[8.14] Updates Attack Discovery images/gif, model recommendation, beta/tech preview label (backport #5263)

docs/attack-discovery/attack-discovery.asciidoc

@@ -7,36 +7,38 @@
 :frontmatter-tags-content-type: [overview]
 :frontmatter-tags-user-goals: [get-started]
 
-beta::[]
+preview::["This feature is in technical preview. It may change in the future, and you should exercise caution when using it in production environments. Elastic will work to fix any issues, but features in technical preview are not subject to the support SLA of GA features."]
 
 NOTE: This feature is available starting with {elastic-sec} version 8.14.0.
 
-Attack discovery leverages large language models (LLMs) to analyze alerts in your environment and identify threats. Each "discovery" represents a potential attack and describes relationships among multiple alerts to tell you which users and hosts are involved, how alerts correspond to the MITRE ATT&CK matrix, and which threat actor might be responsible. This makes the most of each security analyst's time, helps fight alert fatigue, and can reduce your mean time to respond.
-
-NOTE: Attack discovery currently only analyzes alerts from the past 24 hours.
+Attack discovery leverages large language models (LLMs) to analyze alerts in your environment and identify threats. Each "discovery" represents a potential attack and describes relationships among multiple alerts to tell you which users and hosts are involved, how alerts correspond to the MITRE ATT&CK matrix, and which threat actor might be responsible. This can help make the most of each security analyst's time, fight alert fatigue, and reduce your mean time to respond.
 
 This page describes:
 
-* <<attack-discovery-generate-discoveries, How to generate discoveries>>.
-* <<attack-discovery-what-info, What information each discovery includes>>.
-* <<attack-discovery-workflows, How you can interact with discoveries to enhance {elastic-sec} workflows>>.
+* <<attack-discovery-generate-discoveries, How to generate discoveries>>
+* <<attack-discovery-what-info, What information each discovery includes>>
+* <<attack-discovery-workflows, How you can interact with discoveries to enhance {elastic-sec} workflows>>
 
 [[attack-discovery-generate-discoveries]]
 [discrete]
 == Generate discoveries
 
-To use Attack discovery:
+When you access Attack discovery for the first time, you'll need to select an LLM connector before you can analyze alerts. Attack discovery uses the same LLM connectors as <<security-assistant>>. To get started:
 
 . Click the **Attack discovery** page from {elastic-sec}'s navigation menu.
-. When you open the page for the first time, you'll need to select an LLM connector before you can analyze alerts. Select an existing connector from the dropdown menu, or add a new one.
+. Select an existing connector from the dropdown menu, or add a new one.
 +
-NOTE: Attack discovery uses the same LLM connectors as <<security-assistant, Elastic AI Assistant>>. If you've already configured one, you can use it here without further configuration. In general, models with larger context windows are more effective for Attack discovery.
+.Recommended models
+[sidebar]
+--
+While Attack discovery is compatible with many different models, our testing found increased performance with Claude 3 Sonnet and Claude 3 Opus. In general, models with larger context windows are more effective for Attack discovery.
+--
 +
 image::images/select-model-empty-state.png[]
 +
 . Once you've selected a connector, click **Generate** to start the analysis.
 
-It may take from a few seconds up to several minutes to generate discoveries, depending on the number of alerts and the model you selected.
+It may take from a few seconds up to several minutes to generate discoveries, depending on the number of alerts and the model you selected. Note that Attack discovery only analyzes alerts from the past 24 hours.
 
 IMPORTANT: Attack discovery uses the same data anonymization settings as <<security-assistant, Elastic AI Assistant>>. To configure which alert fields are sent to the LLM and which of those fields are obfuscated, use the Elastic AI Assistant settings. Consider the privacy policies of third-party LLMs before sending them sensitive data.