We Asked 4 AI Models to Recommend a TMS (Transport Management System)

Here's a test worth running before you trust an AI chatbot's next software recommendation: we asked ChatGPT, Claude, Perplexity, and Gemini the same 8 realistic buyer questions a logistics company would actually type in, "best CargoWise alternatives," "is CargoWise still worth it," "compare Descartes Aljex, MercuryGate, and Cargoson," and logged exactly what came back.

We didn't cherry-pick. Each question ran fresh, independently, in a new conversation each time, so one answer couldn't bias the next. Then we counted: who got mentioned, how often, in what order, and, critically, who was invisible. 32 test runs in total.

GoFreight was mentioned across all four AI models more than any other platform, 15 of 32 runs, a 46.9% share, and was named the top recommendation in 8 of those 32 runs, more than double the next closest competitor. Whatever GoFreight is doing with its content and positioning, it's working at the AI layer, not just in traditional search.

Riege Scope, a genuinely well-regarded European air/ocean forwarding platform, was named four times, and only by ChatGPT. Zero mentions from Claude, Perplexity, or Gemini. Not "ranked lower." Absent. If you're a European forwarder relying on AI tools to shortlist software, and you happen to ask Claude or Gemini instead of ChatGPT, an entire credible option simply never enters your consideration set.

That's not a fluke. It's the whole thesis: AI-mediated discovery doesn't have one blind spot, it has several, and they're different depending on which model you ask.

We also found something less comfortable: several AI answers cited GoFreight's own blog, a "best CargoWise alternatives" post that naturally features GoFreight in the top spot, as if it were independent commentary. Self-published vendor content, folded into at least two different models' "independent" recommendations, phrasing and all.

This matters more than it sounds. A recent G2 survey found 51% of B2B software buyers now start their research with an AI chatbot instead of Google, up from just 29% a year earlier, and 69% ended up choosing a different vendor than they'd originally planned, based on AI guidance. 33% bought from a vendor they'd never heard of before. If that guidance is quietly built on vendor marketing dressed up as neutral analysis, buyers are being steered without knowing it.

CargoWise remains the second most-mentioned name overall, 12 of 32 runs (37.5%), named top recommendation 4 times, but mostly as the reference point every alternative gets compared against, not as an enthusiastic recommendation. Worth knowing: CargoWise's own Capterra rating sits at 2.9/5 overall, with customer service scoring just 1.9/5. High mention volume and high approval are not the same thing, and it's easy to conflate the two if you're only skimming AI-generated summaries.

This was a first, deliberately small test. 8 questions, 4 models, one run each. It's directional, not a controlled study. But it already surfaced real, checkable patterns: one dominant winner, a genuine blind spot, and a live example of vendor content slipping into "independent" AI advice unlabeled. We're turning this into an ongoing tracker, Agent Recommendation Share, measuring how often real logistics software vendors actually get recommended by AI, and by whom, over time.

Here's the full breakdown.

We ran 8 real buyer questions, the kind a logistics company would actually type into an AI assistant today:

Each question was run independently across ChatGPT, Claude, Perplexity, and Gemini, fresh conversation each time, no shared context between runs. 32 test runs total. This is directional, not a controlled study, a single query can vary with wording, model version, and freshness. Read it as a real, live signal, not a definitive ranking.

*Two related Descartes products counted together, directional, not a single clean figure.

Smaller, single-model mentions also worth knowing about: Neurored and Linbis appeared only in Gemini's answers, never in the other three. Fresa Xpress appeared only once, in an incomplete ChatGPT response that never actually named a final answer, a good reminder that not every AI answer resolves to a clear recommendation.

Every model has its own gap, and none of them overlap:

The practical takeaway: which AI a buyer happens to open determines which vendors even make the shortlist. That's not a small effect, it's the difference between being considered and not existing.

GoFreight's own blog, a self-published "best CargoWise alternatives" post that ranks GoFreight first, was cited as an apparent source in at least two separate models' answers, treated as if it were independent commentary rather than the vendor's own marketing. We're naming this because it's checkable, and because it's exactly the mechanism buyers should be watching for: content that looks like independent analysis but originates from the company being recommended.

CargoWise shows up constantly, 12 of 32 runs, second only to GoFreight, and is named the top pick 4 times. But raw mention volume mostly reflects CargoWise's role as the incumbent every question gets framed against ("alternatives to CargoWise," "is CargoWise still worth it"), not enthusiasm. Its own Capterra profile tells a different story: 2.9/5 overall, 1.9/5 on customer service specifically. Worth remembering: showing up often and being well-regarded are two different things, and AI-generated summaries can blur that line if you're not checking the underlying source data yourself.

A short, practical checklist for sanity-checking an AI's software recommendation before you act on it:

This was a first, small test. We're building this into an ongoing tracker, and we want it to reflect real operator experience, not just what four AI models happen to say. If you've evaluated or switched TMS platforms recently, tell us what actually happened, what worked, what didn't, what the AI got right or wrong, via our short CargoWise Alternatives survey. Your input feeds directly into the next, larger version of this report. Please reply using the email in my signature.

Thanks for reading and please share. Feel free to reach back with any comments or feedback.

Have a lovely weekend.

Derek.
[email protected]