Airabot

Businesses want to answer customer questions instantly across the channels their customers already use, but building an AI chatbot usually means writing code, managing infrastructure and accepting the risk that the model will make things up. Airabot set out to remove those barriers. The goal was a platform where anyone could train a bot on their own documents and URLs, deploy it to a website or WhatsApp, and trust that its answers stay grounded in their content. It also needed to serve agencies and SaaS teams managing many clients, so it had to work as a multi-tenant product with clear, usage-based billing.

We built Airabot as a platform-as-a-service around a simple three-step flow: upload knowledge, configure the bot, deploy everywhere. Retrieval-augmented generation sits at the core. Each bot gets its own vector index, and answers are drawn only from the content the customer has supplied. We kept setup code-free. A single script tag adds the bot to any website, and connecting a WhatsApp number takes the same channel live. For agencies, we designed the platform to handle unlimited clients under one account. We also supported bring-your-own-key (BYOK) so teams can use their own AI provider credentials. Billing follows a straightforward credit model: one credit per message, on a 30-day rolling window.

• Knowledge ingestion: upload PDFs, Word documents and text files, or paste URLs for automatic crawling, chunking, embedding and indexing per bot
• Bot configuration: set the name, persona, colour, welcome message and AI model, plus FAQ shortcuts for guaranteed instant answers
• Website chat widget: deployed with one script tag on any site
• WhatsApp integration via Twilio: includes voice note transcription so customers can ask questions by speaking
• Multi-bot, multi-channel, multi-tenant management: suited to agencies, SaaS teams and enterprise
• Usage-based credit system: transparent billing at one credit per message