Skip to content

Commit a0e85b7

Browse files
committed
task-oriented chat: add ai section and more small fixes
1 parent 0d67d5e commit a0e85b7

File tree

1 file changed

+29
-6
lines changed

1 file changed

+29
-6
lines changed

src/pages/docs/guides/chat/build-task-oriented.mdx

Lines changed: 29 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -233,7 +233,8 @@ Effective moderation ensures your task-oriented chats remain professional, safe,
233233

234234
Ably supports [moderating messages](/docs/chat/moderation) both before and after publication, making it easy to integrate with AI-powered or human moderation systems.
235235

236-
* **After-publish moderation:** Messages appear instantly, then are removed if flagged as inappropriate. Best for most task-oriented scenarios where immediacy matters.
236+
* **After-publish moderation:** Messages appear instantly, then are removed if flagged as inappropriate.
237+
Best for most task-oriented scenarios, where immediacy often matters. It can be used to detect abusive behaviour or inappropriate content after the fact, and can also be used to remove users if necessary.
237238
* **Before-publish moderation:** Messages are held until approved. Use this for high-stakes tasks where every message must be vetted.
238239

239240
### Key moderation considerations for task-oriented chat
@@ -246,7 +247,6 @@ Ably supports [moderating messages](/docs/chat/moderation) both before and after
246247
* **Inappropriate content:** Flagging adult content, violence, or graphic material.
247248
* **Toxicity:** Measuring overall message sentiment and hostility
248249

249-
250250
2. **Technical integration**
251251
* **Latency impact:** AI moderation adds up to 100ms to message delivery.
252252
* **Integration options:** Choose from pre-built integrations or connect your existing moderation systems via webhooks, serverless functions, or queues
@@ -260,10 +260,10 @@ Ably supports [moderating messages](/docs/chat/moderation) both before and after
260260

261261
Ably's flexible moderation system adapts to your task requirements:
262262

263-
* **Per-room policies:** Different moderation rules for different task types or user roles
264-
* **Fallback handling:** Configure what happens when moderation services are unavailable
265-
* **Custom integration:** Connect your existing moderation infrastructure via webhooks, serverless functions, or message queues
266-
* **Role-based permissions:** Give moderators special capabilities to manage conversations
263+
* **Per-room policies:** Different moderation rules for different task types or user roles.
264+
* **Fallback handling:** Configure what happens when moderation services are unavailable.
265+
* **Custom integration:** Connect your existing moderation infrastructure via webhooks, serverless functions, or message queues.
266+
* **Role-based permissions:** Give moderators special capabilities to manage conversations.
267267

268268
<Code>
269269
```javascript
@@ -283,6 +283,29 @@ room.messages.send({text: 'Can you help me with my order?'}).then((message) => {
283283
```
284284
</Code>
285285

286+
## AI Integrations
287+
288+
AI chatbots are revolutionizing the way we interact with digital platforms. Integrating AI functionality into your chat experience can provide automated contextual moderation, personalized interactions, automatically creating training material, and more.
289+
290+
Ably's flexible [integration capabilities](/docs/platform/integrations) make it easy to connect your chat system with AI services. Through Ably's integrations, you can automatically route messages to:
291+
292+
* Serverless functions (AWS Lambda, Azure Functions, etc.),
293+
* Event streaming services including Kafka, SQS and Pulsar,
294+
* Webhooks and custom endpoints.
295+
296+
This enables powerful AI features like:
297+
298+
* Sentiment analysis of chat messages,
299+
* AI-powered responses to common questions and support queries,
300+
* Chat summarization and key point extraction,
301+
* Automatically adding contextual information for your agents to help them be more effective.
302+
303+
For example, you could set up an integration that sends all messages from each chat room to an AI service for sentiment analysis,
304+
which then automatically flags detected negative customer sentiments for review, to ensure they are handled appropriately, which
305+
could increase overall customer satisfaction.
306+
Or route messages to a language model to generate contextual responses to common questions, or to pull relevant information
307+
for your agent to increase their effectiveness in handling support queries.
308+
286309
## Room reactions
287310

288311
Room reactions are a great engagement feature for chats that accompany calls, meetings, collaborative tools, and games. They are a way to quickly express a sentiment to the entire room at a point in time without adding to chat history or being tired to a message.

0 commit comments

Comments
 (0)