The questions people ask before buying AI support
20 questions answered one at a time, with the answer first and the reasoning after. No padding, and no assuming the answer always suits us.
Differences
What is the difference between a chatbot and an AI agent?
Where the answer comes from and what it can do with it. A chatbot answers from what is written in your documentation: it handles the questions whose answer is the same for everybody. An agent also reads the account state of whoever is asking and can run actions, so it resolves the questions that depend on that specific customer too.
Is a knowledge base the same as a help centre?
Almost. The knowledge base is the content — the articles — and the help centre is the public site where they are published, with its search and navigation. In daily conversation the two are used interchangeably and it rarely matters, except when comparing tools: some sell the first and some the second, and the price is not the same.
Does an AI helpdesk replace my support team?
No, and anybody telling you otherwise is selling something. It replaces one specific part of the work: the repeated, the documented, and a good share of what gets resolved by checking one account fact. Anything needing judgement, anything angry and anything nobody had seen still belongs to a person — usually the part you hired that person for.
Can an AI agent see each customer plan and data?
Only if you connect it, and that is exactly the line separating a documentation bot from an agent. By default an AI system sees only what you give it: give it documentation and it answers documentation; give it read access to your API as well and it can check the plan, the usage or the last invoice of whoever is asking, and answer with that fact.
What it can do
Does an AI chatbot make up answers when it does not know?
It can, and that is called hallucination: the model builds an answer that sounds plausible because its job is producing convincing text, not recognising that it lacks the fact. What separates a useful system from a dangerous one is not never failing, it is how it fails: "I do not know, let me get a person" is acceptable; inventing a billing figure is not.
Can an AI agent act without me reviewing it?
It can, and how far is your decision, not the vendor. The sensible split is by reversibility: what only reads breaks nothing and can run alone; what writes something recoverable can run alone with a log; and anything irreversible — cancelling, deleting, refunding — should pass through a person, not because the agent is wrong more often, but because when it is wrong nobody will notice in time.
Can I see which actions an AI agent has run?
You should, and a vendor who will not show you is reason enough to rule them out. Every action an agent runs has to be logged with what it did, on which account, when, and off the back of which conversation. Without that log you cannot audit a mistake, you cannot answer a customer who complains, and you cannot pass any mid-sized customer security questionnaire.
Can an AI agent change a customer plan?
Technically yes, if you give it that tool. The useful question is a different one: whether it should. An upgrade is recoverable and some teams leave it automated; a downgrade or a cancellation touches money and does not undo itself, so the sensible arrangement is the agent preparing it and a person confirming.
Is an AI chatbot any use for complex technical support?
For part of it, and not the part people expect. What it handles well is the initial diagnosis: recognising the error, checking that account state and ruling out the known causes before it reaches anybody. What it does not handle is the case nobody had seen, which is the definition of complex technical support and the reason a third tier exists.
Which languages does an AI agent answer in?
Practically any, and that is the easy part. What shapes the result is not the language the model writes in, it is the language your documentation is in: if you only have articles in English, a customer asking in German gets a translation of your content rather than content written for them. It is usually enough, but it is worth knowing.
Getting started
How long does it take to get an AI support agent running?
Answering from your documentation is a matter of hours, not weeks: what you already publish gets indexed and it starts replying. What takes time is the other half — connecting the reads against your API so it can check the account of whoever is asking — and that depends on your product team, not the vendor. One to two weeks is an honest estimate for having both.
Do I need to integrate all my tools before starting?
No, and it is better not to. With the documentation indexed you already resolve the group of questions that do not depend on the account, and that works on day one with nothing connected. Integrations get added in order of what gets asked most, and normally two or three cover the bulk: subscription state, usage, and the last order or invoice.
Do I need a help centre already to use an AI knowledge base?
No. The articles can live anywhere readable — a portal, a docs repository, even internal documents — and the system indexes them the same. What is required is that they exist in writing: if the answer is nowhere, there is nothing to index, and that is the only real prerequisite.
Do I need the official WhatsApp Business API to support customers there?
Yes, if you want to handle it with a team or with an automated system. The normal app and WhatsApp Business do not allow tools to be connected in any supported way, and the workarounds that manage it risk having Meta shut the number down. The official API is contracted through an authorised provider, who handles the verification.
Pricing
What counts as a conversation when you pay per resolution?
A conversation is a whole thread with one customer, not each message: if they write five times about the same thing, it is still one. What varies between vendors is when that thread gets cut and what counts as resolved, and those two definitions decide your invoice far more than the unit price on the website.
What happens if I go over my plan conversation limit?
It depends on a decision the vendor makes and that is worth knowing before signing: either the agent stops answering, or it keeps answering and the excess gets billed. The first turns a volume spike into a customer service problem; the second, into an unexpected invoice. Neither is bad, but knowing which applies to you changes which plan you buy.
Is Freshdesk cheaper than Zendesk?
On the per-seat rate yes, nearly always, and the gap is noticeable on the entry plans. But the saving narrows as soon as you move into the higher plans of either, and it disappears entirely if what pushed you to switch was the cost of the AI, because there both charge separately and on a similar model.
Is Intercom more expensive than Zendesk?
It depends on how many people answer and how much you use the AI, which is why the comparisons circulating contradict each other. Intercom charges per seat plus per resolution of the automated agent; Zendesk charges per seat plus its AI module separately. With a small team and high automated volume, Intercom usually comes out dearer; with a large team and little AI, it inverts.
Choosing and migrating
Can you migrate from Zendesk to Freshdesk without losing the history?
Yes, and both offer tools for it, but "without losing anything" is a promise worth qualifying. Tickets, contacts and articles move fine. What almost never survives intact are the automations, the custom fields with logic behind them and the historical reports, and that is what makes a migration take weeks rather than an afternoon.
Which helpdesk should I choose with three people in support?
With three people, what decides is not the feature list but two things: the work being visible — who owns what and what is outstanding — and answering not forcing anybody onto another screen to find a fact. Nearly every tool in the category handles the first; the second is where they separate, and it is what you feel every day.
