Most “Chatbase alternatives” posts are a feature matrix with a winner chosen before the table was drawn. This one is narrower and, we hope, more useful: six tools that crawl your site and answer on it, compared on the four things that decide whether you will still trust the bot in month three — whether it shows where an answer came from, what it does when the answer is not in your content, how a visitor reaches a person, and what the entry price is actually metering.
We make one of these. Matter Chat is listed first because this is our site; the criteria are the same for all six, the prices are what the vendors publish, and the last section hands you a test that does not need us in the room.
Why people leave Chatbase
Chatbase is a capable product with a large connector list and its own helpdesk. The reasons people shop around are mostly on the pricing page: answers are metered in credits that vary by model, removing the badge is a separate $99 a month, the helpdesk and API start at the $150 tier, and the free tier deletes an agent after 14 days of inactivity. None of that is hidden. It is simply a pricing shape that suits some teams and not others, and the four alternatives below each have a different shape.
The comparison
| Tool | Entry paid price | What the plan meters | Citations in answers | When it doesn't know |
|---|---|---|---|---|
| Matter Chat | $29/mo yearly, $39 monthly | 2,000 replies, 200 indexed pages, 1 bot; hard caps, no overage | Every answer, linked to the page | Refuses and logs the question |
| Chatbase | $40/mo ($32 yearly) | 700 message credits; credits vary by model; 10 MB training | Not documented | Not documented |
| Intercom Fin | $29/seat/mo + $0.99 per outcome | Seats, then per resolution (50/mo minimum off Intercom) | Inline source links on Messenger and email | Expresses uncertainty, asks to clarify, routes on error |
| Tidio Lyro | $24.17/mo yearly | 100 conversations + 50 one-off Lyro conversations; two meters | Not documented | Passes what it cannot handle to agents |
| DocsBot | $49/mo ($41 yearly) | 5,000 credits, 5,000 source pages, 1 bot, 1 user | “Sources” links, only the ones used | Classifies unanswered and escalates |
| Chatling | $25/mo ($21 yearly) | 1,750 credits (0.5–25 per answer by model), 2 agents | Optional, off by default | Not documented |
Matter Chat
Ours, so the short version. Matter Chat crawls your site and uploads, answers only from that content, and puts a citation under every answer that links to the page it came from. When the content does not cover a question it refuses rather than improvising, and the question is logged so you can write the page. Plans meter replies — one answer to a visitor, however long the back-and-forth — with hard caps and no overage, and every plan includes its team seats (more are an add-on). The badge comes off on Growth; white-label, WhatsApp and SMS are Agency.
What to weigh against it: there is no built-in helpdesk or voice channel, and the pitch is deliberately narrow. If you need a ticketing system, Matter Chat sits beside one rather than replacing it. The pricing page has the three tiers side by side; the trial is free, needs no card, and is enough to run the test at the end of this post.
Chatbase
Best for teams that want the bot, a helpdesk, voice and many channels from one vendor and are comfortable with credit pricing. Sources include files, websites, Notion and Zendesk or Salesforce tickets, up to a 100 MB cap. Escalation creates a ticket in Chatbase's own helpdesk or in Zendesk, Salesforce, Intercom, Zoho, Freshdesk, HubSpot, Help Scout or Gorgias. We could not find citations or a refusal policy documented; test both yourself. Costs stack — credits, $25 per extra agent, $99 to remove branding — so price the month you expect, not the tier.
Intercom Fin
The enterprise option, and the most transparent on grounding: Fin includes source links inline on Messenger and email, and its help centre describes what it does when it is not confident. The trade is the pricing model — a seat, then $0.99 per outcome, where an outcome is a resolution, a procedure handoff or a qualification. Cost scales with success and is hard to forecast; Fin's own pricing page is candid about the 50-outcome monthly minimum if you run it on another helpdesk. If you already live in Intercom, it is the obvious choice.
Tidio Lyro
Best for a small team that wants live chat first and an AI agent second. Every plan, including free, comes with 10 seats, and Lyro is positioned as working only from the content you give it and handing off what it cannot handle. Two things to read twice: Lyro conversations are a separate meter from billable conversations, with 50 one-off on the lower tiers, and file upload for training is a higher-tier feature. Citations are not documented.
DocsBot
The widest connector list of the six — tickets from several helpdesks, Google Drive, GitHub, YouTube — and explicit control over citations in the widget, showing only the sources an answer actually used. There is no native live chat; handoff is a support link or a callback into your own tools. The free tier's bots expire after 30 days, and credits, bots and source pages are metered separately, so the Personal plan at one bot and one user is genuinely personal.
Chatling
The cheapest paid tier here, with live chat, voice minutes and several channels included. Citations exist but are off by default — worth switching on before you judge it. The thing to model is the credit multiplier: an answer costs from 0.5 to 25 credits depending on the model you pick, so 1,750 credits is anywhere from a lot of answers to about seventy. Removing the badge is a $35 add-on, more than the plan itself.
Which one, for whom
- You want cited, grounded answers on a marketing or docs site with a predictable bill: Matter Chat or DocsBot.
- You already run Intercom, or need enterprise connectors and can budget per resolution: Intercom Fin.
- You want live chat for a small team first and some AI on top: Tidio or Chatling.
- You want bot, helpdesk, voice and channels from one vendor: Chatbase.
The test that settles it
Every vendor on this page offers a free tier or trial. Put the same five pages into each, then ask four questions: one the pages answer, the same one paraphrased badly, one the pages do not answer, and “can I talk to a person”. What you are checking is whether the citation resolves to the claim, whether the paraphrase gets the same answer, whether the out-of-scope question gets a refusal or an invention, and whether the handoff keeps the conversation. The long version is how to test an AI support tool in twenty minutes, and what a citation should prove covers the one check most tables skip.
