> 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/october-2026.md).
# October 2026
What changed in Dapta in October 2026: desktop alerts in Inbox, AI Notetaker assistants set up in a minute and failure alerts, correcting the AI's result on a voice call, quick actions and in-place ed
What reached Dapta in October 2026, newest first. Each change links to the page where it is documented in full. The month is still running, so new changes are added here as they ship.
***
## Desktop alerts in Inbox
Inbox can now alert you on your desktop, so you don't miss a customer while you work in another tab.
* **Two alerts.** You get one when a chat is assigned to you, and when a customer replies on WhatsApp in a chat assigned to you while a person handles it.
* **One click to turn them on.** The first time you open Inbox, a banner asks; click **Turn on alerts** and allow notifications in your browser. The bell at the top shows whether alerts are on, and turns them off again.
* **Click an alert to open the chat.** If Dapta is already open, that tab switches to the chat.
* **A clearer header.** The top of the list now has your **Online** / **Offline** switch, the alerts bell, search, **New conversation**, the filter button and the views, in three rows.
→ [Desktop alerts](/ai-text-agents/inbox/desktop-alerts.md) · [Operate conversations in Inbox](/ai-text-agents/inbox.md)
***
## AI Notetaker: automate your meetings in a minute
Sending what was said in your meetings to the tools your team uses now takes a minute, and you hear about it when a meeting couldn't be recorded.
* **Start from the Panel.** Under **Automate your meetings**, pick Slack, HubSpot or Google Sheets and say in your own words what to pull out of each meeting. Dapta turns it into fields you can edit and opens the assistant ready to save.
* **Try it on your last meeting.** A new assistant can run on your latest meeting right away, so you see what your team will get.
* **Send a meeting in one click.** **Send this to** on a meeting sends it to Slack, HubSpot or Google Sheets, and offers to do it after every meeting.
* **Know when a meeting fails.** Turn on **Notify me if the meeting fails** and the assistant tells you when the bot wasn't let in, the meeting wasn't recorded or nobody spoke.
* **A welcome email** after you connect a calendar confirms which meetings the bot will join.
→ [Assistants](/ai-notetaker/assistants.md) · [Meetings](/ai-notetaker/meetings.md) · [Calendars, free hours and billing](/ai-notetaker/calendars-and-billing.md)
***
## Correct the AI's result on a voice call
After every call, the AI decides whether it met your agent's goal. When it gets one wrong, you can now fix it yourself.
* **One click under the recording.** The result reads **Goal met**, **Goal not met** or **No verdict**. Click it, pick the other value and say why in a few words.
* **Everyone sees who changed it.** A corrected result shows a pencil and **Corrected by**, with the name and the date. Hover it to see what the AI said. In **Call History** it reads **· corrected**.
* **Your numbers follow.** **Call History**, its **Successful** filter and the success numbers on the **Voice Agents** page count the corrected result.
* **Nothing is lost.** Every change keeps its author, date and reason in the history. To undo one, pick the other value again.
Anyone who can edit voice agents can correct a result; everyone else sees the result and its history.
→ [Correcting a call result](/ai-voice-agents/analyze-your-calls/correcting-a-call-result.md)
***
## The CRM: add contacts and act on them in one click
The **CRM** is now where your team works its contacts, not only where it finds them.
* **Add a contact by hand.** **New Contact** in the toolbar takes a name and a phone number.
* **One click from an open contact.** Buttons under the contact's name call it, send it a WhatsApp template, add it to a voice or text sequence, or open its notes.
* **Edit in place.** **⋮** → **Edit** turns the name, the phone number and every custom column into fields you change and save right there.
* **Delete several at once.** Tick contacts in the table, or the whole page, and delete them in one go.
→ [CRM: your contacts](/crm.md)
***
## Dapta Auto, the model Dapta keeps up to date for you
**Dapta Auto** is a new choice in the model list of Text Agents and Voice Agents. Dapta picks the model behind it for fast, efficient answers and moves it to a better one after testing it, so your agents get better without you comparing models.
* **New Text Agents start on Dapta Auto**, whether you create them or Dapti does. It is first in the list, marked **Recommended**.
* **40 credits per answer** on Text Agents, and it reads the images and voice notes your customers send.
* **The agents you already have keep their model.** Nothing switches by itself: you move an agent when you decide to.
* **On Voice Agents too.** Choose it under **Settings** → **Voice Settings** → **LLM Model** and save. The per-minute rate of your calls stays the same.
* **GPT-4o retires on 9 November 2026.** If one of your agents uses it, move it to Dapta Auto before that date: **Settings** → **Model** → **Dapta Auto**, then save and publish.
→ [Dapta Auto for Text Agents](/ai-text-agents/setting-up-your-text-agent/dapta-auto.md) · [Dapta Auto for Voice Agents](/ai-voice-agents/how-to-set-up-your-ai-voice-agent/dapta-auto.md)
***
## Your agents book appointments with Dapta Calendars
Voice Agents and Text Agents now book meetings during the conversation, straight into **Dapta Calendars**, Dapta's own scheduling product. The customer never leaves the call or the chat, and never gets a booking link.
* **One action, nothing to copy.** Add **Dapta Calendars Scheduling** to a Voice Agent or **Schedule appointments (Dapta Calendars)** to a Text Agent and pick the meeting. No API key, no event type ID, no Cal.com account.
* **Only times that are really free.** The agent checks your availability before naming a time and books exactly the slot the customer confirms. If someone takes it first, it offers fresh times.
* **Your rule for when to book.** Additional instructions tell the agent when it may offer times, for example only to a qualified lead.
* **The right time zone.** Each action has a default time zone, and the agent switches to the customer's own zone when they say where they are.
* **A confirmation for both sides.** Every booking sends a calendar invite to your customer and to you, and shows up in your Dapta Calendars account.
Text Agents have had it since the end of September; Voice Agents since 8 October. One Calendars account serves every agent in your workspace, with the same meetings and the same availability.
→ [Voice Agents: Schedule Appointments with Dapta Calendars](/ai-voice-agents/agent-actions/schedule-appointments-with-dapta-calendars.md) · [Text Agents: Schedule Appointments with Dapta Calendars](/ai-text-agents/setting-up-your-text-agent/schedule-appointments-with-dapta-calendars.md)
***
## Post-call analysis sends results where you work
The **Call Analysis** tab of every voice agent now ends in **After-call destinations**: as soon as a call ends, its results go where your team already works.
* **Google Sheets.** A new row per call, with the call data you match to each column.
* **HubSpot.** Find the contact by email or another field (or create it), update its fields or its linked deal, and add a note with the summary, the data you extracted and the transcript.
* **Dapta flow.** Run any flow with the call's data as its input.
* **Test with last call.** Run your destinations on the agent's latest call, or on sample data, before a real call comes in.
Up to 25 destinations per agent, each with its own switch to pause it. They send the variables you set up in the same tab, so the data you extract from each call now lands in your systems with no extra step.
→ [Post-call analysis](/ai-voice-agents/analyze-your-calls/post-call-analysis.md) · [Destinations](/ai-voice-agents/analyze-your-calls/post-call-analysis-destinations.md)
***
## WhatsApp through Meta for every Text Agent
Every workspace can now connect WhatsApp to its Text Agents directly through Meta, either with a new number or with the number you already use in the WhatsApp Business app.
* **One place for your numbers.** **Text Agents → Channels → WhatsApp → View Connections** lists each number, the agent that answers it, whether the agent can reply and send business-initiated messages, and the one fix a number needs when something is wrong.
* **A missing payment method is flagged in Dapta.** Since 1 October Meta charges for replies on numbers connected through Meta. If a WhatsApp Business account has no payment method, Dapta stops the agent's replies on its numbers and shows a banner with **Add payment method**. Replies resume on their own once Meta has it.
* **A new template editor.** The form sits beside a live WhatsApp preview. Type `/` to insert a custom field, order **Reply**, **Link** and **Phone** buttons, build carousel cards, and **Submit for approval** lists anything still missing. Template pickers only offer the templates the chosen line can send.
→ [Connect your Agent with WhatsApp](/ai-text-agents/channels/connect-your-agent-with-whatsapp.md) · [Add a payment method](/ai-text-agents/channels/add-whatsapp-payment-method.md) · [Message templates](/ai-text-agents/channels/message-templates.md)
***
## Transfer headers, call context for WhatsApp, AI Notetaker assistants
* **SIP headers filled in at the moment of transfer.** Any `{{name-transfer}}` variable in a transfer's custom SIP headers is written by the agent when it transfers, following an optional line in its prompt. **If your agent uses `{{call_summary_transfer}}`, rename it to `{{call_summary-transfer}}`** to keep sending the call summary.
* **Call context for the WhatsApp agent.** The WhatsApp fallback of a voice sequence and the **Send WhatsApp Template** action pass conversation variables to the Text Agent that answers on WhatsApp, so it knows the context when the customer replies.
* **AI Notetaker assistants.** An assistant can run on your own meetings or on every meeting in the workspace, only on meetings whose title contains your keywords, and send its results to a webhook or a Dapta flow.
→ [Transfer Call](/ai-voice-agents/agent-actions/transfer-call.md) · [WhatsApp fallback](/ai-voice-agents/how-to-create-your-first-campaign-in-dapta/whatsapp-fallback.md) · [Send WhatsApp Template](/ai-voice-agents/agent-actions/send-whatsapp-template.md) · [AI Notetaker assistants](/ai-notetaker/assistants.md)
***
## 1 October: Meta changes how it charges for messages
This is Meta's pricing change and Meta's date, not a Dapta release.
* **Service messages become billable.** Those are the free-form replies your agent sends inside the 24-hour window. Meta charges per delivered message, at the utility rate of the recipient's country.
* **Every business phone number gets 1,000 free service messages a month.** Charging starts at the 1,001st. The allowance resets monthly, does not roll over, and the month begins at midnight in your WhatsApp Business Account's time zone.
* **Utility templates sent inside the 24-hour window become billable.** They had been free since July 2025.
* **Still free:** messages your customers send you, failed messages, and the 72-hour window opened by Click-to-WhatsApp ads.
**If your line was connected the previous way, you are covered.** Dapta is absorbing this increase on legacy lines. There is no surcharge, no price change for now, and no retirement date.
→ [WhatsApp billing: legacy and direct lines](/ai-text-agents/channels/whatsapp-billing-and-meta-charges.md)
***
September 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/october-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.