Chatbase review

Chatbase is a hosted SaaS platform for building and operating customer-facing AI agents. It is not presented as a software library, command-line tool or self-hosted server. Teams configure agents in a web dashboard, connect data sources, define instructions and guardrails, then publish agents to customer channels.

The product sits between a company’s knowledge sources and its customer-facing support or sales interfaces. Its agents can answer questions, use built-in Actions such as order lookup, and route work into a shared helpdesk where AI and human agents handle conversations. The homepage also describes reporting on topics, sentiment and conversation trends, plus a model playground for testing models and settings.

That makes Chatbase an application-layer service rather than an infrastructure component to deploy inside your stack. It is aimed at teams that want to configure an agent without writing code. Developers can connect through the listed API access on the Basic plan, but should verify the API’s scope and integration model before designing around it.

The product is configured through Chatbase’s dashboard: connect data sources, write agent instructions, set guardrails, test scenarios and publish to selected channels. The homepage lists deployment across chat, WhatsApp, email and Slack, and describes integrations with CRMs, helpdesks and other services. Built-in Actions let an agent perform tasks such as retrieving invoices or order information.

The Basic plan lists API access. The homepage does not name an API style, SDKs, supported programming languages, a CLI, webhooks or plugins. Do not assume a particular client library or request format: confirm API endpoints, authentication, event handling and SDK support in the vendor’s docs before implementation. The product also includes a model playground for testing models and settings, but the homepage does not document a code-based testing workflow.

This is a dashboard-first integration surface, with API access as a plan feature. Teams building custom applications should verify whether the API supports their required agent, conversation and helpdesk operations, rather than assuming the full dashboard is exposed programmatically.

Chatbase has no free plan. Its free trial lasts seven days. Hobby costs $32 per month when billed yearly, or $40 month to month, and includes 700 message credits, 200 test credits, one agent and two workspace seats. Basic is $120 per month billed yearly or $150 monthly, with 4,000 message credits, three seats, API access and 10 concurrent calls. Business costs $400 yearly-billed monthly or $500 month to month, with 15,000 message credits, five seats and 20 concurrent calls. Enterprise pricing is custom.

The bill can rise through additional agents at $25 per agent per month, extra storage at $25 per 10 MB per month, or branding removal at $99 per month. Auto-recharge credits cost $40 per 1,000 message credits. The pricing page does not state a separate overage rate beyond that add-on. Check how your chosen model and agent configuration consume message credits before estimating production volume; the listed plans do not translate credits into a fixed number of conversations.

Chatbase is sold as a hosted service. The vendor does not list an open-source licence or a self-hosting option in its product and plan details, so teams should treat it as hosted-only unless the vendor confirms otherwise. There is no listed path to run the agent platform inside your own infrastructure.

The vendor also does not describe data export or migration procedures in the published product details. Before committing, ask how to export source data, conversation history, agent instructions, Actions and analytics, and what formats are available. Those details determine whether you can rebuild the agent elsewhere without manually recreating its configuration.

Lock-in risk is highest when the agent depends on Chatbase-specific configuration, Actions and helpdesk workflows. Keep source knowledge and business logic in systems you control where possible, and document prompts, guardrails and integrations outside the dashboard. Confirm contractual data access and deletion terms before using customer information in production.

The plan details publish concurrent-call limits for Basic and Business: 10 and 20 respectively. The vendor does not list general API rate limits, uptime figures or a service-level commitment for those plans. Enterprise includes SLAs, SSO, audit logs and HIPAA eligibility, according to the pricing details; teams with compliance requirements should confirm the scope and contractual terms directly.

Credit consumption needs monitoring: one Reddit post alleges that model choice can sharply increase credits per response, while another commenter disputes that account and says the public credit breakdown is available. Treat both as user reports, not a reliable calculation; ask the vendor how your model and features consume credits. A separate Reddit thread reports sitemap links failing to train, and comments include complaints about support responsiveness. Another commenter reports that the agent answered many product questions. Test ingestion against representative pages, measure resolution on difficult cases and set a human escalation path before rollout.