What 6 decades of supply chain planning software teach us

Supply chain planning software didn’t suddenly become complicated. Its complexity grew slowly, over decades, as each generation tried to solve a real problem, but layered new complexity on top of old architecture.

‍

If you step back, a clear pattern appears:

‍

Planning software keeps getting more powerful, although not as powerful as the marketing claims, but it has not gotten proportionally easier to use, faster to deploy, or cheaper to implement or change.

‍

To understand the future, we must understand the past.  

‍

1960s–2000s: MRP and the foundations of APS

The earliest planning systems focused on one thing: e.g. making sure materials were available when production needed them, or optimizing a static supply chain model via a black box optimization.

‍

MRP and MRPII introduced bill of materials explosions, reorder logic, with a lot of deterministic rules.

‍

From this, the first Advanced Planning & Scheduling (APS) vendors emerged.

‍

At the risk of missing some, here is a list from the top of my head of some crucial ones that still exist today, although often after going through an acquisition.  

• Gainsystems (1971)  

• Numetrix (1977), acquired by J.D. Edwards, then acquired by Peoplesoft, then acquired by Oracle

• OMP (1985), previously known as OM Partners  

• JDA (1985)  

• Manugistics (1986), acquired by JDA, then acquired by BlueYonder

• Ilog (1987), acquired by IBM in 2009

• I2 Technologies (1988), acquired by JDA, then acquired by BlueYonder

• Toolsgroup (1993)

• Slimstock (1993)

• Arkieva (1993)

• John Galt (1996)  

• Logility (1996), acquired by Aptean in 2025

• Quintiq (1997), acquired by Dassault Systemes in 2014

• Ortems (1989), acquired by Dassault Systemes in 2016

• Futurmaster (1994)

• Dynasys (1985), acquired by QAD Inc. in 2012  

• SAP APO

• Llamasoft (1998), acquired by Coupa in 2020

• Icron (1992)

‍

Quite some of these systems were strong at execution but very rigid. They worked best if the business and data adapted more to the software rather than the other way around.

2000–2010: APS matures

As supply chains globalized, planning needed to go beyond materials planning or black box optimization.

‍

This era introduced time-series forecasting, constraint-based planning, early scenario analysis, value stream planning and cross-company collaboration (in reality, almost never, as cross-functional is already big enough of a challenge)

‍

New players and evolutions appeared:

• E2open (2000)

• Terra Technology (2002), acquired by e2open in 2016

• Relex (2005)

• Kinaxis (1984), with many different names before

• Anaplan (2006)

• Lokad (2008)

• Netstock (2009)

• O9 Solutions (2009)  

• SAP APO (around 2007), planned end of life 2027

‍

This period also marked the start of heavy consolidation. Independent APS vendors were absorbed into larger suites which were and still are technically very difficult to integrate, increasing the breadth of the solution, but also slowing adaptability towards new customer needs.  

2010–2020: machine learning enters quietly

Machine learning didn’t arrive with the GenAI hype. It entered planning way earlier thrugh, better forecast model selection, demand sensing, etc.

‍

At the same time, cloud offerings became way more standard, IBP connected supply chain planning with financial planning and analysis, and digital twins and control towers gained popularity, although after so many years, there is still unclarity on what these exactly mean.

‍

Vendors today will claim they have used machine learning way earlier than this, but many of these older ones still don’t properly use it today and don't have the native architecture to embed it easily.  

‍

SAP announced that it would sunset APO in 2017 (which has been postponed multiple times, and is now planned for 2027), and many implementations became end-to-end projects in which all planning functions and cases had to be incorporated.  

‍

What also did not change was time to value. Implementations still took many months or years. Excel remained the real planning environment in many organizations.

2020–2023: smaller, faster players

Cloud maturity lowered the barrier to entry and a new generation of tools focused on a narrower scope, faster pilots, finally a modern user interface, and on planners instead of only on executives or IT leaders.  

‍

Examples include Garvis.

‍

Garvis, which I saw from up close, was a clear example of a rapid solution that solved demand planning well, and was innovative in its use of machine learning and GenAI when OpenAI launched ChatGPT.

‍

The acquisition by Logility in 2023 showed that legacy players recognized this shift, even if many of them still struggle to make this a reality.

post-2023: GenAI, agents, overclaiming, proliferation of companies

Generative AI changed the narrative overnight.

‍

Suddenly, everyone had copilots, everyone had agents, and everyone promised a form of autonomous planning (here, I suggest you read this article by Lora Cecere).

‍

The reality is that many of these products barely exist beyond demos. Some vendors are trying to build planning systems inside GenAI, which is clearly the wrong direction.

‍

Planning still requires strong data models, deterministic and probabilistic engines, traceability of decisions and changes, the potential to collaborate with other departments and create scenarios, while GenAI is an interface for interaction and an accelerator for development, not the foundation of a system that is crucial for an organization.

What changed and what didn’t change then?

System complexity for smaller vs larger companies

What changed

‍
Advanced planning is no longer technically reserved for the largest enterprises. Cloud infrastructure removed heavy IT barriers, making it possible for small and mid-sized companies to access planning capabilities that were previously out of reach.

What didn’t

‍
Large companies still introduce complexity that has little to do with software capability. Stakeholder management and internal politics quickly dominate the conversation. In practice, systems that work well for smaller companies often work just as well for larger ones, until organizational complexity gets in the way.

‍

Marketing claims

What changed

‍
Marketing evolved massively. Messaging is sharper, more confident, and more ambitious than ever before.

‍

What didn’t

‍
The software itself rarely keeps pace with those claims.  

‍

Core functionality

What changed

‍
Algorithms improved. Forecasting methods became more sophisticated. End-to-end planning ambitions expanded.

‍

What didn’t

‍
Many suites remain collections of acquired solutions that don’t fully fit together or are a patchwork of cloud reengineered and legacy architecture, creating functional depth without conceptual flow.

‍

User interfaces

What changed

‍
Executive dashboards look better and are easier to consume at a high level.

‍

What didn’t

‍
Planner usability and intuitiveness still lags behind. Many tools remain cumbersome for daily work, which is why Excel continues to be the escape.

‍

Cost

What changed

‍
Cloud significantly reduced infrastructure costs. Storage and compute became cheap and scalable, and can be shared across customers in multi-tenant clouds.

‍

What didn’t

‍
Total cost of ownership for customers did not come down in the same way. Dependencies on and shortage of consultants, high cost of change, and long-running projects continue to dominate. Cloud made storage and compute cheap, but organizations didn’t see much of that benefit reflected in what they actually pay.

‍

Time of implementation

What changed

‍
There is constant talk about “faster time to value,” and vendors increasingly acknowledge that long implementations are a problem.

‍

What didn’t

‍
Access to a usable environment still often takes months, and full rollouts still take years, especially in large organizations.  

‍

Data

What changed

‍
Integrations became easier. Connecting systems and ingesting data is less technically complex than it used to be with modern APIs.

‍

What didn’t

‍
Data quality issues persist and will persist, and ERP master data problems remain a structural challenge. Data problems didn’t disappear with AI. AI simply exposes them faster and can provide suggestions to resolve these problems.

‍

ERP as planning

What changed

‍
ERPs added more planning features, and marketing increasingly positioned ERP as an end-to-end planning solution.

‍

What didn’t

‍
ERP remains transactional at its core, while planning requires flexibility, experimentation, and iteration.

‍

ERP is built for control. Planning is built for learning & simulation.

‍
Treating them as the same thing keeps failing, yet the debate keeps repeating.  

‍

IT typically pushes for this, to keep everything in 1 system, never mind what is needed for running supply chain planning properly and ignoring that in many cases an ERP and APS system owned by the same company is not necessarily easier to integrate as it is often built on different legacy architectures.

‍

Market power and analysts

Gartner remains a kingmaker for large enterprises, strongly shaping vendor selection and long-term platform decisions. More commentary and criticism, and even competition (e.g. Supply Chain Movement, keynotion events, cparity events, worldwide business research events), have sprouted and grown during the last years.  

‍

For small and mid-sized companies, those rankings are often irrelevant.

‍

If you put all of this together, a pattern emerges. The problem is no longer missing algorithms or compute, but the problem is that planning software still doesn’t behave like modern software.

‍

Why all of this led to Horizon

If you’re still reading this, allow me to look ahead now that we have looked at the past together.  

‍

After years in this industry, we believe the next breakthrough is not another algorithm or AI claim.

‍

It’s this:

Advanced planning should deliver value before projects, consultants, and long setups.

‍

That belief is why Horizon Solutions exists.

‍

Our vision

Advanced planning that works immediately.

‍

Planning should start when you log in, not after months of setup.

‍

Our mission

Make advanced planning accessible, in time, effort, and ownership.

‍

We help manufacturers’ supply chains move beyond Excel and slow, consultant-heavy systems by giving planners immediate access to a real planning environment on their own data.

Start small.

‍
See value fast.  

Begin with demand and expand into supply planning and scheduling only when ready.

‍

What we are not

• a data science toolbox

• a consulting-led business

• a vague “AI agent” promise with no product to show for it

‍

We deliberately avoid these paths because they delay value instead of delivering it.

‍

The real shift ahead

Advanced planning shouldn’t be hard to start, expensive to change, and impossible to trust.

‍

It should be, usable immediately, flexible as businesses evolve, owned by planners, not projects.

‍

That’s the evolution that still hasn’t fully happened.

‍

And that’s what we’re building for.

‍

If you’re curious, I invite you to try it out yourself, and you can see it happening within hours.