Find and evaluate the right AI and digital supply chain technology
A practical guide to scouting, evaluating, and choosing technology — with recommended vendors across the supply chain stack.
Part 1
Strategic decisions before you scout
The biggest mistakes in technology selection happen before a single vendor is evaluated. Get these decisions right first.
1
Build vs. Buy vs. Partner
Every organization faces this at some point — do we build it ourselves, buy a product from a vendor, or partner with someone who can build or configure something for us? Each path has real tradeoffs: building gives you control and fit but is slow, expensive, and risky. Buying is faster and proven but requires adaptation. Partnering sits in between but depends heavily on the partner's capability and commitment.
My recommendation
Unless your organization has a clear DNA of building software — meaning you've done it before, have experience succeeding and failing, and maintain a large, capable development team — choose buying or at minimum partnering. Building software is a discipline, not a project. Most supply chain organizations should not be in the software business.
2
Point Solution vs. Horizontal Platform
A point solution solves one specific problem extremely well — a demand forecasting tool, a freight audit system, a warehouse drone for cycle counting. A horizontal platform aims to be the operating system for a broader function — an end-to-end supply chain platform or a full-suite TMS/WMS. Platforms promise integration and unified data but require significantly more resources, time, and internal capability to implement and operate.
My recommendation
Most of the time, choose the point solution. It deploys faster, proves value faster, and is easier to exit if it doesn't work. Choose a horizontal platform only when you have a genuine need that spans multiple functions, a dedicated team to implement and run it, and the organizational maturity to operate at that level.
3
Always run a pilot before committing
No vendor demo, reference call, or RFP response will tell you what it is actually like to use a product in your environment, with your data, and with your team. The only way to know is to run a focused, time-boxed pilot before signing a long-term contract.
My recommendation
Make the pilot non-negotiable. Any vendor worth working with will agree to one. Use NorthStar's Pilot Playbook module to structure it properly — define what success and failure look like before you start, not after.
4
Involve multiple people in the decision
Technology decisions made by one person or one team are almost always worse than those made with multiple perspectives. The daily user sees things the project lead misses. IT sees integration risks the business doesn't. Finance sees TCO implications that operations overlooks. Resistance to adoption often starts in the selection process — people who weren't involved feel it was done to them, not with them.
My recommendation
Build a formal selection team with at minimum: the future daily users, an IT representative, a finance stakeholder, and an executive sponsor. Give each group a structured evaluation role, not just a checkbox to sign off on.
Part 2
What to evaluate when choosing a software product
When evaluating any software vendor, look across three dimensions: Technology, Commercials, and Culture. Most organizations focus on Technology alone — and miss the things that actually determine whether a vendor relationship works.
⚙️
Factor 1
Technology
How well does this product answer your needs? Which of those needs are must-haves vs. nice-to-haves? Evaluate functional fit rigorously — but don't stop there. Look at integration capability, data requirements, scalability, and security. And critically: look at the product roadmap. When you choose a product, you usually want to use it for years. You want to know it has a relevant, innovative direction — not just what it does today.
Key questions: What are the must-have features vs. nice-to-haves? How does it integrate with our existing stack? What is on the roadmap for the next 12 months?
💰
Factor 2
Commercials
How much does this product cost and does it fit your budget? This seems obvious but there is a critical detail most teams miss: the cost of growth. Everything might fit the budget today — but what happens when you need to scale? More users, more features, more data volume. You do not want to be surprised by the pricing terms when you are already dependent on the product. Understand the full cost trajectory before you sign.
Key questions: What is the 3-year total cost of ownership? What happens to pricing if we double our users? What are the exit terms and data portability rights?
🤝
Factor 3
Culture
This is the least evaluated factor — and often the most important. When you buy software, you are not just buying a product. You are choosing the company that stands behind it. When issues arise — and they will — how responsive is their support? How much do they genuinely try to help? How customer-obsessed is this company? If there is a significant size gap between you and the vendor, ask yourself honestly: how important will you be to them as a customer?
Do not ignore gut-feel, red flags, or culture mismatches. These things surface during the sales cycle. They are fair game in any evaluation.
Key questions: How does their support model work in practice? Ask a reference customer, not the sales team. Do you feel like a priority to them?
Beyond the three-factor framework, a few additional criteria that are easy to overlook:
🏗️
Reference customers — the right question to ask
Ask for reference customers similar to you in size, industry, and problem. Then ask them not just "are you happy?" but "what didn't work, and what would you do differently?" That second question is where you learn the truth.
📊
Data requirements and readiness
Most AI products require clean, structured, historical data. Before evaluating a vendor, be honest about your data reality. Ask what data they need and what happens when quality is poor. The answer tells you whether the product is built for the real world or the ideal world.
🏢
Vendor viability
A great product from a vendor that fails is an operational crisis. For smaller or newer vendors, ask about funding, customer count, and churn rate. Ask what happens to your data and contract if they are acquired or shut down.
🚪
Exit strategy before you sign
Before committing, understand what it takes to leave. Can you export your data? What are the exit fees and notice period? A vendor that makes it hard to leave knows you won't be happy. Make data portability part of the contract negotiation, not an afterthought.
Use this as a structured vendor questionnaire — send it to vendors during your evaluation to get consistent, comparable answers across all three dimensions.
Part 3
Recommended technologies
Companies I have personally worked with, implemented into enterprises, or run pilots with. Not paid endorsements — recommendations based on direct experience.
A note on these recommendations: Every company listed here is one I have personally worked with, implemented into enterprise environments, or run structured pilots with. This list reflects direct experience, not paid placement or partnerships. Technology evolves quickly — always validate that a vendor's current capabilities match your current needs before committing.