> For the complete documentation index, see [llms.txt](https://docs.dapta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dapta.ai/news/september-2026.md). # September 2026 What changed in Dapta in September 2026: campaigns became Sequences, Dapti sets up calling sequences, WhatsApp lines connect directly through Meta, and Agentic Warm Transfer works on every Voice Agent What reached Dapta in September 2026, grouped by product. Each change links to the page where it is documented in full. * **Campaigns are now Sequences**, and Dapti can set up a calling sequence for you. * **WhatsApp lines connect directly through Meta**, including the number you already use in the WhatsApp Business app. * **Agentic Warm Transfer works on every Voice Agent**, waits through hold queues, and lets you read the screening call. * **Text Agents book appointments with Dapta Calendars**, and you choose what customers get when a person takes over. *** ## Sequences **Campaigns are now called Sequences.** You find them under **Sequences** in the menu, as **Voice Sequences** and **Text Sequences**, in English and Spanish. Only the name changed: your sequences, their settings and their results work as before. → [Sequences (Voice Agents)](/ai-voice-agents/how-to-create-your-first-campaign-in-dapta.md) · [Sequences (Text Agents)](/ai-text-agents/campaigns.md) **Dapti sets up an outbound calling sequence for you.** Tell it you want to call a list of contacts. It walks you through the voice agent, the dates, the retries and the calling hours, calls your test contacts so you can check the agent, and starts the sequence only when you confirm. You can also ask Dapti to edit, start, stop or report on your sequences. → [Create a sequence](/ai-voice-agents/how-to-create-your-first-campaign-in-dapta/create-a-campaign.md) **See and retry the call fallback of each contact.** In a text sequence with call fallback, the conversations table has a **Call fallback** column (**Scheduled**, **Calling…**, **Call made**, **Answered**, **Not called**, **Could not call**) and the sequence metrics add a **Call Fallback** summary. For a contact whose call was not placed or failed, **Retry** calls again right away, even outside the sequence's hours, and never resends the WhatsApp message. → [Call fallback](/ai-text-agents/campaigns/call-fallback.md) **Run a text sequence with no end date.** Leave **End Date (Optional)** empty when you create a sequence and it keeps running and waits for new contacts, which suits a sequence fed by a webhook. You can also clear the end date of an existing sequence from its detail page. → [Create a sequence](/ai-text-agents/campaigns/create-a-campaign.md) **The sequence builder shows what you're missing.** If you start a sequence without a WhatsApp channel or an approved template, the step says so and takes you there with **Connect a channel** or **Create a template**. The prompts only appear when that piece is actually missing. → [Go live on WhatsApp](/text-agents-first-sequence.md) *** ## WhatsApp through Meta **New WhatsApp lines connect directly through Meta.** New lines connect from the **WhatsApp** card in **Text Agents → Channels**: both a standard WhatsApp Business API number and a number already in use on the WhatsApp Business app. → [Connect your Agent with WhatsApp](/ai-text-agents/channels/connect-your-agent-with-whatsapp.md) · [Prepare your Meta account](/ai-text-agents/channels/prepare-your-meta-account.md) **Lines connected the previous way now live under WhatsApp (legacy).** They keep working exactly as they do today, on a single Dapta invoice. You do not have to reconnect. → [WhatsApp billing: legacy and direct lines](/ai-text-agents/channels/whatsapp-billing-and-meta-charges.md) **You can keep the WhatsApp Business app on the phone.** Your team goes on replying from the phone while an agent works the same number, and both show up in Inbox. If Meta's own assistant is switched on for that number, turn it off before connecting. → [Prepare your Meta account](/ai-text-agents/channels/prepare-your-meta-account.md) **Your line is protected while a sequence runs.** Dapta watches quality, holds daily sending limits, warms up a new line gradually, respects opt-outs, and pauses marketing by itself when the evidence says the line is at risk, with the reason shown so you can fix it and resume. → [Message logic and limits](/ai-text-agents/campaigns/message-logic-and-limits.md) **A clear split on direct lines.** Dapta charges for the agent, Meta charges for delivering the messages to your own Meta account. Nothing is charged twice. → [WhatsApp billing](/ai-text-agents/channels/whatsapp-billing-and-meta-charges.md) · [How credits work](/my-account/how-credits-work-in-dapta.md) **Message templates: import, sync and carousels.** Bring in the templates your Meta account already has, refresh their approval status from Dapta, and build carousel templates with up to ten cards. → [WhatsApp message templates](/ai-text-agents/channels/message-templates.md) · [Which template type do you need](/ai-text-agents/channels/template-types.md) **Number status in plain words:** *Active*, *Needs an agent*, *Setup incomplete*, *Reconnect needed* and *Blocked by Meta*, each with the fix it needs. → [Connect your Agent with WhatsApp](/ai-text-agents/channels/connect-your-agent-with-whatsapp.md) **Template header files from a public URL.** A template's image, video or document header can point at a public `https` address instead of an upload. → [WhatsApp message templates](/ai-text-agents/channels/message-templates.md) *** ## Text Agents and Inbox **Text Agents book appointments with Dapta Calendars.** Since the end of September, add **Schedule appointments (Dapta Calendars)** to a Text Agent and pick the meeting. The agent offers only times that are really free and books the one your customer confirms, without sending a booking link. Voice Agents got the same action in October. → [Text Agents: Schedule Appointments with Dapta Calendars](/ai-text-agents/setting-up-your-text-agent/schedule-appointments-with-dapta-calendars.md) **Choose what customers get when a person takes over.** In the **Transfer to a human** action, **Message to the customer at handover** can be **Silent**, a **Fixed message** you write, or **AI-generated**, a short notice that knows whether an advisor was assigned. The AI stops replying as soon as it hands over and sends your **Unavailability message** if no advisor is free. Transfer actions you already have stay **Silent** until you choose. → [Text Agent Actions](/ai-text-agents/setting-up-your-text-agent/text-agent-actions.md) **Filter the Inbox by several statuses and send longer replies.** The **Status** filter takes several statuses at once and starts on **Open**, and **Clear filters** shows every conversation. Your WhatsApp replies can run to 4,096 characters, and long messages, yours and the AI's, arrive in order as consecutive messages with no numbering. → [Inbox](/ai-text-agents/inbox.md) **Find chats your customers haven't answered yet.** The Inbox **Status** filter has **Awaiting first reply** for conversations your business started that the contact hasn't replied to, with a banner that explains the state. These chats also appear by default in chat history and exports, so a conversation that opened and stalled is no longer missing from your records. → [Inbox](/ai-text-agents/inbox.md) **See how many messages your agent sent in a chat.** **Chat History** has an **AI Messages** column next to **Messages**, in both the sessions and the conversations views. It counts only the messages the agent wrote, so you can tell a chat the agent carried from one where the customer did most of the writing. → [Analytics, logs, and exports](/ai-text-agents/analytics-and-logs.md) **Set the reply timeout in hours and minutes.** **Close after no reply for** now takes separate **Hours** and **Minutes**, from 10 minutes to 24 hours, and each **Conversation behavior** switch sits on its own row with its description. New Text Agents start with **Reply in several short messages** and **Show a typing indicator while replying** on. Agents you already have keep their settings. → [Conversation behavior](/ai-text-agents/setting-up-your-text-agent/conversation-behavior.md) **AI follow-ups on WhatsApp can carry a file.** A follow-up with the **AI Generated** content type can attach one file, such as a PDF or an image, with a short caption. Mention the file and its public link in the **AI Prompt** to steer it. Each follow-up is attempted once and isn't sent if a person takes the conversation over. → [Follow-ups](/ai-text-agents/setting-up-your-text-agent/follow-ups.md) **Knowledge sources that say why they were rejected.** A knowledge source that can't be read now fails at once, with the file name and the reason, instead of staying in processing. → [Brains](/ai-voice-agents/how-to-set-up-your-ai-voice-agent/brains.md) **Worked cost examples per use case.** Each of the six Text Agent use cases now carries a monthly estimate, so you can size your own before you build. → [What people use Text Agents for](/ai-text-agents/use-cases.md) *** ## Voice Agents **Agentic Warm Transfer works on every voice agent, and you can read the screening call.** Choose it in any **Transfer Call** action, including on agents that call through your own SIP trunk, and use any of your voice agents as the screening agent. The call detail shows the **Screening conversation** turn by turn, how long each action took, and the audio if you turn on **Record the conversation with the advisor**, which is billed as an additional recording. → [Transfer Call](/ai-voice-agents/agent-actions/transfer-call.md) **Agentic warm transfers wait for a person.** The screening agent says hello on a silent line and waits through hold queues and hold music instead of taking them for a voicemail. The wait for a decision goes up to 300 seconds. → [Transfer Call](/ai-voice-agents/agent-actions/transfer-call.md) **Decide who can open a voice agent's calls.** Turn **Public call link** off under **Settings** → **Call Settings & Behavior** and that agent's call pages open only for signed-in members of your workspace; anyone else is told the call is private. Leave it on and anyone with the link can listen to the recording and read the transcript. → [Public call link](/ai-voice-agents/how-to-set-up-your-ai-voice-agent/voice-agent-settings-guide/configure-call-public-call-link.md) **Recording links follow the Public call link switch.** In the post-call webhook, notification emails and the call-log export, a public agent's recording opens at `files.dapta.ai`, and a private agent's opens inside Dapta after sign-in. → [Public call link](/ai-voice-agents/how-to-set-up-your-ai-voice-agent/voice-agent-settings-guide/configure-call-public-call-link.md) **A daily email with your voice agent's results.** In the agent's **Call Analysis** tab, turn on **Scheduled email report**, choose the **Time to send**, **Time zone** and **Report day**, and add recipients under **Send to**. It sums up that agent's calls, the credits used and a few AI-written insights, and **Send Test Email** shows you a sample first. → [Post Call Analysis: Notifications](/ai-voice-agents/analyze-your-calls/post-call-analysis-notifications.md) **Set the time zone a voice agent works in.** In **Agent Context**, the optional **Timezone** field below **Purpose** gives the agent its local date and time, so greetings and dates like "tomorrow" or "Friday" come out right. Leave it empty and the agent keeps using US Pacific Time, as before. → [Go live with calls](/voice-agents-first-sequence.md) **Encoded SIP headers arrive readable.** On a call that comes in over your SIP trunk, a User-to-User header, or any header that declares hex or base64 encoding, also gives your agent its plain text in `{{sip_header__decoded}}`. In a transfer's **Custom SIP Headers**, add `|decode` to a value you know is encoded to pass it on as text. → [Transfer Call](/ai-voice-agents/agent-actions/transfer-call.md) **Two more AI models for Voice Agents: GPT 5.6 Luna and Gemini 3.8 Flash.** Pick one under **Settings** → **Voice Settings** → **LLM Model** and save. Agents you already have keep their model until you change it. → [Dapta Auto for Voice Agents](/ai-voice-agents/how-to-set-up-your-ai-voice-agent/dapta-auto.md) **Remove a phone number that failed verification yourself.** In **Phone Numbers** → **Verified**, a number whose verification failed has a **Delete** button. Confirm with **Yes, Delete** and it's gone, no support ticket needed. Verified numbers stay as they are. → [Verifying an Existing Phone Number for Outbound Calls](/ai-voice-agents/phone-numbers/verifying-phone-numbers-in-dapta.md) *** ## AI Notetaker **AI Notetaker calendars and billing.** Your first calendar is free, with 100 hours of recording. Each extra calendar costs 30,000 credits a month, and a banner shows the free hours you have left. Each connected calendar also sets its meeting bot's name and image, and who gets the summary email. → [Calendars, free hours and billing](/ai-notetaker/calendars-and-billing.md) · [Meeting bot settings](/ai-notetaker/meeting-bot-settings.md) **Meetings without a title get one written from what was said.** When a meeting reaches AI Notetaker with no title, such as one you sent the bot to with **Send Bot**, it takes the headline of its recap, so your list reads as topics. Meetings from your calendar keep their calendar title. → [AI Notetaker](/ai-notetaker/ai-notetaker.md) *** ## Flow Studio and the Dapta MCP server **A node can take its credential from a variable.** Next to the node's credential, switch from **Option** to **Variable** and type an expression, such as a secret from your **Vault** or a value the flow receives, so one flow can use a different token on each call. Switching back to **Option** brings the credential list back. → [Get a Record from HubSpot](/flow-studio/nodes/get-a-record-from-hubspot.md) **Connect Claude to Dapta with your Dapta login.** Add `https://mcp.dapta.ai/mcp` as a custom connector in Claude and sign in with your Dapta account, with no API key. It works in the workspaces where you are an owner or admin and asks which one when you have several. Your API keys keep working. → [Connect claude.ai & Claude Desktop](/dapta-mcp/setup/connect-claude-desktop.md) **More from the Dapta MCP server.** Claude can work with flows, voice and text agents, sequences, calls and phone numbers, and build a Dapta database with its tables and rows from scratch. → [Tools reference](/dapta-mcp/tools-reference.md) · [Dapta Databases](/dapta-mcp/skills/dapta-databases.md) *** August 2026 October 2026 All news --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.dapta.ai/news/september-2026.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.