Skip to main content

Documentation Index

Fetch the complete documentation index at: https://tinytalk.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

Stay up to date with the latest features, improvements, and fixes.
Release notes are published here as new features ship. Check back regularly for updates.

​
May 2026

  • Cal.com tool — Added Cal.com as a Platform Tool. Paste your cal.com or cal.eu event-type URLs and the agent offers the right one when a visitor wants to book. Renders inline on the website widget; posts a prefilled link on WhatsApp and Slack. EU-hosted cal.eu URLs are supported for data residency requirements. See Cal.com.

​
April 2026

  • Calendly tool — Added Calendly as a Platform Tool: paste your event-type URLs and the agent offers the right one when a visitor wants to book. Renders inline on the website widget; posts a prefilled link on WhatsApp and Slack. See Calendly.
  • Tools — Agents can now call HTTP endpoints mid-conversation. Custom Tools are the first tool type; point Tiny Talk at your own HTTPS endpoint, define parameters and secrets, and the agent invokes it when relevant. Available on all paid plans. See Tools.
  • Thinking steps in the chat — Agents now surface reasoning tokens and tool usage as live “thinking steps” in the messenger while a reply is being generated, where the selected model supports it. Visitors get visible progress during longer responses.
  • Move agents between workspaces — Owners can now move an agent from one workspace to another they own without recreating it. The agent keeps its ID, API keys, integrations, knowledge base, conversations, and historical data. See Moving an agent between workspaces.
  • Reasoning controls on reasoning-capable models — Added Reasoning Effort (none, minimal, low, medium, high, xhigh) and Reasoning Summary (auto, concise, detailed) settings for agents using reasoning-capable models from OpenAI, Google, xAI, DeepSeek, and Groq. Tune the trade-off between answer quality, latency, and credit cost. See Reasoning configuration and the full list.
  • Deprecated models removed — The following models have been deprecated and removed from model selection: o1 Mini, Claude 3.5 Sonnet, Claude 3 Opus, Gemini 3 Pro Preview, Gemini 2.0 Flash, Gemini 1.5 Pro, Gemini Flash 1.5, and Llama 4 Maverick (Groq).

​
February 2026

  • Documentation site launched — Comprehensive docs covering all Tiny Talk features, integrations, and API reference.
  • Notion integration — Import pages directly from your Notion workspace into your Knowledge Base. Connect via OAuth, select pages, and keep your agent up to date with your team’s docs.
  • Website crawler improvements — New crawl modes, concurrency controls, and job status tracking for more reliable and faster website crawling.
  • AI model updates — Added new models and adjusted credit costs for better value across providers.

​
January 2026

  • Slack integration — Connect your agent to Slack workspaces. Your agent responds to messages in Slack channels and threads, with support for message reactions and conversation history.
  • Google Drive integration — Import files directly from Google Drive into your Knowledge Base via OAuth. Supports documents, spreadsheets, and presentations.
  • New AI models — Added Google Gemini models, Llama 4 Maverick via Groq, and updated model paths across providers.

​
December 2025

  • Push notifications — Receive real-time notifications in the dashboard when visitors request human support or send new messages. Powered by FCM with automatic subscription management.
  • Groq provider support — Added Groq as a third inference provider for ultra-fast response times. Connect your Groq API key in Settings to access supported models.
  • Analytics improvements — Faster analytics loading with server-side caching. Improved data accuracy and response handling across all dashboard charts.
  • New webhook events — Added conversation.created and takeover.request webhook events, giving you more visibility into conversation lifecycle.
  • Llama 3.1 8B Instruct — Added via OpenRouter for a lightweight, cost-effective model option.

​
November 2025

  • Workspaces & team collaboration — Organize your agents under workspaces and invite team members with role-based access control. Four roles available: Owner, Admin, Editor, and Viewer.
  • Seat management — Manage team seats from your workspace settings. Invite members by email with customizable roles, resend or cancel invitations, and track active seats against your plan limit.
  • Credit system — Introduced monthly credits that reset each billing cycle. Different AI models consume different credit amounts (1, 2, 10, or 20 credits per response), giving you flexibility to balance cost and capability.

​
October 2025

  • Rate limiting — Protect your agent from spam and abuse with per-user rate limits. Configure the maximum number of messages and time window per user, with customizable feedback messages and localization support.
  • WhatsApp rate limiting — Rate limits now apply to WhatsApp conversations, with localized rate-limit replies and per-channel tracking.
  • Knowledge Base search — Search across your Knowledge Base resources by name to quickly find and manage content.
  • Document processing improvements — Improved reliability for document ingestion with better retry mechanisms and background processing for large uploads.

​
September 2025

  • GPT-5 models — Added support for OpenAI GPT-5 and Azure-hosted GPT-5 models with EU data residency options.
  • Azure EU deployment — Select Azure-hosted OpenAI models with data processed in EU/DE regions for improved data residency compliance.
  • Credit cost per model — Each AI model now displays its credit cost, making it easier to choose the right balance of performance and cost.
  • WhatsApp contact profiles — WhatsApp visitor names now appear in your inbox and contact list, making it easier to identify and follow up with leads.

​
August 2025

  • Analytics event tracking — Message-level analytics tracking across web and WhatsApp channels. View detailed conversation metrics including response times, channel distribution, and usage patterns.
  • Improved bot authentication — Enhanced auth flow to better protect private agents and prevent unauthorized access through contact tokens.
  • Resource management — Improved bulk deletion with background processing, embedding status filtering, and pagination for Knowledge Base resources.

​
July 2025

  • Custom domains — Serve your chat widget from your own subdomain (e.g., chat.yourdomain.com) with automatic SSL. Available on supported plans with TXT record verification.
  • GPT-4.1 series models — Added OpenAI’s GPT-4.1, GPT-4.1 Mini, and GPT-4.1 Nano models.
  • Webhook: message.created — New webhook event fires when your agent sends a reply, enabling real-time integrations with external systems.
  • Contact context in conversations — Agents can now access contact information during conversations for more personalized responses.

​
June 2025

  • Reasoning model support — Added support for OpenAI o1, o1-mini, and o3-mini reasoning models for tasks requiring deeper analysis and multi-step problem solving.
  • Improved domain management — Dynamic origin management for the chat widget with real-time domain verification and cleanup.

​
May 2025

  • Lead form improvements — New contact.lead.updated webhook event fires when a lead submits or updates their information, replacing the previous contact.created event for better integration with CRMs and automation tools.
  • Messenger spaces — Set a default space to launch when the widget opens, directing visitors to specific content or flows.
  • Localization enhancements — Improved bot creation to include localization data from the start, ensuring consistent multi-language support.

​
April 2025

  • API key validation — Verify your API keys directly from the dashboard before going live. Instant feedback on whether your keys are correctly configured and have the required permissions.
  • Knowledge Base character limits — Clearer error messages when approaching storage limits, with improved resource management and cleanup.

​
March 2025

  • Improved document processing — Enhanced document ingestion pipeline with better support for large files, improved chunking, and more reliable embedding across all supported file types (PDF, Word, Excel, PowerPoint, CSV, Markdown, plain text).
  • Date-aware responses — Agents now include the current date in their context, enabling accurate responses to time-sensitive questions.
Follow us on X for announcements.