ChatBaser Review 2026: I Built, Trained and Embedded an AI Chat Agent

Tested: October 2026 | Test type: logged-in, hands-on setup, training, testing and website deployment | Project: Tools for Affiliate Marketing website assistant

I created a fresh AI chat agent inside ChatBaser, imported the BrandBaser knowledge document for my business, trained the chatbot, tested conversations in the playground, configured its website behaviour and embedded it on this blog.

The biggest result was not simply “a chatbot appeared.” I could keep the agent’s source material, conversations, contact capture, deployment controls and performance data in one workspace. That makes ChatBaser especially approachable for beginners who want a useful website assistant without assembling several separate tools.

Important testing rule: an AI assistant is only as dependable as its instructions and approved sources. Always test important questions yourself, review captured information responsibly and fact-check any answer involving prices, policies, guarantees, legal matters or time-sensitive details before relying on it.

My ChatBaser test at a glance

Agent created: Tools for Affiliate Marketing.
Knowledge imported: my BrandBaser document based on 27 business questions.
Additional source test: a text knowledge source of roughly 6.9K characters.
Testing: conversations were run inside the built-in playground.
Deployment: I configured a branded 400 × 600 pixel widget and embedded it on my blog.
Visibility: public access with embedding restricted to my chosen domain.
Lead capture: optional name, email and phone fields with a Skip choice.
Behaviour controls: delay, exit-intent and scroll-depth trigger settings.
Reporting: conversation, message, feedback and contact-capture metrics.

Testing boundary: this review covers the features I accessed and configured in my account. I did not test every advanced developer connection, voice feature or third-party integration. Features can change, so confirm any capability essential to your business before purchasing or deploying it.

What is ChatBaser?

ChatBaser is an AI-agent builder inside the Scale.gg ecosystem. You give an agent a focused purpose, add approved business knowledge, test the resulting conversations and then make the assistant available through a link or embedded website widget.

In practical terms, it can turn your brand documents, support information, product explanations and frequently asked questions into a conversational assistant. Instead of forcing visitors to search through many pages, the agent can help them find relevant information through natural questions.

This is different from opening a blank general-purpose AI chat. ChatBaser keeps the agent’s job, knowledge sources, test conversations, deployment settings and performance data together. You still decide what information it receives and you remain responsible for checking its answers.

Who is ChatBaser good for?

From my test, ChatBaser is a particularly good fit for:

Bloggers and affiliate marketers who want to guide readers towards useful articles, tutorials and product information.
Small businesses that repeatedly answer the same questions about services, availability, policies or next steps.
Course and community owners who want an assistant trained on onboarding and help material.
Agencies that want to create focused assistants for different brands or responsibilities.
Beginners who prefer a visual workspace for sources, testing and deployment rather than building a chatbot from code.
Support teams that want to review real conversations and find gaps in their help content.

It may be less suitable if you need a completely autonomous support system that never requires human supervision, or if a specific integration is essential and you have not yet verified that connection in your account. ChatBaser can help answer and organise questions, but it does not remove the need for accurate source material, escalation rules and human review.

How I created my first ChatBaser agent

Step 1: Name the agent and give it one clear purpose

I named my agent “Tools for Affiliate Marketing.” The setup asks for a description explaining what the agent does and how it helps visitors. This sounds simple, but the purpose statement matters: an assistant trained to guide readers around a review site needs different boundaries from one handling customer support or sales qualification.

Creating a ChatBaser AI agent and importing a BrandBaser knowledge document — Creating my test agent and selecting my BrandBaser document as its first knowledge source.

The optional BrandBaser connection was valuable in my workflow. My existing Brand Base contained answers to 27 questions about the business, audience, offer and positioning. ChatBaser makes clear that it imports the document content; visual Brand Kits and style guides are not automatically imported in this step.

Beginner tip: give each agent one main responsibility. “Help visitors understand our tested AI marketing tools and find the right guide” is clearer than “answer everything about our business.”

Step 2: Add approved knowledge and train the chatbot

The Sources area is where the agent receives the information it should use. In my test, the imported BrandBaser document appeared as a text source of about 6.9K characters, and the workspace showed the selected source size before I pressed the training button.

Training a ChatBaser chatbot with a selected text knowledge source — The source workspace showed my BrandBaser text knowledge before training the chatbot.

You can use direct text to supply precise facts, instructions and frequently asked questions. Keep sources focused and current. If a price, policy or product capability changes, update the source and retrain before expecting the agent to know the new information.

A useful knowledge pack might include:

a concise business and brand overview;
verified product or service descriptions;
frequently asked questions;
support and escalation instructions;
approved links and calls to action;
important boundaries—what the agent must not claim or advise.

If you need help building the business context first, read my Scale.gg guide, which explains how the connected app ecosystem fits together.

You can also compare this focused chatbot workflow with my ClickCampaigns.ai 18-page website test, where I used Scale.gg’s multi-agent campaign system to plan and generate a much larger marketing project.

Step 3: Test conversations in the playground

After training, I moved to the Playground. The workspace includes recent chats, so I could begin a fresh conversation and return to earlier tests. This is where you should deliberately ask easy, difficult and ambiguous questions before exposing the assistant to real visitors.

ChatBaser playground ready to start a new chatbot conversation — The playground provided a controlled place to start a new conversation and review recent chats.

I recommend testing at least these question types:

Direct fact: “What does this business help with?”
Recommendation: “Which guide should I read first?”
Boundary: ask about something absent from the approved sources.
Follow-up: refer to an earlier message without repeating the subject.
High-risk fact: ask about pricing or a policy, then compare the answer with the current source.
Escalation: ask a question that should be handed to a human.

When an answer is weak, do not merely rephrase the same question repeatedly. Identify whether the missing piece belongs in the knowledge source, the agent instructions or the website itself. Then correct that layer and retest.

Deploying and embedding the chatbot

The Deploy area was one of the strongest parts of my test because it turned the trained agent into something I could actually place on my website.

Choose the widget size and preview it

My account offered desktop and mobile previews along with standard, compact, large and custom sizing. The recommended standard option was 400 × 600 pixels, which gave enough room for a useful conversation without taking over the entire page.

ChatBaser deployment screen with responsive widget sizes and website preview — I previewed the standard 400 × 600 pixel widget before deploying it.

The same deployment area also provides the embed code and a link option. For beginners, this means you can configure the presentation visually and then place the generated code in a website area that accepts custom code.

Control public access and allowed domains

I set the agent to Public and restricted embedding to danielpopa.com. The visibility controls also showed a Private option for an agent that should remain accessible only inside the account. Domain restrictions are useful because they reduce the chance of somebody copying your embed to an unrelated site.

ChatBaser visibility settings for public or private access and allowed website domains — The agent was public, while embedding was limited to my selected domain.

The interface accepts exact domains, subdomains and wildcard subdomains. If you run the same assistant on a main domain and a separate blog subdomain, check that both are permitted before testing the live widget.

Configure load strategy and automatic opening

ChatBaser exposed script-loading choices including Async, Defer and Lazy. The interface marked Async as recommended, while I could also choose lazy loading. These settings can affect when the widget becomes available and how it competes with the rest of the page, so test both performance and usability rather than choosing solely by label.

ChatBaser performance settings with Async Defer and Lazy script loading strategies — The performance settings offered Async, Defer and Lazy script-loading strategies.

I also found three auto-open controls: a time delay, exit intent and scroll depth. The settings screen allowed a delay in seconds and a percentage for scroll depth. These are useful conversion tools, but enabling all of them aggressively can annoy visitors.

ChatBaser widget auto-open controls for delay exit intent and scroll depth — Auto-open behaviour can be triggered by a delay, exit intent or visitor scroll depth.

My recommendation: start with the launcher closed and let visitors open it themselves. After you have real usage data, test one gentle trigger at a time—for example, opening after meaningful scroll depth on a long guide.

The final embedded result

After adding the embed, the Tools for Affiliate Marketing assistant appeared as a floating chat button. Opening it displayed the branded header, greeting and question field over the blog page. This completed the full workflow from brand context to a functioning website assistant.

ChatBaser AI chatbot embedded and open on the Tools for Affiliate Marketing blog — The finished ChatBaser widget running on my BlogBaser-hosted website.

This blog itself was created with another Scale.gg application. You can see that workflow in my hands-on BlogBaser review.

Contact capture: useful, but configure it responsibly

The widget can request a visitor’s name, email address and phone number. In my account, the visitor could press Continue or Skip. That optional path is friendlier than blocking every conversation behind a form.

ChatBaser optional contact capture form for name email and phone — The contact form could request a name, email and phone number while still allowing the visitor to skip.

Before collecting personal data, update your privacy information and explain why the information is requested. Only request fields you genuinely need, protect access to the data and follow the privacy and consent rules that apply to your audience and location.

Conversation analytics and feedback

The analytics screen turned conversations into something I could review. My early test showed two conversations and four messages. It also displayed thumbs-up and thumbs-down counts plus contact-capture totals and capture rate.

ChatBaser analytics for conversations messages feedback and contact capture — My initial test activity appeared in the conversation and contact-capture dashboard.

The numbers are not meaningful performance results yet—the sample is far too small. Their value is that the reporting structure is already available. Over time, you can use repeated questions and negative feedback to decide what the agent’s knowledge or your website needs to explain more clearly.

Conversation and agent deletion controls

The Agent Settings area also contains a clearly labelled Danger Zone. It separates deleting all conversations from deleting the entire agent, and warns that both actions are irreversible. That distinction is useful, but it deserves care: export or record anything you need before using either control, confirm you are working on the correct agent and never treat a destructive button as a routine cleanup shortcut.

ChatBaser danger zone for permanently deleting conversations or an AI agent — ChatBaser separates permanent conversation deletion from deleting the complete agent.

What I liked most

Beginner-friendly workflow: create, add knowledge, train, test and deploy in a logical order.
BrandBaser handoff: I could reuse business context instead of retyping the foundation.
Real deployment controls: responsive sizing, preview, embed code, public/private visibility and domain restrictions.
Practical behaviour settings: load strategy and auto-open choices provide useful control.
Optional contact capture: the visitor can skip rather than being forced to submit details.
Review loop: conversations, feedback and capture metrics help turn questions into improvements.

What to plan before using it

Prepare accurate, focused source material before training.
Write clear boundaries for topics the agent should not answer.
Test pricing, policies and important links manually.
Decide when a visitor should be directed to a person.
Update your privacy notice before enabling contact capture.
Test the widget on desktop and mobile after embedding.
Review conversation history regularly rather than treating deployment as the finish line.

ChatBaser pricing and access

The official Scale.gg product information listed ChatBaser as included with the Pro plan when I reviewed it in October 2026. Scale.gg plans, credit allowances and included applications can change, so use the current checkout page as the final source of truth.

Affiliate disclosure: the button below uses my affiliate link. If you purchase through it, I may earn a commission at no extra cost to you. This does not change how I test or describe the platform.

See the current Scale.gg plans and ChatBaser access →

My verdict

My first ChatBaser test was a positive one. I went from a new agent to an embedded website assistant using my real business context, not a fictional demo. The most useful part is the connected workflow: BrandBaser supplied the foundation, ChatBaser turned that knowledge into conversations, and the live widget made the result available to readers.

For a beginner, that removes much of the technical uncertainty around creating a branded AI assistant. The work you still need to do is the important work: choose accurate sources, define sensible boundaries, test difficult questions and keep improving the agent from real conversations.

I will continue testing the chatbot with more visitor questions and will update this guide when I have enough real activity to report meaningful results rather than small-sample numbers.

Frequently asked questions

What is ChatBaser?

ChatBaser is a workspace for creating AI chat agents grounded in approved business knowledge. It combines agent setup, sources, testing, deployment and conversation reporting.

Do I need to code?

The main setup is visual. To place the chatbot on a website, you need somewhere that accepts the provided embed code. The exact installation step depends on your website platform.

Can ChatBaser use my BrandBaser document?

Yes. In my test I selected my BrandBaser knowledge document while creating the agent. The interface stated that document content is imported, while visual Brand Kits and style guides are not imported through that specific step.

Can I keep an agent private?

Yes. My visibility settings offered Public and Private options. Public agents could be made available through a link or embed, while Private restricted access to the account.

Can I restrict where the chatbot is embedded?

Yes. I restricted embedding to danielpopa.com. The settings also explained support for exact domains, subdomains and wildcard subdomains.

Does it capture leads?

My account included optional fields for name, email and phone, plus a Skip button. Configure this responsibly and make your privacy information easy to find before collecting personal details.

Can I control when the widget opens?

The tested settings included opening after a delay, on exit intent or after a selected scroll depth. Start conservatively and measure the visitor experience before enabling aggressive triggers.

Does ChatBaser show analytics?

Yes. I saw totals for conversations and messages, thumbs-up and thumbs-down feedback, and contact-capture metrics. Treat early numbers as setup confirmation, not proof of business performance.

Can AI answers be wrong?

Yes. Any AI system can misunderstand a question, use incomplete context or repeat outdated information. Keep sources current, define boundaries, test important questions and verify consequential facts with an authoritative source.

How do I get ChatBaser?

ChatBaser was listed as part of Scale.gg Pro when this review was prepared. Check the current Scale.gg plans for the latest access, pricing and inclusions.