Review of Asper.ai, Supply Chain Software Vendor

By Léon Levinas-Ménard

Last updated: November, 2025

Asper.ai is an AI-centric software vendor focused on consumer businesses (primarily CPG, retail and adjacent manufacturing) that positions itself as an “interconnected decision” platform at the intersection of demand and supply. Commercially, the brand sits on top of Asper.AI Technologies Private Limited, an Indian private company incorporated in 2019 and now a subsidiary of analytics group Fractal, with additional entities in the US and UK. Operationally, Asper.ai offers a relatively narrow but deep product suite built around two modules: Dynamic Demand.ai (probabilistic demand forecasting and demand sensing) and Pricing & Promotion (revenue growth management, price and promo optimization). Technically, public evidence shows a cloud-delivered SaaS platform deployed on AWS and Microsoft Azure, using a modern web stack (Go, Python, Kafka, AWS services, Postgres, React) and a mix of deep learning and more conventional machine-learning for forecasting, plus rule- and workflow-driven automation for planning. The company reports multi-digit gains in forecast accuracy and promo ROI for anonymised customers, but provides very little verifiable detail about its optimization layer, decision architecture, or the exact nature of its “autonomic decisioning” and Gen-AI components. From financial filings and third-party databases, Asper.ai appears to be a mid-size, fast-growing but still commercially young vendor (revenues in the low tens of crores of INR in FY 2022-23, i.e. roughly USD 1–3m, with modest negative margins), strongly dependent on Fractal’s capital and sales channels. Overall, Asper.ai is best understood as a domain-specific AI application for CPG-style planning and revenue growth, not as a generalist supply chain optimization platform.

Asper.ai overview

At the product level, Asper.ai presents itself as a SaaS platform that helps “consumer businesses” make interconnected, AI-driven decisions across demand sensing, forecasting, pricing and promotional investment, with the promise of faster time-to-value and reduced capital tied in inventory.¹² The core user-facing artifacts are:

Dynamic Demand.ai – a module focused on demand forecasting and demand sensing, positioning itself as a way to anticipate demand “risks and opportunities” and unlock revenue growth.
Pricing & Promotion – a revenue growth management (RGM) module that aims to optimise strategic pricing and promotion portfolios and quantify the ROI of trade investments.¹²

The platform is marketed as AI-native and cloud-native. Asper.ai is offered both via AWS Marketplace (as “Asper.ai: Demand Forecasting at Scale”) and Microsoft Azure Marketplace / AppSource, indicating multi-cloud deployment options and integrations with common data and analytics stacks.³⁴ Target customers are mid-to-large CPG, food & beverage, and other consumer brands that already have significant data infrastructure and are looking to upgrade forecasting and RGM capabilities without building in-house data science teams.

Legally and financially, Asper.AI Technologies Private Limited is an unlisted private subsidiary of a foreign company, incorporated on 18 September 2019 in Bangalore, India, classified under computer-related services and IT consulting.⁵⁶⁷ Tofler and similar registries report authorised capital of INR 10 crore and paid-up capital of ~INR 9.67 crore, with FY 2022-23 revenue in the ₹10–25 crore band (roughly USD 1–3m) and modestly negative operating margins.⁵⁶ Fractal’s financial statements and LEI records confirm that this entity was formerly known as Samya.AI Artificial Intelligence Technologies Private Limited and later rebranded to Asper.ai.⁷⁸⁹ Public corporate databases and employment sites also show related US/UK entities and a small to mid-size headcount (roughly 50–200 people globally), with a leadership team including CEO Mohit Agarwal and other executives previously associated with Samya.ai.¹⁰¹¹¹²

In late February 2025, Fractal announced a USD 20m strategic investment in Asper.ai, explicitly positioning Asper as its AI platform for revenue growth and commercial decision-making for global brands.¹⁰¹³¹⁴ Press coverage and Fractal’s own statements describe four primary “growth levers”: demand forecasting and planning, revenue growth management, inventory planning, and sales execution—although only the first two are clearly productised on the public website today.¹⁰¹³¹⁴

From a technology standpoint, the platform is implemented as a cloud-hosted multi-tenant SaaS application on AWS (Redshift, EMR, ElastiCache) and Azure, with a backend in Go and Python, event streaming through Kafka, and a relational persistence layer (PostgreSQL or similar), as evidenced by engineering job ads and marketplace listings.³⁴¹⁵⁹¹⁶¹⁷ The AI layer is described as deep-learning-based demand forecasting, with additional Gen-AI components (large-language models) for some analytical and UX tasks, but there is no public technical documentation of model architectures, training regimes, or decision optimization algorithms beyond high-level marketing descriptions.³⁴⁹¹⁶

The rest of this report unpacks each of these aspects in detail, with a deliberately skeptical stance: we accept no AI or optimization claim that is not backed by concrete, reproducible-in-principle evidence and we treat anonymised or uncorroborated case-study claims as weak evidence only.

Asper.ai vs Lokad

Lokad (the host of this market-research series) and Asper.ai both operate in the broad “AI for planning” space, but they embody fairly different philosophies, architectures and commercial focuses.

Scope and vertical focus

Asper.ai focuses tightly on consumer brands—especially CPG and food & beverage—and on decisions around forecasting and revenue growth management: demand sensing, baseline and incremental promotion forecasts, pricing and promo portfolio design, and, by extension, some inventory and sales execution decisions.¹²¹⁰¹³³⁴ Its product modules and case studies are almost entirely within this CPG/RGM envelope.¹⁸¹⁹
Lokad, by contrast, is a horizontal quantitative supply chain platform used across retail, manufacturing, aerospace/MRO and other sectors. Its primary deliverable is an in-house DSL (Envision) and optimization stack that can be programmed into bespoke applications for demand forecasting, inventory optimization, production scheduling, network flows, and pricing across very different industries (from fashion retail to aircraft maintenance), as summarised in the Lokad brief.

In practical terms: Asper.ai offers pre-packaged applications for a fairly specific problem family (CPG forecasting & RGM), whereas Lokad offers a programmable platform for a broader class of supply chain problems.

Architecture and modelling approach

Asper.ai appears to rely on a conventional modern SaaS architecture: microservices, REST APIs, a relational data store, and separate ML services running deep learning models on cloud infrastructure.³⁴¹⁵⁹¹⁶ The business logic seems embedded in the application code and configuration of the Dynamic Demand.ai and Pricing & Promotion modules. There is no sign of an exposed modelling language or end-user programmable layer; instead, customers configure packaged workflows and dashboards.
Lokad builds around a domain-specific language (Envision) and a custom distributed VM. All forecasting and optimization logic is expressed in Envision scripts that are compiled and run on Lokad’s own execution engine, with an algebra of random variables and probabilistic optimization primitives. The “product” is effectively a programmable environment plus expert-built scripts, not a fixed app.

In other words, Asper.ai is closer to a vertically-focused AI application, while Lokad is closer to a supply chain programming environment.

Treatment of uncertainty and optimization

Public sources indicate that Asper.ai uses deep learning for demand forecasting and talks about “autonomic decisioning” and “automated, interconnected decisions” across demand and revenue levers.¹²³⁴¹² Case studies mention sizeable improvements in forecast accuracy (e.g. double-digit increases in forecast quality) and higher promo ROI.¹⁸ However, there is no detailed description of how the forecasts are turned into optimized decisions: whether they use full demand distributions, what objective functions are optimised, or whether they employ explicit stochastic or mathematical programming methods. The optimization appears to be presented as a black-box capability attached to the forecasting engine.
Lokad, by contrast, explicitly builds on probabilistic forecasting (full demand distributions) feeding into stochastic optimization algorithms (e.g. Stochastic Discrete Descent), articulated around economic drivers (stock-out penalties, holding costs, etc.). This is described at the level of algorithms and language primitives in its public technical material (see Lokad brief). Decisions (“order this much of SKU X, transfer Y units from DC A to B”) are explicit optimization outputs derived from those probabilistic models.

Thus, Asper.ai’s decision layer is opaque and under-specified in public sources; Lokad’s is explicitly modelled and auditable via code.

AI claims and transparency

Asper.ai makes strong claims around being an AI-native, autonomic decision platform and, more recently, a Gen-AI-enabled system.¹²¹⁰¹³⁹¹⁶¹² Job ads confirm the use of deep learning frameworks and LLMs, but there are no algorithmic explanations, technical blogs, benchmarks, or open-source artifacts that would allow independent assessment of these claims.⁹¹⁶ Case studies are anonymised and largely qualitative.¹⁸¹⁹
Lokad also uses advanced ML (including deep learning) but emphasizes white-box modelling: customers can inspect the Envision scripts, and Lokad has participated in public forecasting competitions and academic collaborations, providing some external validation of its technical stack (again, per the brief).

From a skeptical perspective, Asper.ai looks like a modern black-box AI application tailored to CPG planning; Lokad is a white-box probabilistic optimization platform with a heavier engineering and modelling burden but higher transparency.

Decision workflow and human role

Asper.ai’s UX narrative focuses on collaborative workflows for commercial and planning teams: marketing, sales, finance, supply chain. The promise is to automate much of the baseline forecasting and scenario analysis so that teams can spend more time on strategy and negotiation.¹²³⁴¹² The emphasis is more on “autonomic” recommendations embedded in business workflows than on exposing raw probability distributions.
Lokad positions itself as a copilot for supply chain teams: it produces ranked lists of actions (orders, transfers, schedule changes) and associated economic metrics, while leaving planners and executives to validate and execute. The locus of configuration is the Envision code; the UX is dashboards plus ranked action lists rather than a commercial-planning cockpit.

For a CPG revenue management director, Asper.ai may feel like a domain-specific cockpit; Lokad feels more like a quantitative engine sitting behind a more generic analytics UI.

Commercial maturity and go-to-market

Asper.ai is young as a brand (launch late 2022; USD 20m infusion in 2025) and has few publicly named clients. Case studies refer to “a multinational manufacturer of pet nutrition products” and “a packaged foods company” without naming the brands.¹⁸¹⁹ The vendor appears to be in a scale-up phase, backed by Fractal’s capital and relationships rather than a long independent track record.
Lokad has been operating since 2008 with a slower, largely organic growth model and a documented set of named customers (retailers, distributors, aerospace players). Its go-to-market centres on a small number of high-value accounts per industry, supported by in-house “supply chain scientists”.

In summary: Asper.ai is a Fractal-backed, CPG-focused AI application for demand and revenue planning with a largely black-box optimization layer and limited public technical detail. Lokad is a cross-industry probabilistic optimization platform with an exposed modelling language and more documented technical lineage, but also a steeper modelling and integration curve. They are not direct substitutes: Asper.ai is best evaluated as an RGM/forecasting app, Lokad as a general supply chain decision engine.

Company history, structure and funding

Legal entities and origins

Multiple independent registries agree that Asper.AI Technologies Private Limited:

was incorporated on 18 September 2019 under CIN U72900KA2019FTC128045,
is classified as a private company limited by shares, unlisted, and a subsidiary of a company incorporated outside India,
operates primarily in computer-related services / IT consulting & support.⁵⁶⁷⁸³

Tofler, Instafinancials, QuickCompany and TheCompanyCheck all corroborate this basic profile, with small differences in wording but consistent dates, CIN, and capital structure.⁵⁶⁹ LEI records and Fractal’s FY2022-23 audited financial statements further specify that the entity was previously named Samya.AI Technologies Private Limited / Samya.AI Artificial Intelligence Technologies Private Limited, before adopting the Asper.ai name.⁷⁸⁹ External sources such as SignalHire and LeadIQ also describe Samya.ai as “now Asper.ai, a Fractal company”, reinforcing the continuity.⁹

A separate ASPER.AI LIMITED exists in the UK, and various sources list a US HQ address in Chicago, IL, with London as a secondary location.⁹¹² Craft.co, for example, lists Asper.ai as a subsidiary, founded (as a brand) in 2022 with HQ in Chicago and a UK presence, consistent with the 2022 product launch narrative.¹²

Relationship with Fractal and funding

Fractal is a larger, long-standing analytics and AI company. Public information indicates that Samya.ai was originally incubated within or around Fractal and later fully integrated. The 2022 Fractal press release “Fractal announces launch of Asper.ai” explicitly describes Asper.ai as “a Fractal company” and positions it as a growth and decision-intelligence platform for consumer businesses.²⁰

In February 2025, Fractal announced a USD 20m strategic investment in Asper.ai, with the goal of accelerating product development and market expansion. The press release and follow-up coverage (SaaS news sites and regional tech media) emphasize:

the CPG / consumer focus,
the four “revenue growth levers” (demand forecasting & planning, RGM, inventory planning, sales execution),
and the use of AI to automate and interconnect these decisions.¹⁰¹³¹⁴

Tofler’s financials and Fractal’s subsidiary statements show that Asper.AI Technologies has revenues in the ₹10–25 crore band for FY 2022-23, with year-on-year growth in revenue and EBITDA but still modest negative operating margins.⁵⁸ This is consistent with a scale-up stage: meaningful revenue but not yet at the profitability and size of a mature enterprise vendor.

Leadership and headcount

Corporate and recruiting sites list:

Mohit Agarwal as CEO / co-founder for Asper.ai, previously CEO of Samya.ai,¹⁹
additional executives such as Chief Commercial Officer and Chief Product Officer with backgrounds in analytics and CPG.¹²

Glassdoor and various job portals classify Asper.ai in the 51–200 employee range, headquartered in Chicago with substantial engineering in Bangalore.¹¹¹² While exact numbers are not verifiable, the picture is of a moderately sized team—large enough to sustain a product but far from the scale of multi-thousand-employee planning incumbents.

Product suite and functional scope

Public positioning

On its website and in third-party directories, Asper.ai describes its mission as enabling “interconnected decisions at the intersection of demand & supply, powered by AI” for consumer businesses.¹² EliteAI.tools summarises the offering as a platform that:

improves demand sensing,
drives revenue growth,
optimizes pricing and promo investments,
and reduces capital and operational costs by automating decision workflows.²

The two clearly identifiable product modules are:¹²

Dynamic Demand.ai
- Demand forecasting and demand sensing.
- Detection of demand “risks and opportunities”.
- Use cases such as improving forecast accuracy, reducing stockouts, and enabling more agile planning.
Pricing & Promotion
- Revenue growth management, focusing on pricing and trade promotion decisions.
- Analytical support to allocate promo budgets, evaluate promo ROI, and optimise price packs and promotional calendars.

Third-party fund-raising news expands the scope to four levers (demand forecasting & planning, RGM, inventory planning, sales execution), but the public product UI still surfaces only the first two as distinct modules.¹⁰¹³¹⁴

Marketplace offerings (AWS & Azure)

The AWS Marketplace listing “Asper.ai: Demand Forecasting at Scale” describes Dynamic Demand.ai as a SaaS application focused on demand forecasting at scale for consumer goods, built on AWS Redshift, Amazon EMR and Amazon ElastiCache, and integrated with existing customer data stores.³ This confirms that at least one deployment pattern is tightly coupled to AWS analytical infrastructure.

The Microsoft Azure Marketplace / AppSource listing for Asper emphasises that the platform:⁴

unifies demand forecasting and revenue growth management around a single forecast model,
uses deep learning to incorporate multiple demand drivers (promotions, pricing, seasonality, etc.),
claims 10–20 percentage-point improvements in forecast accuracy and up to 80% automation of the portfolio (presumably, SKUs or planning decisions),
provides collaborative workflows for cross-functional teams (sales, marketing, finance, supply chain).

These marketplace descriptions are marketing-oriented but at least specify some outcomes and architectural context (AWS/Azure, deep learning, automation levels).

Case studies and white papers

Asper.ai provides a small number of anonymised case studies:

“Digitizing and automating demand planning for a pet nutrition company” – describes work with a multinational pet nutrition manufacturer, reporting large improvements in forecast accuracy and substantial automation of demand planning workflows, including consolidation of multiple legacy forecasting tools into Dynamic Demand.ai.¹⁸
“Future-proofing the demand planning process for a packaged foods company” – outlines a similar engagement in packaged foods, again with anonymised client identity, focusing on standardisation of demand planning, accuracy gains and process automation.¹⁹

A related white paper / blog (“From forecasting to fulfilment: using AI to optimize demand”, linked from the case study pages) elaborates conceptually on AI-driven demand planning but does not expose additional technical specifics beyond what is already in marketplace listings.¹⁸¹⁹

Crucially, no public case study names the end client. All references are at the level of “a multinational manufacturer…” rather than verifiable logos. These documents are therefore weak evidence from a due-diligence standpoint: they show that Asper.ai has run projects with at least a handful of sizeable companies, but they do not permit independent corroboration or customer cross-reference.

Technology stack and architecture

Cloud infrastructure and delivery model

From marketplace listings and job postings, we can infer the following about Asper.ai’s technical architecture:

The product is delivered as a multi-tenant SaaS application, with customers typically connecting to it via the web and integrating via APIs or batch data feeds.³⁴
On AWS, the platform leverages Amazon Redshift for data warehousing, Amazon EMR for large-scale processing, and Amazon ElastiCache for caching.³
On Azure, the AppSource listing suggests integration with common Microsoft data and analytics services, although implementation details are not disclosed.⁴

There is no indication of an on-premise deployment option; the stack appears fully cloud-hosted.

Application and data layer

Engineering job descriptions for Senior Backend Engineer and similar roles mention:

Golang (Go) as a primary backend language,
PostgreSQL or other relational databases,
Kafka for streaming / event handling,
microservices architecture and REST APIs,
containerisation (Docker, Kubernetes) and infrastructure-as-code.¹⁵¹⁶¹⁷

This is all consistent with a modern, but fairly standard, SaaS stack.

There is no sign of a domain-specific language or modelling environment akin to Lokad’s Envision; instead, the data model and decision rules are presumably encoded in the service code and configurations.

Front-end and UX

Public material is sparse on front-end technology, but the UX appears to be:

web-based dashboards and reports,
collaborative workspaces for planning cycles,
scenario and simulation interfaces for pricing and promotion portfolios.¹²⁴¹²

The EliteAI.tools directory and AppSource listing both emphasise collaborative workflows and “streamlined decision-making”, but they do not name specific frameworks.²⁴ It is safe to assume a SPA (React / Angular / Vue) front end, but this is an inference from industry norms and should be treated as such.

Data ingestion and integration

Although detailed technical docs are not public, marketplace descriptions and case studies imply that Asper.ai:

ingests historical sales, promotions, prices, distribution and external drivers (e.g. macro indicators),
connects to existing data warehouses / data lakes (AWS Redshift, Azure equivalents) rather than acting as the primary system of record,³⁴
outputs forecasts and recommendations which can be exported or integrated back into ERP / planning systems.

We do not see any claim of being an ERP replacement; like Lokad, Asper.ai appears to be an analytical layer sitting on top of transactional systems.

AI, machine learning and optimization claims

Forecasting models

Asper.ai’s public messaging and marketplace listings repeatedly refer to deep learning for demand forecasting and demand sensing.³⁴ The Azure listing states that the platform “leverages deep learning” to build a single forecast model incorporating multiple demand drivers (price, promo, distribution, etc.), with claimed improvements of 10–20 points in forecast accuracy compared to legacy baselines.⁴

Job postings for Data Scientist – Gen AI and similar roles list experience with:

Python and ML frameworks (TensorFlow, PyTorch, scikit-learn),
time-series forecasting and causal modelling,
large-language models and generative AI,
cloud ML pipelines and MLOps.⁹¹⁶

Taken together, this is credible evidence that non-trivial ML pipelines exist within the product, and that Asper.ai uses mainstream deep-learning toolchains. However:

There is no publication of model architectures (e.g. whether they use temporal convolutional networks, transformers, DeepAR-like models, etc.).
There is no benchmarking against public datasets or competitions.
The claimed accuracy gains are not backed by detailed statistical analysis (e.g. error distributions, statistical tests).

We must therefore treat the “10–20 point accuracy improvement” as plausible but unverified marketing, not as independently reproducible evidence.

Gen-AI and “autonomic decisioning”

Recent job ads and marketing language emphasise Gen-AI and autonomic decisioning:

Roles include “Data Scientist – Gen AI” to build LLM-enhanced features for planning and commercial decisioning.⁹¹⁶
EliteAI.tools and other directories highlight “automated workflows” and “streamlined decision-making” using AI.²¹²

In practice, Gen-AI could be used for:

natural-language exploration of forecasts and scenarios,
semi-automatic narrative reporting (e.g. explaining why a forecast changed),
classification and enrichment tasks.

However, there is no public explanation of how Gen-AI is concretely integrated into the decision pipeline. It is not clear whether LLMs are used in core optimization loops or only for peripheral UX features. Given the current industry trend, a conservative assumption is that Gen-AI is adjacent (explanations, UX) rather than central to the numerical optimization.

Optimization and decision logic

The critical question for this review is: does Asper.ai go beyond “forecasting + dashboarding” into genuine decision optimization?

Public materials claim:

automated, interconnected decisions across demand, pricing, promotion and inventory,
high levels of automation (80% of the portfolio “on autopilot”),⁴
ROI-driven recommendation of promo and pricing strategies.²¹⁰¹³¹⁸

But they do not:

describe objective functions (e.g. expected profit vs. service level),
expose whether decisions are based on full demand distributions or on point forecasts,
mention stochastic optimization, mathematical programming, or heuristics.

The pet-nutrition case study describes consolidation of multiple demand-planning tools into Dynamic Demand.ai, with improvements in forecast accuracy, reduced manual effort and better promo planning.¹⁸ However, the narrative is qualitative; it does not reveal whether the “recommendations” are primarily:

rule-based (e.g. thresholds and heuristics based on forecast outputs),
simple optimizations (e.g. greedy ROI sorting of promos under budget constraints),
or more advanced stochastic decision models.

In absence of explicit evidence, the most conservative reasonable interpretation is that Asper.ai provides:

advanced forecasting (deep learning, feature-rich models),
combined with workflow-embedded decision rules and some optimization over budgets and constraints,

rather than the type of explicit stochastic optimization stack that Lokad or OR-heavy vendors advertise. This does not make Asper.ai “not AI”, but it does suggest that the key technical innovation is in forecasting and RGM analytics, not in novel optimization algorithms.

Deployment, rollout and usage

Implementation pattern

While Asper.ai does not publish detailed implementation playbooks, we can infer a typical pattern from case studies, marketplace materials and job descriptions:

Data onboarding – ingest historical transactional data (sales, prices, promotions, distribution, external drivers) from the client’s data warehouse or data lake.³⁴¹⁸¹⁹
Model training and configuration – configure Dynamic Demand.ai and Pricing & Promotion to the client’s hierarchy (brands, SKUs, customers, channels) and calibrate models to historical patterns.
Workflow design – set up planning workflows for demand planning cycles and RGM cycles, including collaboration between sales, marketing, finance and supply chain.
Rollout and automation – progressively move more of the portfolio to “autopilot” where the system generates baseline plans and recommendations, with humans reviewing exceptions.

The pet-nutrition and packaged-foods case studies both describe this sort of staged rollout: pilot on a subset of categories, validation of accuracy and business impact, then progressive expansion and automation.¹⁸¹⁹

User roles

Public messaging and directories emphasise that Asper.ai is designed for:

Supply chain managers,
Sales and marketing managers,
Finance managers and executives,

who need to collaborate on demand planning and revenue decisions.²⁴¹² The UX seems deliberately business-user-oriented, not data-scientist-oriented.

Customer-facing job descriptions (Customer Success, Solution Consultant) stress the need to bridge between technical configuration and business value, suggesting that Asper.ai’s own team plays a significant role in implementation and ongoing support.¹⁷

Integration into execution systems

There is no detailed public information on ERP / TPM integration, but given that the product is delivered via AWS/Azure and sits on top of existing data warehouses, it is reasonable to infer:

upstream data flows from ERP, TPM, CRM and POS systems into the data warehouse, then into Asper.ai;
downstream flows of forecasts and recommendations back into planning systems (via file exports, APIs or connectors).

Again, this positions Asper.ai as an analytical overlay, not as a transaction system.

Clients, sectors and commercial maturity

Named vs anonymised clients

A key issue for a skeptical review is verifiable customer evidence.

The official website and case studies do not name any clients. Instead, they refer to “a multinational manufacturer of pet nutrition products” or “a leading packaged foods company”.¹⁸¹⁹
Public logos or detailed testimonials from recognisable brands are absent from the materials reviewed.

This does not mean Asper.ai has no real customers, but it does mean that as outside observers we cannot independently verify claims of impact or cross-check customer satisfaction.

Sectors and geography

From product positioning and case examples, Asper.ai is clearly targeting:

CPG and food & beverage manufacturers,
possibly retail and distribution arms of these brands,
with an emphasis on global or multi-regional operations.¹²¹⁰¹³¹⁸¹⁹¹²

Corporate records place the legal backbone in India (Bangalore) with front-office presence in Chicago and London, suggesting a go-to-market focus on North America and Europe for revenue-generating clients, with India as the main engineering and delivery hub.⁵⁷⁹¹²

Scale and maturity

Financial and corporate data indicate:

a 2019 incorporation and 2022 brand launch,
revenue in the ₹10–25 crore band (~USD 1–3m) for FY 2022-23, growing but with negative operating margins,⁵
USD 20m funding from Fractal in 2025 to scale the business.¹⁰¹³¹⁴

Glassdoor shows a small but non-trivial employee base, mixed reviews (including some comments referencing “RGM” and “Gen-AI” focus), and a work environment typical of a growth-phase SaaS firm.¹¹

On that basis, Asper.ai should be considered an early-scale vendor:

technically credible enough to have a working product and paying customers,
but without the depth of reference base or financial robustness of a long-established APS/ERP provider.

Assessment of how “state-of-the-art” Asper.ai is

Where Asper.ai appears modern and credible

On the forecasting and data-platform side, Asper.ai looks technically up-to-date:

Use of deep learning for demand forecasting and demand sensing, with multi-driver models, is state-of-practice in 2025, especially in CPG.³⁴⁹
The cloud-native, microservices-based architecture on AWS/Azure, using Go, Kafka and Postgres, is a standard, robust pattern for scalable SaaS analytics.³⁴¹⁵¹⁶¹⁷
The focus on revenue growth management (pricing & promotion) tightly coupled to forecasting is aligned with how advanced CPG players think about value: not just minimising forecast error but maximising revenue and margin across price and promo levers.²¹⁰¹³¹⁸¹²

From this angle, Asper.ai is not behind the curve; if anything, it is well aligned with current best practice in ML-driven planning for CPG, at least in concept.

Where evidence is weak or absent

However, several critical aspects lack hard evidence:

Decision optimization There is no detailed account of how forecasts become decisions under uncertainty (order quantities, promotional calendars, price ladders) and what mathematical or algorithmic techniques are used. Without this, claims of “autonomic decisioning” and “interconnected, automated decisions” must be treated as marketing assertions, not verified technical facts.³⁴¹⁸
Handling of uncertainty Public materials do not clarify whether Asper.ai operates on:
- full demand distributions,
- quantiles,
- or point forecasts plus heuristics.
This matters for assessing how well the system deals with uncertainty. In contrast, vendors who explicitly discuss probabilistic forecasting and stochastic optimization provide stronger, inspectable evidence.
Gen-AI depth While job ads show genuine investment in Gen-AI skills, there is no visible evidence that LLMs are being used beyond the UX and analytics narrative layer. At present, the Gen-AI angle looks real but peripheral, not core to the optimization logic.⁹¹⁶
External validation There are no public benchmarks, peer-reviewed papers, open-source artifacts or named reference clients that would allow external parties to test or replicate Asper.ai’s claims. All impact numbers come from vendor-produced, anonymised materials.¹⁸¹⁹

Overall technical verdict

Under a skeptical, evidence-based lens:

Asper.ai is almost certainly technically competent on the ML and cloud-engineering side: the stack, roles and marketplace integrations all align with current practices in AI-enabled SaaS for planning.
It is not yet demonstrably state-of-the-art in terms of published and verifiable decision-optimization methods for supply chain and revenue planning. The forecasting story is modern; the optimization story is opaque.
The vendor’s positioning as an AI-native, autonomic decisioning platform is directionally plausible but insufficiently documented to be taken at face value without direct technical access or customer validation.

For a sophisticated buyer, Asper.ai should be evaluated via hands-on pilots with carefully designed A/B tests and a clear view of how recommendations are produced and governed, rather than purely on the basis of marketing claims.

Direct answers to the key questions

What does Asper.ai’s solution deliver, in precise terms?

Based on public evidence, Asper.ai delivers:

Probabilistic / ML-driven demand forecasts and demand sensing for CPG-style product and customer hierarchies.
Revenue growth management analytics for pricing and promotion: baselines, uplifts, promo effectiveness and ROI measurement.
Workflow-embedded recommendations and partial automation for:
- demand planning cycles,
- promo and pricing calendars,
- potentially some inventory and sales-execution decisions.

These are delivered as cloud SaaS applications (Dynamic Demand.ai and Pricing & Promotion) integrated with the customer’s data warehouse and planning processes.¹²³⁴¹⁸¹²

Through what mechanisms and architectures are outcomes achieved?

Mechanisms (as far as can be inferred):

Data ingestion from the customer’s data stack (AWS Redshift, Azure, etc.).³⁴
Deep-learning-based forecast models trained on historical sales, price, promo, distribution and other drivers.³⁴⁹
Analytical layers that compute promo uplifts, ROI and scenario impacts.
A web-based application embedding these analytics into workflows for planning teams.

Architecture:

Multi-tenant SaaS on AWS/Azure.
Backend services in Go/Python with Kafka and relational storage.
ML pipelines using standard Python ML frameworks (TensorFlow/PyTorch/scikit-learn).[^[11]]¹⁵⁹¹⁶
Business-user-oriented UIs for planning and RGM.²⁴¹²

How substantiated are Asper.ai’s AI / optimization claims?

AI / ML – reasonably substantiated:
- Marketplace listings explicitly cite deep learning.
- Job ads list mainstream ML frameworks and Gen-AI requirements.³⁴⁹¹⁶
Optimization / autonomic decisioning – weakly substantiated:
- No visible technical description of objective functions, constraints, or algorithms.
- No external benchmarks or named customer validations.
- Case studies are anonymised and qualitative.¹⁸¹⁹

A cautious buyer should treat optimization claims as hypotheses to test in pilots, not as established facts.

Commercial maturity

Legally and financially, Asper.ai is a 6-year-old corporate entity with ~5–6 years of operations, but the Asper.ai brand and current product line date effectively from 2022.²⁰⁵⁷⁸⁹¹²
Revenues appear to be in the low single-digit million USD range, growing but not yet at large-scale enterprise-vendor levels.⁵
The firm has significant backing from Fractal (USD 20m investment, corporate integration), which supports its ability to continue product development and go-to-market.¹⁰¹³¹⁴
Lack of named client references and limited public documentation indicate a vendor that is still in early commercial scaling rather than a fully mature APS incumbent.

Conclusion

Asper.ai is best characterised as a Fractal-backed, CPG-focused AI application for demand forecasting and revenue growth management, delivered as cloud SaaS on AWS and Azure. Public information clearly supports the existence of a modern ML stack (deep learning, Gen-AI-adjacent features), a contemporary cloud architecture, and a product suite aligned with how consumer brands think about forecasting and RGM.

However, from a maximally skeptical, evidence-based perspective, several caveats are important:

The decision-optimization layer—how forecasts become concrete, economically rational decisions under uncertainty—is essentially undocumented in public sources. Claims about “autonomic decisioning” and high automation should therefore be treated as unproven until a buyer can inspect models and outputs directly.
Client evidence is weak: case studies are anonymised, and no verifiable client logos or independent testimonials are available. All impact numbers come from vendor-produced material.
The vendor is commercially young and still reliant on Fractal’s capital and ecosystem, with financials consistent with a scale-up rather than a fully mature enterprise vendor.

Relative to Lokad, Asper.ai looks like a vertical AI app with strong CPG/RGM orientation and black-box ML, whereas Lokad is a horizontal probabilistic optimization platform with a programmable modelling layer and more explicit, inspectable treatment of uncertainty.

For prospective buyers, the practical implication is:

Asper.ai may be a good fit if you are a CPG / consumer brand looking for a ready-made demand & RGM cockpit, willing to run pilots and judge impact empirically, and comfortable with a largely black-box optimization layer.
If you require transparent, programmable, cross-vertical supply chain optimization—with explicit control over models and decisions—Asper.ai’s current public footprint suggests it is not a substitute for platforms like Lokad.

In all cases, due diligence should include: a proof-of-value pilot with clear KPIs, access to detailed configuration and modelling explanations, and robust governance over how “autonomic” decisions are validated and overridden by human experts.