Hallucination is the number-one objection to support AI, and it is largely a retrieval and permissions problem rather than a model problem. The controls that actually reduce invented answers.
Also available in Dutch: Zo voorkom je dat je support-AI antwoorden verzint
Ask any support leader what stops them deploying AI on their front line and the answer is some version of: what if it invents a policy we do not have. It is the right worry. It is also, in a support context specifically, one of the more tractable ones — because the failure almost never originates in the model's imagination. It originates in what the model was handed and what it was allowed to do when the material ran out.
Lumping them together is why the problem feels unfixable. Separated, each has a different and fairly boring remedy.
| Failure | What it looks like | The actual fix |
|---|---|---|
| Retrieval gap | The customer asks about a policy that exists, but the search step returned nothing relevant, so the model filled the space from general knowledge. | A content problem, not a model problem. Find the questions that return nothing and write the missing article. |
| Stale source | The answer is confidently given and was correct last quarter. The article was never updated. | Review dates and one canonical page per fact. The model is faithfully repeating what you told it. |
| Unconstrained generation | The material was thin, the prompt did not forbid extrapolation, and the model produced a plausible-sounding policy. | Explicit grounding instructions plus a rewarded escape hatch. This is the only one of the three that is genuinely a model-behaviour issue. |
A model with no acceptable way to fail will fail unacceptably. If every turn must produce an answer, and the reference material does not contain one, something has to give — and what gives is accuracy. The fix is unglamorous: instruct the system to answer only from supplied material, never to invent policies, prices or order details, and when it does not know, to say so plainly and hand off to a person. That instruction is only worth anything if the handoff actually exists and works; an escape hatch that leads to a dead end trains the same behaviour as having none.
The most damaging invented answers in support are not wrong policies but wrong specifics: a delivery date, a refund status, an account tier. These should never come from a model's reasoning. They come from a system lookup, after the customer's identity is verified, or they do not get stated at all. The distinction matters enough that we built the identity check and the generic response on a mismatch into the retrieval path rather than the prompt — a prompt instruction is a request, not a control. There is more on where that sits in the features overview.
Most demos are run on curated content where retrieval never fails, so the interesting behaviour never appears. Ask instead: what does the system do when retrieval returns nothing — does it decline or improvise? Can I see the source passages behind a given answer? Are those sources visible to my agents, and separately, can they be shown to customers? Can I set a confidence point below which it escalates rather than answers? And what does the log retain, so a complaint can be reconstructed? Our own answers to those are in the documentation, and the measurement side is on how we measure.
One caveat worth stating plainly, since vendors rarely do: none of this reduces the error rate to zero, and any vendor implying otherwise is selling. The realistic goal is that errors are rare, visible, bounded to reversible actions, and traceable to a source you can fix. That is achievable. Perfection is not, which is why the approval queue described in AI agent vs chatbot matters more than any prompt.
In a support context the cause is usually that the retrieval step returned nothing relevant or returned an outdated article, and the system was not constrained to decline when its reference material came up short. Far less often it is the model reasoning badly. This matters because the first two causes are fixed by writing and maintaining content, which is ordinary operational work rather than a machine-learning problem.
Constrain it to answer only from supplied reference material, name the prohibited categories explicitly such as policies, prices and order details, and give it an explicit instruction to say it does not know and escalate to a person when the material does not cover the question. That escape route only works if the handoff genuinely reaches a human, otherwise the system learns that declining is not a viable option.
Yes, and this is the more tractable half of the problem. Order and account specifics should be produced by a verified system lookup rather than by the model, meaning the customer's identity is confirmed first and a mismatch returns a neutral response rather than a guess. Enforcing that in the retrieval path rather than in prompt wording is what makes it a control instead of a request.
Ask it about a policy you know you have never documented anywhere. A correctly configured system will say it is not certain and offer a human; a poorly configured one will produce a confident and plausible answer. Repeat this with questions whose answers changed recently, since stale-source errors are invisible to the model and will be delivered with the same confidence as correct ones.
Reopen rate within a stated window, because a wrong answer delivered confidently scores identically to a correct one on resolution rate, first-response time and often on immediate satisfaction ratings. The customer discovers the error later and comes back, so any resolution count that is not subject to a settling window will overstate quality precisely where it is worst.
2026-08-20 · 10 min read
Google retired FAQ rich results in June 2026, but answer engines still quote support content constantly. What actually earns a citation, and the structural changes that make your docs quotable.
2026-08-09 · 10 min read
Tickets closed, handle time and raw CSAT survive in reports because they always look fine. Here is the replacement set, how each number is gamed, and the weekly page a support lead can defend.
2026-08-09 · 9 min read
You can answer customers in their own language without maintaining a knowledge base per language — but only if you know which parts genuinely translate and which ones quietly break. Setup order, the failure modes, and what to measure.
See how honestly-measured AI resolutions cut your support load — start on the Free plan, no credit card, no sales call to get started.