> For the complete documentation index, see [llms.txt](https://docs.aisera.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.aisera.com/product-release-notes/2026/aisera-product-release-notes-october-12-2026.md).

# Aisera Product Release Notes - October 12, 2026

Review what's new and changed, and the fixes and limitations for the Aisera platform in the latest release.

## Agent Assist <a href="#conversational-ai" id="conversational-ai"></a>

### Known Issues

For customers who are using Agent Assist with a ServiceNow data source, when multiple incidents are opened in ServiceNow consecutively, Agent Assist may display the summary/answer generated for a previously viewed incident.

## Campaigns <a href="#conversational-ai" id="conversational-ai"></a>

### What's New <a href="#whats-new.4" id="whats-new.4"></a>

#### Low Priority Campaign Batches

Campaign messages are no longer sent all at once. A campaign run now sends messages in small batches at a steady pace of about **50 users per minute**, on a low-priority worker pool. The pacing does not depend on the channel (Web Chat, Slack, Teams, Webex, and so on), so it protects every channel from rate limits.

**Visible effect:** a large campaign takes longer to finish. It reaches the same users with the same content.

[Low Priority Batch Delivery](https://docs.aisera.com/aisera-platform/what-can-my-bot-do/ai-automation-optimize-flow/campaigns/low-priority-batch-delivery)

## Channels <a href="#conversational-ai" id="conversational-ai"></a>

### What's New <a href="#whats-new.4" id="whats-new.4"></a>

#### Secondary Channel for Voice (IVR)

When you are using a voice channel with your application/bot, you can add a secondary channel, such as email, or messaging to deliver a long or detailed version of the response while your voice channel provides a summary of the fulfillment. This secondary channel serves as a backup or specialized routing path.&#x20;

The secondary channel feature is an asynchronous delivery experience for **Voice** channels when a query produces a **RAG-generated response**. The caller immediately hears a brief, high-level summary over the phone. The complete grounded answer, including detailed instructions and citations, is generated in the background and delivered to a configured secondary channel, such as email or messaging, through a Customization Flow.

**Scope and applicability:** This feature applies only to **IVR channels** and to v2 (and later) applications/bots. It does not trigger for other channels, casual conversations, CDS clarifications, or fallback dialogs.

#### Key capabilities <a href="#key-capabilities" id="key-capabilities"></a>

1. **Optimized voice experience:** Callers hear an approximately 50-word Summary response, reducing text-to-speech playback time and keeping calls natural.
2. **Asynchronous re-query:** The platform performs a background re-query to generate the complete grounded answer according to the bot’s Conversational AI Policy.
3. **Automated delivery:** The full answer, citations, and original user query are sent to the configured secondary channel.
4. **Configurable playback:** Administrators can play the summary with a custom suffix or play only the suffix message.
5. **Auditing and tracking:** Delivery status is recorded through `VoiceSecondaryChannelDeliverySucceeded` and `VoiceSecondaryChannelDeliveryFailed` events.

[Using a Secondary Channel with IVR](https://docs.aisera.com/aisera-platform/channels/interactive-voice-response-ivr/channel-settings-for-ivr#using-a-secondary-channel-with-ivr)

#### Using the Secondary Channel in Workflows

You can use a secondary channel in a workflow if:

* The interaction uses a **Voice** channel.
* The turn produces a **RAG response** through knowledge retrieval.
* A secondary-channel Customization Flow is configured for dispatch.

[Secondary Channel Workflow Requirements](https://docs.aisera.com/aisera-platform/channels/interactive-voice-response-ivr/channel-settings-for-ivr#customization-flow-requirements)

#### Enable Webex Native SDK in Tenant Configuration

You can now set a Feature Flag to exable your Webex Channel to use native SDK calls on the backend. Setting this flag changes how the chatbot **talks to Webex**. It does not change what the user sees in Webex, except for text formatting.

Choose the link below for more information about the behavior of the Webex channel when the flag is enabled.

[Enable Webex Native SDK](https://docs.aisera.com/aisera-platform/tenant-setup/aisera-platform-configuration/tenant-configuration-settings/feature-flags#use-webex-native-sdk)

## Conversational AI <a href="#conversational-ai" id="conversational-ai"></a>

### What's New <a href="#whats-new.4" id="whats-new.4"></a>

#### Set Length for Conversational AI Responses

With this release, the binary **Provide Concise Answers** checkbox in the **Conversational AI Policy** has been replaced by three choices for **Primary RAG response type.**

{% hint style="info" %}
This feature applies to v2 (and later) applications/bots.
{% endhint %}

Choose the response verbosity/length that best fits their application/bot use case:

* **Detailed response:** Full grounded answer with no length limit.
* **Summary response:** Approximately 50 words for quick overviews and voice experiences.
* **Concise:** Approximately 400 words for balanced brevity.

This feature can be used with the Summary option in the Secondary IVR channel feature.

[Set Conversational Response Length](https://docs.aisera.com/aiseragpt/how-to-set-up-an-aiseragpt-bot/view-conversational-ai-policy#agent-response-length-and-verbosity)

### Fixes <a href="#fixes.3" id="fixes.3"></a>

* Corrected issue where Deflection Question was being asked multiple times to the End user in same session.

## Flows <a href="#flows" id="flows"></a>

### What's New <a href="#whats-new.4" id="whats-new.4"></a>

#### Built-in variable picker in the Hyperflow prompt editor

The Hyperflow prompt editor now includes a variable picker for built-in variables, so you can insert them into any prompt section without typing their names. The picker currently offers `${flowUser}` and `${conversationContext.channelType}`, and the variables use the same syntax as in Workflows.

[Hyperflow Editor Variable Picker](https://docs.aisera.com/aisera-platform/what-can-my-bot-do/ai-automation-optimize-flow/hyperflow-studio/structured-hyperflow-prompt-editor/work-with-the-structured-hyperflow-prompt-editor#insert-a-built-in-variable)

#### New MCP Tool node in Workflows

The new Model Context Protocol (MCP) node runs tools from an MCP server configured for your application inside an app-level workflow, so you can use external capabilities without mapping or maintaining the server’s REST API endpoints. Tools you sync and register to the bot appear in the workflow designer without further setup. Hyperflows can call these tools through a workflow that contains an MCP node, which lets upstream and downstream workflow nodes check user permissions and filter large tool output before it reaches the Hyperflow.

[MCP Tool Node](https://docs.aisera.com/aisera-platform/what-can-my-bot-do/ai-automation-optimize-flow/ai-workflow-studio/workflow-reference-guides/workflow-nodes/action-nodes/mcp-tool-node)

### Fixes <a href="#fixes.4" id="fixes.4"></a>

* Fixed an issue where Hyperflows did not save tool responses displayed as cards to the conversation history, so agents could not reference information already shown to users in later turns, which led to duplicate responses and inconsistent agent behavior.

### Limitations

* Workflow export and import do not support workflows that contain an MCP node, and attempting either operation fails.
* The MCP node is not supported in tenant-level workflows, even though it currently appears in the tenant-level workflow builder.
* The Request Details page shows an MCP node in the workflow execution path but does not display its payload or output data.
* The platform does not alert you when a schema change on an external MCP tool breaks an app-level workflow. Workaround: Use the Request Details page to identify the MCP node where execution breaks.
* The Salesforce integration in tenant successfully authenticates via OAuth and functions without issue when called through direct **HTTP Request Nodes**. However, it consistently fails when invoked through the native **Execute REST Call Action Node**.

## Ingestion and Connectors <a href="#ingestion-and-connectors" id="ingestion-and-connectors"></a>

### What's New <a href="#whats-new.5" id="whats-new.5"></a>

#### Remote Executor API key expiry notifications

Credential notifications now cover Remote Executor API keys. The platform sends notifications up to 30 days before a key expires, using your existing credential notification settings, so you can rotate the key before bot execution is interrupted.

[Configuring Credential Expiration Alerts](https://docs.aisera.com/aisera-platform/adding-data-to-your-tenant/integrations-and-data-sources/creating-a-connector/configuring-credential-expiration-alerts)

### What's Changed

Selecting `LearnUsers` in the ServiceNow connector now syncs users only. To ingest user profiles, also select `LearnUserProfiles`. If you relied on `LearnUsers` to ingest user profiles, add `LearnUserProfiles` to your connector configuration.

[ServiceNow Connector](https://docs.aisera.com/aisera-platform/adding-data-to-your-tenant/integrations-and-data-sources/connectors/servicenow-connector)

### Fixes

* Fixed an issue where the ServiceNow connector ingested user profiles twice when both `LearnUsers` and `LearnUserProfiles` were selected.
* Fixed an issue where the ServiceNow connector job failed when both `LearnTickets` and `LearnInteractions` were selected.

## Knowledge Base Article Generation <a href="#knowledge-base-generation" id="knowledge-base-generation"></a>

### What's New <a href="#whats-new.6" id="whats-new.6"></a>

#### New Tenant Configuration to Enable Summarization for Copilot

In the neural-search feature of the Aisera application, v2 versions of RAG and File Copilot indices support summarization.

To facilitate different use cases, the summarization is now toggle-able.

The 2 new tenant configuration options are:

* `ConvAIRAGTenantLevelIndexSummarizationEnabled`
* `ConvAIFileCopilotTenantLevelIndexSummarizationEnabled`

To modify these settings, navigate to **Settings > Configuration > Neural Search in the Aisera Admin UI.**

Both of the options are Boolean. If they are set to `True` (default for both) then summarization is enabled, assuming that the index is on a version that supports summarization (`v2` and onwards for both.

When summarization is enabled, documents are summarized during indexing, and the summaries are added next to the chunks as extra context during embedding generation. Summaries are implicitly used during inference via nearest-neighbor (semantic) search (as the embeddings contain the summary of the document) and explicitly during bm25 (keyword) search.

[Enabling RAG Index Summarization](https://docs.aisera.com/aisera-platform/tenant-setup/aisera-platform-configuration/tenant-configuration-settings/neural-search#convairagtenantlevelindexsummarizationenabled)

[Enabling File Copilot Summarization](https://docs.aisera.com/aisera-platform/tenant-setup/aisera-platform-configuration/tenant-configuration-settings/neural-search#convaifilecopilottenantlevelindexsummarizationenabled)

**Disabling Summarization**

If you disable summarization, any new documents that are indexed won’t be summarized, and the embeddings generated for their chunks won’t contain the summary of the document. Also, the summaries won’t be used during the search.

Any already-indexed documents will still implicitly use the summaries only during nearest-neighbor search, as the embeddings already contain info from the document’s summary.&#x20;

Re-indexing (running indexing via the UI, with incremental updates disabled) is required to re-generate the embeddings without document summaries.

[Disabling Summarization](https://docs.aisera.com/aisera-platform/tenant-setup/aisera-platform-configuration/tenant-configuration-settings/neural-search#disabling-summarization)

#### New Neural Search File System Job for Copilot

Currently, the only way to re-generate RAG index summaries is to run indexing jobs that will re-index the particular document.&#x20;

Instead of re-indexing whole data sources, this new System Job re-indexes only documents without summaries.

The job is named `Neural Search File Copilot - Summary backfill`, and it:

1. Checks whether the tenant is using an index version that supports summaries (`v1` doesn’t, `v2` does).
   1. If you're using a v1.0 application/bot, the job will Exit.
   2. RAG Index Summarization is enabled by default in your Aisera tenant configuration.\
      The job checks to see if you have disabled this option. If so, the job will Exit.
2. Finds the document keys for chunks whose `summary` field is null or non-existent.
3. Re-indexes these documents.

[Neural Search Summary System Job](https://docs.aisera.com/aisera-platform/adding-data-to-your-tenant/post-ingestion-tasks-system-jobs#neural-search-file-system-job-for-copilot)

### Fixes

* Fixed an issue where the Application/Bot user sees the URL of their Aisera tenant application instead of the Knowledge Base Article URL.

### Limitations

* The Knowledge Generation Ticket Uploader may time out when processing a large batch of more than 50,000 tickets. When this happens, the uploader job times out and shows a failed status in the jobs status view.\
  \
  **Workaround:** Split the input into batches of approximately 10,000 tickets and run the knowledge generation job. If a large batch has failed, retry its input in smaller batches.

## Test Suite <a href="#conversational-ai" id="conversational-ai"></a>

### Limitations <a href="#whats-new.4" id="whats-new.4"></a>

If you need to run a test that needs to impersonate a real user, implement the test suite either using AI Lens or if real SSO is required, then attach it to a user session in a chatbot.