evidence density behind technical claims — 4/10. Specific historical releases and customer mechanisms support the inventory capability. The digital twin, joint optimization, and broad autonomy claims extend beyond the published validation. (24, 6)
Vendor seriousness
| Criterion | Score |
|---|---|
| Technical seriousness of public communication | 4/10 |
| Technology roadmap integrity | 4/10 |
| Conceptual sharpness | 5/10 |
| Economic accountability | 4/10 |
| Defensibility in an agentic-software world | 5/10 |
Average: 4.4/10.
Technical seriousness of public communication — 4/10. Documentation of real inventory mechanisms coexists with claims that learning removes statistical assumptions and that broad autonomy is already available. These formulations obscure the engineering limitations a buyer needs to understand. (16, 2)
Technology roadmap integrity — 4/10. Cloud migration, named ML applications, and acquired product integration are traceable. Social-sensing outcomes and the incremental delivery behind successive automation promises remain unresolved. (17, 16, 20)
Conceptual sharpness — 5/10. ToolsGroup recognizes demand uncertainty and differentiated inventory needs. Its operational definition of success still shifts between service targets, scenario selection, and a much broader economic steering claim. (28, 23)
Economic accountability — 4/10. Named cases report operating gains, but the public proposition leaves net returns, bounded customer obligations, and responsibility for an unsuccessful economic result undefined. Emphasizing adoption does not discharge those obligations. (9, 23)
Defensibility in an agentic-software world — 5/10. Inventory algorithms, operational integrations, and deployment experience give ToolsGroup more substance than a conversational interface. Decy’s disclosed workflow does not establish a durable advantage in eliminating work or improving decision logic. (15, 21)
Overall: 118/25 = 4.72, rounded to 4.7/10.
Sources
Sources accessed November 4, 2026. Links use verified Wayback snapshots where available. Customer outcomes are attributed to the published accounts.
ToolsGroup. Current product page; accessed November 4, 2026.
ToolsGroup. May 27, 2026. Launch scope and autonomous-execution claims.
[3] Inventory optimization datasheet
ToolsGroup. 2020, pp. 2–3. Objective options, stock/service curves, and planning inputs.
[4] SO99+ supply-planning enhancements
ToolsGroup. March 7, 2023. Supplier alternatives, probabilistic bills of materials, and constraints.
[5] Probabilistic Planning: The Supply Chain Advantage
ToolsGroup. October 21, 2025. Supply variability, objectives, and operating claims.
[6] SO99+ and PriceAI integration
ToolsGroup. December 2, 2025. Data exchange, navigation, and pricing scenarios.
[7] Kotsovolos customer account
ToolsGroup. Undated; accessed November 4, 2026. Daily integration, nightly replenishment, and reported results.
[8] Mitsubishi Electric customer account
ToolsGroup. Undated. Implementation, inventory corridors, and three-year results.
[9] Ackermans customer account
ToolsGroup. Undated. Style-versus-size forecasting and reported stock/sales changes.
ToolsGroup. September 27, 2023. Pricing and prescriptive-analytics acquisition.
[11] Mi9 demand-management acquisition
ToolsGroup. November 8, 2021. Former JustEnough application perimeter.
ToolsGroup. June 7, 2022. Retail execution and inventory-data expansion.
ToolsGroup. May 3, 2018. Investment history.
ToolsGroup. May 6, 2024. Collaborative scenarios and decision alignment.
[15] Embedded machine learning in SO99+
ToolsGroup. March 2014. Version 7.3.1 and named seasonal-clustering deployment.
[16] Groover and other machine-learning applications
ToolsGroup. February 15, 2017. Social signals, external data, and statistical-assumption claims.
ToolsGroup. September 13, 2017. Azure, Microsoft AI, and in-memory S&OP.
[18] Blockchain and planning data
ToolsGroup. March 29, 2018. Claims about trust, authentication, and shared records.
[19] Production-planning datasheet
ToolsGroup. 2021. In-memory simulation and interactive scheduling.
[20] Machine learning for demand planning infographic
ToolsGroup. 2019. Self-adaptation and 50–90% planner-workload claim.
[21] Decy demonstration: retail newsletter
ToolsGroup. 2026. Publicly indexed transcript; the live page now requires a subscription.
[22] Industrial product newsletter
ToolsGroup. October 2026. Publicly indexed SO99+ 2026.2 release details; the live page now requires a subscription.
ToolsGroup. October 15, 2026. Product-leadership account of adoption, scenarios, and human approval.
[24] Digital supply chain twin
ToolsGroup. Current product page; accessed November 4, 2026. Simulation and predictive claims.
[25] Software-engineer advertisement
ToolsGroup / JobLeads. Undated employer advertisement, reproduced by JobLeads. Colombia engineering team and technology list.
ToolsGroup. Current security documentation; accessed November 4, 2026.
ToolsGroup. 2026, especially p. 6. Recovery targets, backups, and availability commitments.
[28] Critique of ABC inventory classification
ToolsGroup. March 15, 2023. Stock/service optimization and aggregate service targets.
ToolsGroup. Current product page; accessed November 4, 2026. Elasticity, inventory constraints, and commercial rules.
ToolsGroup. July 2023. Fulfillment decisions and order-management integration.
ToolsGroup. Current product page; accessed November 4, 2026. Forecasting, external signals, and demand sensing.
[32] Groover in a technology-award announcement
ToolsGroup. May 16, 2017. Follow-up promotion of social sensing.