Meta Tech ProviderChat AI Agent
Build chat AI agents,
no code required
ThinnestAI is a chat AI agent builder built in India for global teams. Configure an agent in a tabbed settings screen, ground its answers in your knowledge base (site crawl, sitemap, pasted text, uploaded files) with source citations, call tools in your CRM and APIs, reply in all 22 scheduled Indian languages, and run the same agent on the phone.
What you can build
Chat AI agents, LLM chatbots with tool calling, RAG-grounded help centers, WhatsApp bots, internal co-pilots. One agent, every surface.
Use cases
What you can build
Customer support
24/7 bilingual support across chat, WhatsApp and voice — with ticket handoff to humans when needed.
Knowledge base answers
Upload your product docs, help center articles, PDFs or website. The agent answers only from your content, with a link to the exact source on every reply.
Agent with tools
Connect to CRMs, databases and internal APIs over MCP — the agent uses tools to do real work, not just answer questions. Consent is per tool, off by default.
Multilingual bot
One agent, 37 reply languages including all 22 scheduled Indian languages, each in its native script. Hinglish code-switching handled natively. No per-language surcharge.
Lead qualification
Qualify inbound leads via chat widget, hand warm leads to sales with full context.
Internal co-pilots
Give your team a chat agent over internal docs and tools. Works on the web widget, WhatsApp and Telegram.
Features
Everything you need
01
Knowledge base with citations
Crawl your website 1–3 levels deep, import a sitemap, paste text, or upload txt, md, csv and text-layer PDFs. The agent grounds every answer in retrieved context with source links.
02
37 reply languages
All 22 scheduled Indian languages (Hindi, Marathi, Tamil, Telugu, Bengali, Gujarati, and more) in native script, plus global languages. One knowledge base serves every one of them.
03
Unified voice + chat
Same agent logic powers the web widget, WhatsApp, Telegram and voice phone calls — no duplication.
04
Configured in settings, not code
An agent is a tabbed settings screen — prompt, knowledge, channels, actions, templates, voice and campaigns. For product teams, not just engineers.
05
Curated LLM providers
Models are chosen and managed by the platform and covered by your rate — no keys to configure, nothing to wire up.
06
Tool calling
Connect to your CRM, database, REST APIs, webhooks. The agent uses tools to take real actions, not just chat.
Models
A curated model stack
Pick a model per agent from the set we run and maintain — platform-managed, no keys to configure, no provider bills to reconcile, no hidden extras.
OpenAI GPT-4o
Anthropic Claude
Groq GPT-OSS
Google Gemini
Sarvam LLM
Mistral
DeepSeek
FAQ
Frequently asked questions
Can the chat agent see and fill in forms on my website?
+
Yes, with the In-App AI Agent (/in-app-ai-agent) switched on. The agent sees the screen your customer is on — the form, its errors, the buttons — fills in fields, points at the next step and steps in when someone gets stuck. Field values stay on the customer's device unless you allow them, and anything that pays or submits waits for the customer's Allow.
What's the difference between a chat AI agent and a regular chatbot?
+
A regular chatbot follows scripted decision trees — user clicks a button, bot follows a branch. A chat AI agent uses an LLM to understand natural language, ground its answers in your knowledge base via RAG, call tools (CRM, database, APIs) to take real actions, and maintain memory across a conversation. ThinnestAI builds the second kind.
Can I build a WhatsApp chatbot on ThinnestAI?
+
Yes. ThinnestAI supports WhatsApp Business API integration — the same agent logic that powers your web chat widget can run on WhatsApp. We have a dedicated guide at /whatsapp-ai-agent for WhatsApp-specific deployments.
Which LLMs can I use?
+
A curated, platform-managed set — you pick one per agent and we run it, so there are no keys to configure and no provider bills to reconcile. The free tier includes Prana LLM and gpt-4o-mini; premium models are on the paid tiers.
Does ThinnestAI support RAG (retrieval augmented generation)?
+
Yes. Crawl your website one to three levels deep, import a sitemap, paste text, or upload txt, md, csv and text-layer PDF files. The platform chunks, embeds and retrieves relevant context for every message, and the LLM cites the source in its answer. No vector database configuration required. There is no OCR, so scanned PDFs will not import.
Can one agent handle Hindi, English and Hinglish?
+
Yes. Our routed providers handle Hinglish code-switching natively — no pre-processing or segmentation required. Indian users switch languages mid-sentence and the agent follows along the way real conversations actually work.
What's the pricing for chat-only agents?
+
The flat voice rate — ₹2/min on standard voices, ₹2.5/min on premium voices, telephony included either way — applies to voice agents. Chat replies are usage-based, not a flat per-message rate, with 200 free replies to cover most small-business trials. Talk to us for enterprise volume.
Is there a no-code builder or do I need to write code?
+
You don't need to write code. An agent is configured in a tabbed settings UI — prompt, knowledge, channels, actions, templates, voice and campaigns — which is the fastest path for product teams. Engineering teams can also drive the platform over our REST API and receive signed webhooks for leads, escalations, resolutions and finished calls or campaigns.
Can the chat agent also make phone calls?
+
Yes. ThinnestAI is unified — the same agent definition can power a chat widget AND a phone voice agent. When voice is needed, we route audio through LiveKit over a number you buy from us or an imported Plivo, Vobiz, Twilio or Telnyx number, and hand audio to the same underlying LLM and tools.
Start building your first chat AI agent
Free trial — 200 chat replies and 25 voice minutes, no card required. Deploy to the web widget, WhatsApp, Telegram or voice.