The Independent AI Support Layer Is Disappearing

For a few years there was a distinct product category: AI layers that sat on top of whichever helpdesk you already ran. Ada, Forethought, Ultimate and others sold into teams who wanted resolution capability without deepening their commitment to a platform vendor.

That category is consolidating. Zendesk acquired Forethought in March 2026 and folded it into its Resolution Platform, calling it the largest acquisition in its twenty-year history. Meanwhile Intercom, Freshdesk, HubSpot and Salesforce all ship first-party AI with outcome-based pricing.

Why the category existed

Because helpdesk AI used to be weak.

For a while, the built-in option was a scripted bot with a decision tree, and a specialist layer was genuinely better at resolving things. That capability gap was the entire commercial argument for the category.

The gap is closing. Not closed, and closing. Which means the argument for a separate layer has to shift from capability to something else.

What consolidation actually costs you

The benefits are real and immediate: one contract, one security review, one integration you no longer maintain, one vendor relationship to manage. Those are genuine costs and removing them is worth money.

The cost is deferred and shows up at renewal.

When your helpdesk and your AI layer are the same line from the same vendor, the cost of saying no goes up. You are no longer negotiating one component, you are negotiating your entire support infrastructure. Vendors know this, and the pricing conversation reflects it.

That is not a conspiracy, it is just how leverage works. It is worth pricing in before you consolidate rather than discovering it eighteen months later.

What to verify now

Ask any layer vendor whether they intend to stay independent. The answer is information, not a commitment, and a vendor that will not engage with the question has told you something.

Establish your built-in baseline. Before buying any specialist layer, measure what your current helpdesk tier already resolves on your real questions. That number is what a second vendor has to beat by enough to justify itself. Almost nobody measures it, which is how teams end up paying twice for overlapping capability.

Keep switching cost knowable. Export your data on a schedule. Document your configuration. Be deliberate about building irreversible dependencies on proprietary workflow. None of this is about planning to leave, it is about knowing what leaving would cost so the number is available when you need it.

Make sure an alternative exists before renewal. A credible alternative evaluated in advance changes a renewal conversation. One assembled in a panic during it does not.

The structural position

There is still a reason for a layer to be independent, and it is not capability.

It is that a vendor with no helpdesk to sell you into has no incentive to make your data harder to move, no reason to bundle the AI with a platform commitment, and nothing to gain from your switching cost rising. That is a structural property rather than a promise.

Fidiora is in that position: an independent resolution layer running alongside whichever helpdesk you keep, with no platform of our own to consolidate you into. That is obviously a self-interested thing to point out, so treat it as a disclosure rather than an argument.

The part worth taking regardless: after March 2026, independence from your platform vendor is a thing to check rather than a thing to assume, and the check takes one question.

Sources: Zendesk newsroom, TechCrunch.