Three Types of AI Vendors. Only One Is Real. Here's How to Tell the Difference Before You Sign.

By Geoff McDonald, CEO, Ambassador

I'm having the same conversation over and over with brand leaders this year. It always starts with some version of the same question:

"How do we know what's real AI and what's a vendor selling us a chatbot in a trench coat?"

It's the right question. The market is full of vendors making AI claims, and the gap between what they pitch and what they deliver is wider than it has ever been. The brands that buy wrong in 2026 are going to pay for that mistake for the next three to five years, because once you wire a platform into your customer experience, the switching cost is brutal.

So I want to give you a framework. Three types of vendors in the market right now. How to tell them apart. And five questions that will reveal which type you are talking to in under ten minutes.

Type 1: Legacy SaaS with an AI Sticker

These are the vendors who have been in your category for five, ten, sometimes fifteen years. They built their platform on technology that was state of the art in 2015. They have a large installed base, brand recognition, and the kind of enterprise sales motion that gets them on shortlists by default.

In the last 18 months, they realized they had to do something about AI. So they added an "AI tier" to their pricing page, shipped a chatbot inside their existing UI, and started using words like "agent-first" in their marketing.

The reality underneath: nothing has changed. The data model is the same one from 2015. The architecture cannot route customer outcomes back into platform decisions because it was never built to. The chatbot answers questions and stops there. It cannot act, it cannot decide, it cannot connect a customer signal in one part of the platform to an action in another part.

How they'll pitch you: "We've been the leader in this category for a decade and now we're bringing AI to it."

What you're actually buying: A pricing increase on the same product you could have bought three years ago, with a chatbot that adds friction more often than it removes it.

Type 2: The Flashy AI Startup

The opposite end of the spectrum. These are companies that started in the last 18 to 24 months, raised aggressively on the AI hype cycle, and have a demo that is genuinely impressive. The founders are often credible. The pitch deck is sharp. The agent does cool things on stage.

The reality underneath: there is no platform. There is a wrapper around a foundation model, some custom prompts, and a clean UI. There is no Customer Outcome Graph, no data integration, no orchestration layer. The agent can answer questions and execute simple tasks in isolation. It cannot tie outcomes back to the rest of your business because it has nothing to tie them to.

How they'll pitch you: "We were built for AI from day one. The legacy vendors can't catch us."

What you're actually buying: A demo that worked on a clean dataset, that will fail to scale into a real enterprise environment because the platform underneath was never built. In 24 months, this vendor is either acquired, pivoted, or out of business.

Type 3: The Platform That Rebuilt for This

These are the rarest. Vendors who looked at where AI was going three or four years ago and made the hard decision to tear down what they had and rebuild. Not refactor. Not bolt on. Rebuild from the foundation.

You can usually tell because the rebuild was visible. Existing customers went through a migration. The roadmap slowed for a while. The vendor made architectural bets that did not pay off for two years but are paying off now.

The reality underneath: every customer signal connects. Every action feeds the next decision. The agents act because there is a Customer Outcome Graph underneath them telling them what the customer has done, where they are in their lifecycle, and what to do next. The platform isn't just "AI-powered." The platform was designed for AI to be the orchestration layer, not the marketing layer.

How they'll pitch you: "Here is what we tore down. Here is what we rebuilt. Here is what it means for your customer experience."

What you're actually buying: A platform that compounds every dollar you spend on customer growth, because every signal, every action, every outcome routes back into the system and makes the next decision smarter.

Five Questions to Ask Any AI Vendor Before You Sign

You can identify which type you are talking to in under ten minutes if you ask the right questions. Here are the five I'd run on every AI vendor in your pipeline.

1. "Did you rebuild your platform in the last three years, or did you add AI features to your existing platform?"

This is the most important question. If the answer is "we added AI features," you are talking to Type 1. If the answer is "we rebuilt the foundation," ask the follow-up: what specifically did you rebuild, what did it cost you in terms of existing customers, and how long did it take. A real rebuild will have a real story. A fake rebuild will have buzzwords.

2. "Show me a customer journey where one agent's action changes what another agent does."

This is the orchestration test. If the agents in their platform act in isolation, the answer will be vague. If there is real orchestration underneath, they should be able to walk you through a concrete example: customer does X, agent A flags Y, agent B acts on Z, the outcome routes back into the next decision.

If they can't show you that, they don't have it.

3. "Does your agent know whether the action it just took moved the outcome?"

This is the closed-loop test. A real agent owns the outcome. It takes an action, watches what happens, and updates its next decision based on whether the action worked. A chatbot in a trench coat takes the action and stops there.

If the answer is some version of "we're working on attribution," you are buying a chatbot.

4. "What is your underlying data model and how long ago was it designed?"

Vendors built on data models from 2015 cannot do what AI requires in 2026. The questions you should be asking force them to either be honest about the architecture or expose that they're hiding it. If they can't or won't answer this, that's the answer.

5. "Who are three customers running your platform end to end, not just one module, and can I talk to them?"

The vendors who actually have a platform will produce these names quickly. The vendors who don't will give you references for a single feature or product line. If every reference is "they use our X module," the platform isn't really one platform. It's a portfolio of point solutions wearing the same logo.

The Real Choice in Front of You

Every brand evaluating AI vendors right now is making a five-year bet. The vendor you pick this year is the one that will shape your customer experience, your data architecture, and your team's daily workflow for the rest of this decade.

If you pick Type 1, you are paying premium prices for incremental improvement on top of incumbent technology. You will spend the next five years watching the gap between what your vendor promised and what they actually deliver get wider.

If you pick Type 2, you are betting on a startup that has a great demo and no platform. The demo will not survive contact with your enterprise environment. In 18 to 24 months, you will be doing this evaluation again.

If you pick Type 3, you are betting on architecture. The vendor isn't necessarily the loudest, the cheapest, or the most aggressively marketed. But every dollar you spend on customer growth will compound, because the platform underneath is designed to make every action smarter than the last.

This is the moment to ask harder questions. The market is going to consolidate fast over the next 18 months. Some vendors will get acquired and absorbed as features. Others will quietly disappear. The platforms that built for the future will be the ones still standing.

At Ambassador, we rebuilt our platform from the ground up over the last 18 months because we saw this moment coming. We built a Lifecycle Operating System with nine connected engines, seven specialist agents, and an orchestration layer called HiroAI that sits underneath every customer interaction. We didn't add AI to our platform. We rebuilt the platform so AI could actually do something.

If you'd like to see what that looks like in production, reach out. We'll walk you through what real architecture actually looks like, and we'll give you honest answers to the five questions above.

Pick the leader. Built for what comes next.

NOTES

Blog length: ~1,450 words. Substantive enough to be a real reference piece. Buyers will share it internally. That's the point. You want this on Slack threads at brands evaluating AI vendors.

Title alternatives if you want to A/B:

• "Three Types of AI Vendors. Only One Is Real."
• "Five Questions That Reveal Whether Your AI Vendor Is Lying"
• "The Vendor Evaluation Framework Every Brand Needs Right Now"

My pick: the version above. It promises the framework AND the test in one line.

SEO and shareability: "Three types of AI vendors" and "five questions to ask AI vendors" are both highly searchable phrasings. The piece will rank for AI vendor evaluation queries within weeks of indexing.

Suggested hero image: Three-column comparison visual. Column 1: "Legacy SaaS + AI Sticker." Column 2: "Flashy AI Startup." Column 3: "Real Platform." Each with a couple of one-line attributes underneath. Lands the framework visually before the reader scrolls.

Call to action on the blog: Soft and consultative ("we'll walk you through... give you honest answers"). Fits the post tone. Avoid hard "book a demo" CTAs on this piece. The frame is "we are the trusted advisor," not "we are the vendor selling you something."

Video CTA wording: When you record the video, the final beat should be something like: "I wrote up five questions you can ask any AI vendor to figure out which type they are. The link is below this video. Read it before your next discovery call. Pick the leader. Built for what comes next."