Platform Power
What makes simulation differentiated at Lyric vs any other competitor?
Evaluator-based simulation with high scenario control, including asset-aware route simulation, plus auto-scenario generation and auto logs for scenario comparison. Built-in stochastic uncertainty modeling and resilience analytics across the supply chain (not just demand and lead time!), with distribution fitting (120+ distributions) and auto best-fit detection. Operational depth + scale features: shelf-life modeling (product-level expiry + replenishment logic), tariff and landed cost calculator, and parallel multi-scenario runs. Connected to optimization for risk-aware planning via scenario evaluation.
Customer asks how Lyric selects which forecasting algorithm to use for each SKU. Walk through the "tournament" approach.
Multiple algorithms (statistical, ML, foundational models) compete on historical data. Best-performing algorithm wins for each SKU automatically. Hyperparameters auto-tuned.
What's a common limitation with legacy planning software with respect to implementation?
Legacy platforms (Kinaxis, Blue Yonder, SAP IBP, o9) are template-based - pre-built for common scenarios but unique business logic doesn't fit templates. Requires: custom coding/configuration, expensive consultants, implementation work, PS dependency for any changes. Examples of "non-standard" scenarios: seasonal labor constraints, multi-tier customer pricing, complex allocation logic.
Global prospect needs to deploy across US and EU with data residency requirements. Does Lyric support this?
Yes. Lyric can be hosted in customer-selected cloud (AWS, Azure, GCP) with choice of region. Supports geographic access restrictions and enforces data residency by region to meet regulatory requirements.
Name 2 feature requests that you get the most from customers.
NO RIGHT ANSWER!
Customer's data science team has proprietary ML models in Python. How does Lyric's approach differ in integrating these from legacy software?
Lyric supports "bring your own Python" - data science teams can deploy custom ML models that run alongside Lyric's built-in algorithm. Models integrate directly into planning workflows. Use cases: proprietary demand sensing algorithms, custom feature engineering, industry-specific forecasting logic. Other competitors require their forecasting methods or expensive customization through consultants. Lyric lets teams leverage existing ML IP.
What is a common Kinaxis pain point related to demand forecasting?
Kinaxis has a limited selection of algorithms (mainly statistical) because of which it doesn't perform well in cases where there is limited history like NPI - statistical algorithms require ~2 years of data to provide good forecasts. Lyric has the full breadth of algorithms - Statitical, ML, Foundation suited for different situations.
What's the biggest limitation for o9 implementations?
o9 uses their own proprietary language "IBPL" which leads to limited choice of implementation partners and ramp up time, vendor lock-in, any customization will need PS element.
Why is Lyric a better fit than Coupa or Optilogic for turning network design into an ongoing operating process, not a one-off project?
Lyric supports “always-on” runs and repeatable workflows (embedded ETL capability), so teams can refresh models on a cadence and compare scenarios continuously. Example: weekly network refresh plus a what-if on capacity changes, then reuse the same workflow next week without rebuilding.
Name 4 planning use cases Lyric is targeting in Q1 FY27
Phase-in/phase-out, MOQ consolidation, safety stock optimization, transportation optimization.
How is building on Lyric different from Palantir?
Palantir Foundry is powerful for data integration and analytics, requires developers to build anything, platform complexity is high and not customized to supply chain only vs Lyric has a built-in supply chain data model, low code/ no-code apps
Can Pigment do optimizations/ inventory planning?
Pigment acts as a visualization layer on top of other planning engines. It ingests planned orders and other outputs to create interactive dashboards for visual scenario analysis. Pigment doesn't solve large constrained combinatorial optimization problems (e.g., mixed-integer optimization for supply chain decisions) as a first-class product. It can do financial inventory planning (dollars), but can't do operational inventory optimization (units, locations, service levels, constraints)
What's the biggest disadvantage with SAP IBP?
SAP IBP implementation quality, duration depends on the consultant implementing it - users cannot build anything on the solution - it has to be done by consultants.
Can Aera do supply chain planning?
Aera is “decision layer” that can help planners act faster and automate decisions using integrated data, it sits on top of a planning system
Aera quote - "Aera is deployed for many customers who already have a planning tool in place, and the two go hand in hand. Planning typically operates in time horizons beyond the frozen period. You deploy decision intelligence within the frozen horizon, where a lot of changes are happening and you need to respond quickly."
Examples of Aera "Skills": Inventory rebalancing (move excess stock automatically), Aging inventory management (prevent spoilage)
Is SAP IBP (Planning module) the main business for SAP?
No. SAP's core business is ERP (SAP S/4HANA, SAP ECC) - enterprise resource planning for finance, HR, procurement, manufacturing. SAP IBP (Integrated Business Planning) is an add-on module for supply chain planning. This matters because: IBP gets lower priority than core ERP products, support can be slower (add-on vs flagship), innovation focuses on ERP not planning modules. When selling against SAP IBP, acknowledge their ERP dominance but position Lyric as purpose-built for planning (not an ERP add-on). Avoid "SAP bashing" with ex-SAP executives at the prospect.
Blue Yonder customers praise comprehensive suite but complains modules don't integrate well. Why?
Blue Yonder is suite of acquired products stitched together (Manugistics, i2, JDA, Blue Yonder), not built as unified platform from the start. Different modules acquired at different times with different architectures. Hence the integration challenges across modules.
o9 markets AI/ML forecasting heavily. What's the limitation that Lyric addresses?
o9 uses lacks transparency about which algorithms are selected, why they're chosen, and how to validate results. Technical documentation is limited - difficult to reproduce or audit forecasting behavior. Lyric's tournament approach shows which algorithm won for each SKU, performance metrics, and why it was selected. Transparent explainability builds trust with planners and enables continuous improvement.
o9 has no retraining capabilities built in, so o9 developers/FDE's are required to update and maintain models even once they are deployed.
Why does legacy software struggle to show proof of value fast?
Legacy tools often require full implementation before dashboards and performance metrics are usable. Lyric can launch an MVP in months with dashboards and measurable outcomes early.
What are the challenges of building custom planning applications on Databricks?
Databricks great for data engineering and ML R&D, not for production planning applications with business workflows. Building planning apps on Databricks requires massive engineering lift beyond data science: models drift and need continuous retraining, feature engineering breaks when source data changes, data scientists become app maintainers instead of doing strategic work. Still need to build entire application layer (UI, scenario management, workflow orchestration, approval processes, audit trails, RBAC, ).
In which Gartner MQ is Anaplan a leader and what does it imply about them?
Anaplan is a leader in the Financial Planning Software category. It is a financial planning tool, with good UI and collaboration features - more suited for driving alignment between different departments on financial planning. Limited supply chain planning capabilties.
What's the architectural advantage of having ML predictions, optimization, and simulation in the same platform at Lyric?
Lyric creates a closed-loop: (1) ML predictions (lead times, stock-outs, order ETAs) feed into optimization, (2) Simulation tests optimization results under uncertainty, (3) Learn from simulation and re-optimize. All in one platform. Competitors separate these - forecast in one tool, optimize in another, simulate in third tool, no feedback loop. Lyric integrates predictive ML, prescriptive optimization, and simulation testing automatically.
Customer complains Lyric scenario refresh is slower than Kinaxis. How do you explain this?
Kinaxis uses in-memory computing and incremental copies (efficient but complex setup) but also, Kinaxis has specialized heuristics for their solutions (speed) rather than MIP optimization algorithms fueling their solutions. Lyric is working building faster scenario capabilities but already is out in front regarding accuracy and constraint consideration as well as bring in options for heuristic approaches for times where speed is a priority.
Customer switching from legacy software has years of customizations, integrations, and trained users. What makes migration complex beyond just technology?
1) Organizational inertia - years of muscle memory, workarounds, and tribal knowledge creates resistance ("why change what works?")
(2) Business process dependencies - workflows, KPIs, and cross-functional processes designed around Blue Yonder's logic must be redesigned
(3) Integration archaeology - dozens of integrations with outdated/missing documentation and data formats downstream systems expect
(4) Knowledge loss - original implementers gone, rationale for customizations lost
(5) Political challenges - careers built on Blue Yonder expertise, executives who championed it resist change
(6) Change management - retraining users takes months beyond technical work.
Migrations fail more from change management than technical issues. De-risk with phased approach, parallel runs, and executive sponsorship.
In retail, what Kinaxis limitation shows up when you forecast?
Kinaxis can hit data-model guardrails on forecastable part-site combinations and performance limits at ship-to granularity, which can force forecasting at higher aggregation levels. Lyric’s flexible compute can scale up for heavy runs to support detailed forecasting.
What was the RAM size of the machine used for the Pepsico model?
1.5TB RAM machine (200+ suppliers, 100+ plants, 500+ warehouses, 3K SKUs. 79M+ continuous variables, 300K integer/binary variables, 27M+ constraints)