Review of Optilogic, Supply Chain Design Technology Vendor

By Léon Levinas-Ménard

Last updated: December, 2025

Optilogic is a privately held, Ann Arbor–based software vendor focused on supply chain network design and “always-on” digital twins, built around its flagship cloud-native platform Cosmic Frog and a structured data layer known as Anura; the platform combines several distinct optimization engines (Neo for mixed-integer programming, Throg for simulation, Dendro for inventory strategy, Triad for greenfield analysis, Hopper for transportation design), wrapped in a scenario-driven modeling environment and increasingly augmented with generative AI interfaces such as Leapfrog AI (a natural-language-to-SQL assistant) and DataStar (an AI-based data transformation/workflow layer).¹²³⁴⁵ Optilogic was founded in 2018 and is led by CEO Don Hicks, who previously founded LLamasoft; the company positions itself squarely in the supply chain design segment rather than day-to-day execution planning and has raised at least $53M in equity funding, including a $40M Series B closed in April 2025 led by NewRoad Capital Partners.⁶⁷⁸⁹¹⁰ Optilogic has expanded its capabilities via the 2024 acquisition of INSIGHT, the firm behind the long-standing SAILS network design platform, and publicly references customers such as General Motors and Henkel Adhesive Technologies using Cosmic Frog for large-scale digital twin and network optimization initiatives.¹¹¹²¹³¹⁴¹⁵¹⁶¹⁷¹⁸

Optilogic overview

Company profile and positioning

Optilogic Inc. is described in analyst listings and company databases as a supply chain design software company headquartered in Ann Arbor, Michigan, founded in 2018.⁶⁷ Public profiles emphasize that its main product is Cosmic Frog, a cloud-native platform for modeling, optimizing, and simulating supply chains, with a focus on strategic and tactical network design rather than operational replenishment or execution.¹¹⁹

Market-intelligence sites and competitor mappings consistently categorize Optilogic in the “supply chain network design / digital twin” niche alongside tools like GAINS, Coupa Supply Chain Design & Planning (ex-LLamasoft), and others, rather than in traditional APS (Advanced Planning & Scheduling) or day-to-day demand-planning segments.¹⁹ This is broadly consistent with the company’s own messaging, which stresses decisions such as facility location, flows, inventory strategies by echelon, and transportation policies, and the ability to run large numbers of “what-if” scenarios across complex networks.¹²⁸²⁰

Cosmic Frog is marketed as an “always-on” design platform capable of combining optimization, simulation, and risk analysis in a single model, with built-in support for CO₂ and resilience metrics in addition to cost and service.¹²²⁰ This positioning is important: compared to execution-focused vendors, Optilogic is primarily about structural decisions and scenario analysis (e.g., how many DCs, where, which flows, what policies), not about daily order-level automation.

History, funding, and acquisitions

Most public references agree that Optilogic was founded in 2018; several mention that CEO Don Hicks previously founded LLamasoft (a network design vendor acquired by Coupa in 2020 for roughly $1.5B), which provides important context for Optilogic’s focus and positioning.⁶⁷¹³

Funding information from VC-news aggregators and company databases indicates at least three rounds of external funding: early seed/Series A rounds and a Series B in April 2025.⁸⁹¹⁰²¹ The 2025 Series B round is reported at $40M, led by NewRoad Capital Partners with participation from MK Capital, Mercury, and other investors, bringing total disclosed funding to approximately $53M.⁸⁹¹⁰²¹ This level of funding is consistent with a growth-stage SaaS vendor: large enough to finance aggressive product development and go-to-market, but still far from the scale (and constraints) of the largest enterprise players.

In January 2024 Optilogic announced the acquisition of INSIGHT, maker of the SAILS supply chain design platform.¹¹ The press release states that INSIGHT’s technology and team would be integrated into Optilogic to accelerate delivery of “Supply Chain Design as a Service” and to expand professional services capacity.¹¹¹² Independent coverage in DBusiness, Supply & Demand Chain Executive, and Outsource Accelerator corroborates the acquisition and frames it as a transition of a decades-old network design player (INSIGHT/SAILS) into the newer cloud-native environment of Cosmic Frog.¹³¹⁴²²⁵

There is no evidence of Optilogic itself being acquired as of late 2025; the firm remains independent.

Customer references and market footprint

Optilogic names several recognizable brands in public materials, the most detailed being:

General Motors (GM) – GM describes using Cosmic Frog to create a supply chain digital twin for its global logistics network, modeling flows for more than 3 million vehicles and over 300 million part numbers.¹⁵¹⁶²³²⁴ Optilogic’s case study claims the platform enables GM to stress-test scenarios (disruptions, policy changes) and analyze cost, service, and emissions trade-offs, while a SupplyChainDive article confirms GM’s use of Optilogic for improving end-to-end visibility and scenario analysis.¹⁵¹⁶²³²⁴
Henkel Adhesive Technologies – A case study and third-party coverage indicate Henkel leverages Cosmic Frog to redesign and stress-test its global supply chain, with specific emphasis on CO₂ emissions and resilience metrics alongside cost and service.¹⁷¹⁸

Employee-count estimations from contact-enrichment tools suggest on the order of dozens of staff (roughly 40–100), with most located in North America.²⁵ This is consistent with a specialized, not-massive SaaS vendor. Salary and job-role data on Glassdoor and Salary.com show typical roles (software engineers, optimization analysts, data scientists, professional services consultants) but do not materially change the technical picture.¹⁰¹²⁴

In short, Optilogic appears commercially established but not large: credible named references in complex environments (GM, Henkel), one notable acquisition (INSIGHT), and enough funding and staff to sustain ongoing R&D, but nowhere near the scale of the largest APS vendors.

Optilogic vs Lokad

Although both Optilogic and Lokad operate in the broad space of “analytics for supply chains,” their scope, architecture, and decision focus are materially different, which matters when comparing them.²⁶²⁷²⁸

Decision horizon and problem class
- Optilogic is fundamentally a network design and digital twin platform. Cosmic Frog is built for structural and policy questions: facility locations, flow paths, modal choices, inventory strategies by echelon, transportation policies, and long-range scenario planning (including CO₂ and resilience metrics).¹²²⁰ Day-to-day order release, detailed production scheduling, and operational replenishment are not the primary focus.
- Lokad, by contrast, is primarily an operational decision engine: it focuses on probabilistic demand forecasting, daily replenishment, allocation, and, where relevant, production scheduling and pricing, implemented via its domain-specific language Envision and custom optimization paradigms (probabilistic forecasts, Stochastic Discrete Descent, Latent Optimization) in a single forecasting–optimization pipeline.²⁶²⁸²⁹³⁰³¹ Lokad’s own technology overview explicitly presents a generational roadmap from classic forecasts to quantile forecasts, probabilistic forecasts, deep learning, differentiable programming, Stochastic Discrete Descent (2021), and Latent Optimization (2024), all geared toward supply chain decision optimization.²⁶
Modeling interface and extensibility
- Optilogic exposes a modeling GUI and configuration-based model definitions: users define models, data tables (backed by the Anura schema), and scenarios through a UI, selecting engines and configuring parameters; Leapfrog AI adds a natural-language layer over Anura, translating user prompts into SQL queries and scenario operations.²³³²³³ The platform is extensible within the repertoire of engines Optilogic provides (MIP, simulation, inventory, greenfield, routing) but does not offer a general-purpose programming language for arbitrary computations.
- Lokad exposes a general DSL (Envision) where data transformations, probabilistic modeling, and optimization objectives are written as code dedicated to predictive optimization of supply chains.²⁸¹³¹⁴ Official documentation describes Envision as Lokad’s domain-specific language for supply chain analytics, with most platform capabilities delivered through this DSL.²⁸¹³ Lokad’s architecture pages further explain that Envision scripts are compiled and executed on a distributed virtual machine (“Thunks”) inside a multi-tenant SaaS environment.²⁷¹⁴ This makes Lokad closer to a programmable analytics platform where bespoke models are written as programs, at the cost of a steeper learning curve and a heavier reliance on “supply chain scientists.”
Treatment of uncertainty and risk
- Optilogic incorporates risk primarily at the scenario and scorecard level: Cosmic Frog computes metrics across dimensions such as cost, service, risk/resilience, and sustainability; models can be simulated under different assumptions via Throg, and results are summarized in “risk ratings” across several categories.¹²²⁰ The emphasis is on scenario comparison and multi-criteria scoring.
- Lokad centers on probabilistic forecasting and stochastic optimization, explicitly modeling full demand distributions and using them inside optimization algorithms.²⁶²⁹³⁰ Lokad’s public material describes a progression from quantile forecasts to probabilistic forecasts and then to integrated optimization, with probabilistic forecasts estimated at SKU level and used directly to compute prioritized replenishment decisions.²⁶³⁰³⁴ Demand- and inventory-related FAQs explain that probabilistic forecasts drive economic trade-offs between stock-out risk and carrying cost, and that safety stocks and service levels are optimized automatically rather than via fixed formulas.²⁹³⁵³⁶ This makes uncertainty a first-class numerical input, not just a qualitative scenario dimension.
Data strategy and AI usage
- Optilogic has recently invested heavily in AI-enabled data access and data transformation: Leapfrog AI provides text-to-SQL interaction with Anura, effectively “democratizing” model interrogation; DataStar is presented as an “agentic AI” data orchestration layer, meant to automate ingestion, transformation, and publishing of data into Cosmic Frog.⁴⁵³⁷³⁸³⁹³²³³ These features are primarily about usability and data plumbing (making it easier to get data in and insights out) rather than about fundamentally new optimization mathematics.
- Lokad’s AI investment is skewed toward numerical modeling and optimization: probabilistic forecasting, deep learning, differentiable programming, and stochastic search are core to its forecasting–optimization stack rather than to its user interface.²⁶³⁰³¹ Lokad also explicitly frames its methodology as “quantitative supply chain,” insisting that optimization be driven by economic drivers such as holding cost, stockout penalty, spoilage, and obsolescence, which are embedded in Envision programs.³⁶³⁴⁴⁰ In other words, Lokad’s AI is concentrated in the math that turns data into decisions; Optilogic’s AI is, at least currently, more visible in the data and UX layers that surround its engines.
Commercial engagement model
- Optilogic sells Cosmic Frog as a tooling-heavy design platform, and via the INSIGHT acquisition also offers “design as a service” through its team (outsourced modeling, scenario building). This is inherently project-centric: build a model, run scenarios, interpret trade-offs, then feed results into other systems.¹¹¹²¹³¹⁴
- Lokad delivers a continuous, code-based optimization service: its supply chain scientists use Envision to encode the client’s business, and the platform produces daily or weekly prioritized lists of operational decisions—purchase orders, allocations, production plans, pricing moves—optimizing financial outcomes under uncertainty.²⁶²⁸⁴⁰ Lokad’s technology and solution pages explicitly describe automated “predictive optimization” for routine challenges such as purchasing, production planning, stocking, and pricing, with recommendations expressed as financially ranked actions rather than as one-off studies.³⁸⁴⁰

In practice, a large organization might reasonably use both: Optilogic for strategic network design and digital twin scenario work; Lokad for operational forecasting and replenishment optimization. They overlap only partially: Optilogic has inventory and transportation engines that can touch operational policy design, and Lokad’s Latent Optimization reaches into planning/scheduling; but their respective center-of-gravity is different.

From a technology assessment perspective, Optilogic’s architecture is contemporary and credible for its design remit (cloud-native multi-engine platform with structured data layer and LLM-based assistants), while Lokad’s stack is more idiosyncratic and code-centric, optimized for high-frequency probabilistic optimization rather than scenario design.¹²³²⁶²⁷²⁸ Which is preferable depends entirely on whether the primary problem is “What should my network look like?” (Optilogic’s strength) or “What exactly should I order/allocate/produce today under uncertainty?” (Lokad’s strength).

Product and architecture

Cosmic Frog and its engines

Cosmic Frog is Optilogic’s central product: a multi-tenant SaaS application where users build supply chain models, define data tables mapped into the Anura schema, and run different engines to analyze scenarios.¹²³ The documentation outlines several built-in engines:¹²³

Neo – a mixed-integer optimization engine used for classic network design (facility location, flows, capacities, policies, etc.). Users configure objective functions (typically cost minimization) and constraints (capacity, service levels, etc.) and obtain optimized network designs.²
Throg – a simulation engine for dynamic behavior over time, capable of running scenarios under demand variability, lead-time assumptions, and policy changes, producing time-series metrics such as service levels and inventory trajectories.²
Dendro – an inventory planning engine oriented toward inventory strategy design (e.g., where to stock, at what levels, by echelon), complementing Neo’s structural optimization with policy-focused analysis.²³
Triad – a greenfield/center-of-gravity engine used to rapidly identify candidate facility locations before running more detailed Neo models.²⁸
Hopper – a transportation optimization engine for routing and flow decisions.²

All engines share the same Anura data model, which is a Postgres-based schema capturing entities like facilities, customers, products, lanes, demand, costs, and constraints; the “Anura 2.8 Outputs” document details standardized result tables for each engine type, suggesting a reasonably mature internal API and data-contract approach.³

The combination of several engines linked through a shared schema is technically orthodox but solid for network design: MIP engines for structural decisions, greenfield to narrow candidate sets, simulation to test dynamics, and specialized inventory and routing engines to refine policies.

Anura and Leapfrog AI: data access layer

Anura acts as the persistence layer and logical schema behind Cosmic Frog models. Users can upload data to tables, define transformations, and then use either the UI or SQL to query results. The Leapfrog AI module extends Anura with a natural-language interface: users can type prompts such as “Show me the top 10 lanes by transportation cost last year” or “Create a scenario increasing demand by 10% in Europe” and Leapfrog translates these prompts into SQL and scenario operations.³²³³

Documentation indicates that Leapfrog AI stores conversations and generated SQL, and that users can inspect and edit the generated queries—so this is effectively a Text2SQL assistant with scenario macros, not a hidden black-box.³²³³ From a technical standpoint, the sophistication lies less in the SQL side (standard) and more in prompt-engineering and the mapping from a user’s business language into the Anura schema; this is exactly the kind of problem for which LLMs are well suited, and Optilogic’s design matches industry patterns.

DataStar: agentic data transformation layer

In late 2024 / 2025 Optilogic announced DataStar, described as an “agentic AI” platform that automates data preparation and workflow orchestration for supply chain design.⁴⁵³⁷³⁸³⁹ Press releases and coverage state that DataStar uses AI agents to connect to different data sources, transform and clean data, and publish it into Anura and Cosmic Frog, with the goal of making design “always on” instead of episodic.⁴⁵³⁷³⁸³⁹

Third-party tech news articles generally echo Optilogic’s positioning: DataStar is pitched as a way to replace manual ETL scripts and spreadsheets with a more automated, AI-assisted layer that keeps models fed with live or frequently refreshed data.⁵³⁸³⁹ What is less clear (and not well documented publicly) is to what extent DataStar’s “agents” are merely wrappers around common integration tools (e.g., scheduled connectors + transformation rules) vs. genuinely adaptive AI agents capable of autonomously adjusting to schema drift or semantic changes. Current descriptions are high-level, and there is no technical white paper describing learning algorithms, error-detection strategies, or how agent behavior is validated. For now, the most cautious interpretation is that DataStar is a modern, AI-assisted ETL/workflow layer whose internal sophistication cannot be independently assessed beyond marketing copy and high-level diagrams.

Technology stack and engineering signals

Optilogic does not publicly document its full technology stack (languages, frameworks, deployment architecture). However, a few indirect signals are visible:

An AI/ML engineer job description mentions “revolutionizing supply chain network design with our cloud-native solutions” and emphasizes handling “enterprise data at scale,” with Cosmic Frog requiring “no IT footprint”—language consistent with a multi-tenant SaaS platform deployed on hyperscaler infrastructure.⁴¹
A full-stack developer’s personal site lists TypeScript, Python, React, and modern web technologies as their stack while working at Optilogic, indicating at least some services and the front-end are implemented with typical modern web tooling.⁴²

Given the presence of MIP optimization and simulation engines, it is highly likely (though not publicly confirmed) that commercial solvers (e.g., Gurobi, CPLEX) or industrial-grade open-source solvers underlie Neo, and that Throg uses discrete-event or time-stepped simulation frameworks; however, this is inferential and not explicitly stated in documentation, so any stronger claim would be speculation. No information is available on whether Optilogic uses container orchestration (e.g., Kubernetes), what its multi-tenancy isolation model is, or how it handles cost scaling for large scenario batches.

From a critical standpoint, Optilogic’s technical story is consistent but not deeply transparent: we have clear evidence of the product’s functional architecture (engines + schema + AI layer), but relatively little about the implementation details that would allow fine-grained benchmarking against state-of-the-art numerical or architectural practices.

Deployment model and usage in practice

Public case studies and collateral portray a deployment and usage model typical for network design platforms:

Modeling and baseline build – Users (often supported by Optilogic consultants) ingest data into Anura (customers, facilities, products, costs, historical flows) and configure a baseline Cosmic Frog model using Neo and, optionally, Dendro and Hopper.¹²³
Scenario building and simulation – Multiple candidate scenarios are created (e.g., facility closures, sourcing changes, demand shifts), solved under Neo, and then tested in Throg to simulate operational dynamics (inventory oscillations, service performance, etc.).²²⁰
Risk and CO₂ analysis – Cosmic Frog exposes scorecards summarizing performance across cost, service, risk/resilience, and sustainability, enabling decision-makers to visualize trade-offs.¹²⁰
Iterative refinement and decision support – The model is iteratively refined, potentially with Leapfrog AI helping non-technical stakeholders query and visualize results; final decisions are then implemented via other systems (ERP, TMS, WMS, etc.).³²³³²⁰

The GM case suggests that, once built, the digital twin can be run repeatedly to test new policies and disruptions, indicating a shift from “one-off network studies” toward continuous design.¹⁵¹⁶²³²⁴ Henkel’s case similarly emphasizes repeated “stress testing” of the network against demand and sourcing shocks, with sustainability metrics included.¹⁷¹⁸

Unlike operational planning tools, there is no indication that Cosmic Frog directly executes decisions (e.g., generating POs automatically). Instead, it acts as an analytical layer whose outputs—optimized flows, inventory policies, network structures—are interpreted and then applied in downstream systems. This is a standard separation between design and execution, and consistent with the vendor’s stated remit.

Assessment of analytics and optimization capabilities

Strengths: coherent multi-engine design platform

From an optimization perspective, the multi-engine architecture is a sensible and fairly modern response to the heterogeneous nature of network design problems:

Structural network design is best handled by MIP solvers (Neo).
Temporal dynamics, policy interactions, and stochastic effects are better explored via simulation (Throg).
Inventory strategies and decoupling-point placement benefit from specialized inventory models (Dendro).
Greenfield site selection is computationally simpler but benefits from dedicated tooling (Triad).
Transportation decisions can be layered through routing and flow engines (Hopper).

This decomposition aligns with academic and industrial practice. The presence of a standardized output schema (Anura 2.8) further indicates that Optilogic has industrialized its modeling pipeline, making it easier to connect engines and build repeatable analyses.³

The risk metric framing (cost, service, risk, sustainability) is also aligned with contemporary expectations: post-COVID, extreme events and sustainability targets have become first-class design criteria, and Cosmic Frog’s UX showcases multi-axis comparisons rather than single-objective outputs.¹²⁰

AI claims: mostly around access and data, not new math (so far)

Optilogic’s prominent AI claims currently center on Leapfrog AI and DataStar. Based on available documentation and coverage:

Leapfrog AI appears to be a Text2SQL + scenario automation assistant over Anura and Cosmic Frog; this is useful and technically non-trivial, but it does not fundamentally change the underlying optimization algorithms. It mainly lowers the barrier for interacting with data and models.³²³³
DataStar is positioned as an AI agent-based data transformation/orchestration layer. Public material does not describe, for example, how agents learn mappings, detect anomalies, or adapt to changing upstream data; the safest reading is that it is an ETL/workflow product with LLM-driven interface and some heuristics around mapping and schema alignment.⁴⁵³⁷³⁸³⁹

In both cases, the AI component is on the periphery (data access and UX), not clearly in the core optimization mathematics. This is not a criticism per se—most practical value in network design comes from good models and good data rather than exotic algorithms—but it does mean that current AI branding should be interpreted cautiously. Without technical papers or benchmarks, one should treat “agentic AI” as a usability improvement, not a proven step-change in optimization quality.

Gaps and unknowns

Several aspects remain opaque:

Solver stack – no public confirmation of which MIP or simulation engines are used, or how they are tuned for large-scale, multi-scenario workloads.
Scalability and performance – case studies demonstrate non-trivial scale (GM’s 3M vehicles / 300M part numbers), but we lack independent metrics on solve times, convergence properties, or scenarios-per-day limits.¹⁵¹⁶²³²⁴
Uncertainty modeling – while simulation and “risk ratings” indicate some treatment of uncertainty, there is no public evidence of full probabilistic modeling (e.g., demand and lead-time distributions calibrated from historical data) integrated directly into optimization in the way probabilistic vendors do; uncertainty treatment appears concentrated in scenario variability and simulation, which is standard but not cutting-edge.
Open extensibility – there is no general-purpose scripting or DSL exposed, so extending Cosmic Frog beyond the provided engines likely depends on Optilogic’s roadmap or custom professional services.

Overall, the technical maturity of Optilogic’s platform appears solid and up-to-date relative to mainstream network design practice: a cloud-native, multi-engine, schema-based design system with integrated simulation and growing AI-assisted data access. It is harder to substantiate claims of being “state-of-the-art” in algorithmic innovation without more transparency or independent benchmarks, but nothing in the available material suggests outdated or simplistic methods either.

Commercial maturity

Combining funding, headcount signals, and customer references:

Stage – With >$50M in funding, a well-known founder, and a substantive acquisition (INSIGHT), Optilogic is clearly beyond early-stage startup, best described as a growth-stage, specialized design platform vendor.⁸⁹¹⁰²¹¹¹¹²¹³¹⁴
Client footprint – The presence of GM and Henkel as named references, each with non-trivial complexity, indicates credible enterprise adoption in automotive and chemicals manufacturing.¹⁵¹⁶²³²⁴¹⁷¹⁸
Ecosystem – Analyst listings and competitor maps place Optilogic among recognized network design tools, often as a newer, cloud-native alternative to legacy packages.¹⁹

That said, there is no indication of hundreds of customers or a very large services ecosystem. Prospective buyers should treat Optilogic as a focused specialist rather than as a one-stop APS suite vendor. For strategic design, this specialization can be a positive; for companies seeking to consolidate all planning into a single vendor, it implies a multi-vendor landscape.

Conclusion

What does Optilogic’s solution actually deliver? In concrete terms, Optilogic delivers a cloud-native supply chain design platform (Cosmic Frog) with multiple optimization and simulation engines (Neo, Throg, Dendro, Triad, Hopper) integrated via a relational data schema (Anura), plus emerging AI layers (Leapfrog AI for Text2SQL-style interaction, DataStar for AI-assisted data workflows). The platform is used to design and stress-test supply chain structures and policies—facility locations, flows, inventory strategies, transportation configurations—and to compare scenarios along cost, service, resilience, and sustainability dimensions. It does not execute operational transactions; it outputs designs and policies that must be implemented elsewhere.

Through what mechanisms and architectures does it achieve this? Mechanistically, the platform relies on: (1) MIP-based optimization for structural decisions; (2) discrete-event or time-stepped simulation to explore dynamics; (3) specialized engines for inventory strategy, greenfield site selection, and routing; (4) a standardized Postgres-backed schema (Anura) to store models and results; and (5) LLM-based components to improve data access and orchestration (Leapfrog AI, DataStar). The architecture and feature set are consistent with contemporary SaaS design for this domain. However, there is limited public detail on solver implementations, uncertainty modeling in the math itself, or performance characteristics at extreme scale.

How commercially mature is Optilogic? The company is commercially credible but still relatively small: founded in 2018, funded with at least $53M, with one notable acquisition and several high-profile customers (GM, Henkel) and a headcount likely in the dozens rather than hundreds. It occupies a clear niche in supply chain network design and digital twin modeling, differentiated by a modern cloud-native architecture and AI-enabled data tooling, but it is not a full-spectrum supply chain planning suite.

From a skeptical, evidence-based standpoint, Optilogic appears to offer a technically coherent and up-to-date network design stack, with some real innovation in user interaction (Leapfrog AI) and data workflows (DataStar), layered on top of orthodox but solid optimization and simulation engines. Its AI claims are best interpreted as enhancements to data and UX rather than radical new optimization mathematics—at least based on what is publicly documented today. Organizations evaluating Optilogic should focus their due diligence on (a) the quality and transparency of its models for their specific use cases, (b) the maturity of DataStar’s automation beyond marketing claims, and (c) practical integration patterns with their execution systems. For strategic network design and digital twin work, however, Optilogic is a serious, modern contender that deserves to be on the shortlist alongside more established names.