WebApi.ai

Powers multi-turn conversations using GPT-3 or GPT-4o language models with context retention across dialogue turns. The system maintains conversation state and applies custom domain knowledge injected via document uploads (PDF, DOCX, CSV) to ground responses in business-specific information. Dialogue scenarios enable sample-based learning where builders define conversation flows and expected outcomes, which the model uses to adapt response patterns.

Accepts incoming messages from 8+ communication channels (website widget, Instagram, Facebook Messenger, WhatsApp, Telegram, Twilio SMS, Twilio WhatsApp) and routes them to a unified chatbot backend. Each channel integration handles protocol-specific authentication and message formatting, converting diverse input formats into a normalized message schema for the conversational engine. Channel-specific response formatting ensures replies are adapted to each platform's constraints (e.g., character limits, media support).

Enables chatbots to invoke external APIs and trigger business logic in response to user intents. The system supports outbound API calls to customer systems (e.g., booking confirmations, order modifications, ticket cancellations) and integrates with Zapier and Pabbly for no-code workflow automation. Builders can define action mappings in the UI (e.g., 'when user asks to cancel order, call /api/orders/{id}/cancel'), and the chatbot automatically extracts parameters from conversation context and executes the call. Response handling allows conditional follow-up messages based on API success/failure.

Accepts business documents (PDF, DOCX, CSV, website pages, articles) and indexes them for retrieval during conversations. The system extracts text from uploaded files, chunks content into retrievable segments, and uses semantic search or keyword matching to surface relevant passages when the chatbot needs to answer user questions. Retrieved passages are injected into the LLM prompt as context, grounding responses in authoritative business information. Supports knowledge bases from Zendesk KB and Intercom KB via API integration.

Allows builders to define conversation flows and expected outcomes via 'dialogue scenarios' — sample conversations that teach the chatbot how to handle specific user intents. Each scenario includes example user messages, expected chatbot responses, and desired actions (e.g., 'when user says they want to cancel, extract order ID and trigger cancellation API'). The system uses these scenarios as few-shot examples or fine-tuning data to adapt the base LLM's behavior without requiring prompt engineering or model retraining. Scenarios are stored in the builder UI and applied to all conversations.

Automatically classifies user messages into predefined intent categories (e.g., 'product inquiry', 'support request', 'sales lead', 'complaint') and extracts structured data (name, email, phone, company, budget) from conversations. The system uses the base LLM to perform intent classification and entity extraction, optionally routing qualified leads to human agents or CRM systems via API integration. Tutorial references a 'Lead Qualifier chatbot' template, suggesting pre-built classification schemas for common use cases.

Triggers email notifications to business users based on chatbot events (e.g., new lead captured, support ticket created, order cancellation requested). Builders can define email templates and conditions in the UI (e.g., 'send email to sales@company.com when a qualified lead is captured'). The system supports dynamic content injection from conversation context (e.g., customer name, email, inquiry details) into email templates. Emails are sent via WebApi.ai's mail service or integrated with external email providers.

Provides a 14-day free trial with limited quotas (500 article views, 1 admin user) to allow businesses to test the platform before committing to paid plans. Paid tiers use usage-based pricing (exact unit unclear from documentation — appears to be per-token or per-request, ranging $0.15-$4 per unit). The system enforces quotas at runtime, preventing chatbot operations when limits are exceeded. Pricing varies by model selection (GPT-4o vs Llama 3.2), with higher-cost models available on paid tiers.

Provides a web-based UI for non-technical users to design chatbots without writing code. The builder includes drag-and-drop workflow editors for defining conversation flows, intent handlers, and action triggers. Users can configure channels, upload knowledge documents, define dialogue scenarios, and set up API integrations through forms and visual editors. The UI abstracts technical complexity (API authentication, prompt engineering, model selection) behind user-friendly controls, allowing business teams to deploy chatbots independently.

Provides basic analytics on chatbot performance including conversation volume, user engagement, and intent distribution. The dashboard displays metrics like total conversations, average conversation length, and top intents. Trial and paid plans include analytics access, though the editorial summary notes that WebApi.ai lacks 'advanced analytics and conversation insights needed for serious customer support optimization' compared to competitors.