ChatBot by Text Review (2026)
A twenty-year-old public company mid-rebrand, whose two help pages disagree about whether its AI can call an API at all.
Some links on this page can be affiliate links. We may earn a commission at no cost to you. This never affects our rankings. How we work.
What the score measures. Five things we can verify from published material: whether pricing is transparent, whether the product can act on your systems or only answer from documents, whether the vendor still owns its own roadmap, how much independent review evidence exists, and whether the bill stays predictable as volume grows. Weighted, then scored against an ideal platform that scores 5.
BEST FOR: SMALL TEAMS WANTING A SEAT PRICE WITH A RESOLUTION ALLOWANCE ATTACHED · LAST UPDATED 2026-08-21
ChatBot by Text in depth
The company started life as LiveChat in Wrocław in May 2002, listed on the Warsaw Stock Exchange in April 2014, launched ChatBot in August 2017, and renamed itself Text in September 2023. Twenty-four years of trading history, a published net profit margin, dividends and SOC 2 Type 2. On the shortlists we see, it barely appears.
You are buying into a rebrand that is roughly two-thirds finished, and the seam is visible from the outside. The pricing page at chatbot.com is titled for Text's AI Agent and sells Text's plans. Every help article ends with an upsell to a different product. The onboarding guide, updated two days before we read it, is split in two: how to build an AI Agent in Text, then how to build a chatbot with the previous ChatBot.com workflow, described as remaining available for existing users. Signup on either domain lands on accounts.livechat.com.
What it does
- AI agent answering from a scanned website, uploaded files or written articles
- Custom Skills written in plain English and triggered on intent
- Twenty named actions across customers, chats and tickets
- Workflows with 20 Salesforce actions, triggered by events rather than by the agent
- Per-run audit log showing how each skill behaved in real conversations
- Twelve channel and platform integrations, with Instagram and Telegram listed as coming
More detail on how it works
The two properties do not even agree on how many customers there are. chatbot.com says over 35,000 brands, text.com says 30,000+ teams, both read on 21 August 2026.
None of that is disqualifying on its own. What matters is that the same seam runs through the documentation of the one capability a support buyer has to get right.
Two vendor pages, opposite answers, same week
The question is whether the AI can reach outside Text. One help page, Custom Skills vs Instructions, opens its list of what a Custom Skill can do with “Call external services via API (webhooks support)”. Another, Custom Skills, says “Below is a list of all the actions your AI agent can perform” and then enumerates twenty of them: get and update a customer, add a customer event, transfer, tag, close and supervise a chat, pull a transcript, create, find, update, tag and summarise a ticket, match an agent, detect a language. We fetched that page and counted occurrences of the word webhook in all 175KB of it. Zero. One page says the agent can call your API; the page that lists every action it can perform does not contain the mechanism.
What the marketing sides with, and what the docs actually document
Text's AI Agent feature page shows skill templates that read like tool calls, including one that calls a discount-service webhook to generate a retention offer and another that calls an order-status webhook with a collected order ID. So the marketing backs the permissive reading. Meanwhile the external writes that are unambiguously documented sit in Workflows, which are event-triggered rather than agent-chosen: a trigger fires on chat rated, ticket created, a keyword, a schedule or an app event, and then twenty named Salesforce actions become available. Create an account, update a contact, create a case, bulk import. Real writes to a real system of record, initiated by an event rather than by the agent deciding one is needed. Until the webhook claim is confirmed in-product, we classify this as answering from documentation and writing to Text's own records.
Ten skills, and that is the ceiling
An AI agent can have up to ten custom skills. For a business with four or five action-taking scenarios that is invisible. For anyone mapping a real support taxonomy onto it, ten is the number you should test your requirements against before signing anything, because no amount of plan upgrade moves it.
The resolution definition is the best thing on the product
One resolution means the AI directly solved a customer question, and multiple questions in one chat still count once. Then come the exclusions, and they are the unusual part. No charge when no issue is solved, when the answer is irrelevant, or for test, spam, greeting and browsing chats. No charge when the agent transfers to a human without having answered anything first. Answer one question and then transfer, and it bills. There is even a carve-out for ticket fallback: a ticket the customer asked for counts as a resolution, a ticket created because no human was available does not. Most vendors in this directory do not publish that level of detail, and several bill for exactly the cases Text excludes.
Setting ChatBot by Text up
Realistic time to a working deployment: An afternoon for answers. Read from the vendor's own documentation.
- 01
Open AI Agent settings
Name, tone, length, languages and custom instructions. Owners and Admins only — nobody else can edit or switch the agent on. Custom instructions cap at 10,000 characters.
- 02
Add knowledge sources
Scan a website on a repeating interval, upload a file, or write an article. Files cap at 50MB.
- 03
Configure Skills
Plain-English rules that trigger on intent rather than keywords. Hard ceiling of ten custom skills per agent, which is the constraint you meet first.
- 04
Preview, then accept chats
Test in preview before flipping the agent to live.
- 05
Install the widget
An HTML snippet before the closing body tag, via Google Tag Manager, or through the React, Vue or Angular package.
- 06
Decide about auto-refill before you launch, not after
It is on by default. Hit your monthly allowance and Text adds 50 more resolutions for $49.50 without asking. You can only reduce packages before renewal, and a reduction counts as a downgrade.
The step that takes longer than they imply. The webhook, if you need one. It takes a verification-request handshake on your own server, the vendor's own page says it needs some technical knowledge, and its documentation disagrees with itself about whether the AI can call one at all.
What the AI actually does
Every platform here says “AI agent”. It covers a bot that reads your help centre and a system that can refund a customer. This is which one you are buying.
Answers from documentation and writes freely to Text's own records — customers, chats, tickets.
Whether it can write anywhere else is a question two of Text's own help pages answer differently, so we do not classify it as acting on systems.
- Custom Skills are invoked by the AI on intent rather than keywords, which is the right mechanism
- Twenty documented actions, all of them internal to Text: get and update a customer, transfer, tag, close and supervise chats, create, find, update and summarise tickets
- The page listing all those actions contains no webhook or HTTP action; a different help page lists calling external services via API as the first thing a skill can do
- Unambiguous external writes exist through Workflows, including 20 named Salesforce actions, but a Workflow fires on a trigger rather than being chosen by the agent
- Hard ceiling of ten custom skills per agent
- A Runs section logs how each skill behaved in real conversations, which is better observability than most of this directory offers
Who Should Use It
The second list is the more useful one.
Worth a look if
You want a published seat price with a resolution allowance attached and a vendor with a two-decade trading history and audited accounts behind it. The skill model is genuinely easy for non-engineers, the per-run audit log is better than most, and the resolution definition will not surprise you on the first invoice. Small teams with a handful of well-defined automations are the fit.
Who should look elsewhere. Anyone whose central requirement is the AI writing to an external system of record. The external write is documented as a flow step, and the single page claiming the agent itself can call an API is contradicted by the page that enumerates every action it can take. Also wrong above ten distinct action-taking scenarios per agent, and wrong at volume: 200 included resolutions on a $79 seat, then $0.99 each, charged automatically.
Strengths and Weaknesses
What Works
- The resolution definition is the most honest in this directory, down to excluding no-answer handoffs and the ticket fallback.
- Twenty named internal actions with a per-run log you can inspect.
- A public company with real financials and SOC 2 Type 2, which is not nothing in a category full of two-year-old startups.
What Does Not
- Two help pages disagree on the most consequential capability question, and the one that lists every action has no webhook in it.
- The pricing page answers its own pricing model three different ways.
- The integrations FAQ names Zendesk and Zapier, neither of which is in the directory it points at, and the API-call FAQ contradicts the table above it.
- Two live brands with different customer counts.
ChatBot by Text pricing
Essential is $19 per user per month billed annually or $25 monthly, with 10 included resolutions. Growth is $79 or $99 with 200. Enterprise is quoted. Minimum one user, so there is a seat floor underneath the usage meter.
Overage is auto-refill, it is on by default, and nobody asks you first. Hit the monthly limit and Text adds 50 more resolutions for $49.50 so the agent keeps working. That is $0.99 a resolution, the same rate Intercom charges, arrived at through a mechanism you did not opt into. Reductions can only be made before renewal and count as a downgrade.
The full arithmetic
Then the pricing page contradicts itself three ways. One FAQ answer says “We don't charge per ticket or per resolution, so scaling support doesn't trigger surprise bills.” Another on the same page says “We charge per resolution, not per conversation.” A third says “You are not paying a usage meter every time your AI answers a question.” All three were live on 21 August 2026. The second one is the accurate one.
A smaller version of the same problem sits further down: the FAQ says Essential starts at 20,000 free API calls while the comparison table on the same page gives Essential 10,000 and Growth 20,000.
INTEGRATIONS
Our Recommendation
Who we would tell to buy this, and who we would not.
A solid, cheap, well-run small-team chatbot with the clearest billing definition we have read, sold by a company that cannot presently tell you whether its AI can call your API. Buy it for answering, with a seat price you can forecast. Do not buy it for acting until someone at Text reconciles two of their own help pages, and ask them to do that during the trial rather than after.
Where we write about ChatBot by Text
One piece on this site names it.
- Chatfuel ships four autonomy levels and starts every bot on the one where the AI writes nothing.Every vendor here will tell you their agent takes action, and almost none will say who approves it or where the record lands. One small vendor publishes a four-level autonomy slider the rest of the category should be measured against.
Similar Tools
Others in ai agents & chatbots.
Fin (formerly Intercom)
Teams already on Fin who can absorb variable per-resolution billing
4.2 / 5 · $0.99/resDecagon
Large support organisations that need the agent shaped to existing procedures
3.1 / 5 · Not publishedSierra
Enterprises wanting one vendor across chat and phone
3.1 / 5 · Not published
Ada
Global brands needing many languages and channels at once
3.5 / 5 · Not publishedFrequently Asked
The questions people actually search for about ChatBot by Text.
Can the ChatBot AI agent call my API?
Text's own documentation disagrees with itself. The Custom Skills vs Instructions page lists “Call external services via API (webhooks support)” as the first thing a skill can do. The Custom Skills page says it lists all the actions the agent can perform, enumerates twenty of them, and contains the word webhook zero times in 175KB. The external writes that are unambiguously documented run through Workflows, which fire on an event rather than being chosen by the agent. Treat the capability as unconfirmed and test it in trial.
What counts as a resolution?
The AI directly solving a customer question. Several things explicitly do not count: a chat where no issue is solved, an irrelevant answer, test, spam, greeting and browsing chats, and a transfer to a human where the agent had not answered anything first. Answer one question and then transfer, and you are billed. A ticket the customer asked for counts; a ticket created because no agent was available does not.
How much does it cost after the included resolutions run out?
$0.99 each, charged automatically. Auto-refill is on by default and adds 50 resolutions for $49.50 when you hit the cap. You can only reduce packages before renewal, and doing so counts as a downgrade. Essential includes 10 resolutions, Growth 200.
Is ChatBot.com the same company as Text?
Yes. LiveChat Software renamed itself Text in September 2023 and trades on the Warsaw Stock Exchange as TXT. Both domains are live, signup on either lands on accounts.livechat.com, and the two sites publish different customer counts, 35,000 brands against 30,000 teams as read on 21 August 2026.
How many custom skills can one agent have?
Ten. It is a hard cap, no plan raises it, and it is the constraint most likely to matter if you are mapping a real support taxonomy onto the product.
Does it integrate with Zendesk or Zapier?
The pricing FAQ says it does. The integrations directory lists twelve available integrations and neither is among them, though Zapier appears in the site footer. The same page also claims you can connect thousands of tools with webhooks, which returns you to the unresolved question above.