You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: sdk/ai/azure-ai-agents/CHANGELOG.md
+17Lines changed: 17 additions & 0 deletions
Original file line number
Diff line number
Diff line change
@@ -2,6 +2,22 @@
2
2
3
3
# Release History
4
4
5
+
## 1.1.0b3 (2025-06-30)
6
+
7
+
### Features Added
8
+
9
+
- Added support for Deep Research tool. For more information, see https://aka.ms/agents-deep-research.
10
+
11
+
### Bugs Fixed
12
+
13
+
- Fixed a tracing related bug that caused an error when process was ending if messages or run steps were listed and the resulting list was not iterated completely.
14
+
15
+
### Sample updates
16
+
17
+
- The file search samples were updated to demonstrate retrieving text associated with citations.
18
+
- Added samples for file search citation with streaming.
19
+
- Added samples showing usage of Deep Research tool (sync and async).
20
+
5
21
## 1.1.0b2 (2025-06-09)
6
22
7
23
### Bugs Fixed
@@ -11,6 +27,7 @@
11
27
- Fixed a tracing related bug that caused process_thread_run span to not appear when streaming is used without event handler.
12
28
13
29
### Sample updates
30
+
14
31
- Changed all samples to use `AIProjectClient` which is recommended to specify endpoint and credential.
To use this SDK, start by creating an `AIProjectClient`. For more information on `azure-ai-projects`, refer to its [README](https://github.com/Azure/azure-sdk-for-python/blob/main/sdk/ai/azure-ai-projects/README.md).
86
+
87
+
Here is an example of creating a synchronous `AIProjectClient`:
85
88
86
89
```python
87
90
import os
88
-
from azure.ai.agentsimportAgentsClient
91
+
from azure.ai.projectsimportAIProjectClient
89
92
from azure.identity import DefaultAzureCredential
90
93
91
-
agents_client=AgentsClient(
94
+
project_client=AIProjectClient(
92
95
endpoint=os.environ["PROJECT_ENDPOINT"],
93
96
credential=DefaultAzureCredential(),
94
97
)
95
98
```
96
99
97
-
To construct an asynchronous client, Install the additional package[aiohttp](https://pypi.org/project/aiohttp/):
100
+
To construct an asynchronous client, install the `aiohttp` package:
98
101
99
102
```bash
100
103
pip install aiohttp
101
104
```
102
105
103
-
and update the code above to import `asyncio`, and import `AgentsClient` from the `azure.ai.agents.aio` namespace:
106
+
Then use the code below with `AIProjectClient` and `DefaultAzureCredential` in `aio` packages:
107
+
108
+
```python
109
+
import asyncio
110
+
import os
111
+
from azure.ai.projects.aio import AIProjectClient
112
+
from azure.identity.aio import DefaultAzureCredential
113
+
114
+
asyncdefmain() -> None:
115
+
project_client = AIProjectClient(
116
+
endpoint=os.environ["PROJECT_ENDPOINT"],
117
+
credential=DefaultAzureCredential(),
118
+
)
119
+
120
+
if__name__=="__main__":
121
+
asyncio.run(main())
122
+
```
123
+
124
+
Once you have an `AIProjectClient`, you can obtain an `AgentsClient` like this:
125
+
126
+
**Synchronous Client:**
127
+
```python
128
+
with project_client:
129
+
agents_client = project_client.agents
130
+
```
131
+
132
+
**Asynchronous Client:**
133
+
```python
134
+
asyncwith project_client:
135
+
agents_client = project_client.agents
136
+
```
137
+
138
+
Alternatively, you can instantiate an AgentsClient directly as a standalone approach without using `azure-ai-projects`. However, this is not recommended, as it has limitations and lacks the integrated capabilities provided by using an `AIProjectClient`. Here is is the example:
104
139
140
+
**Synchronous Client:**
105
141
```python
106
142
import os
143
+
from azure.ai.agents import AgentsClient
144
+
from azure.identity import DefaultAzureCredential
145
+
146
+
agents_client = AgentsClient(
147
+
endpoint=os.environ["PROJECT_ENDPOINT"],
148
+
credential=DefaultAzureCredential()
149
+
)
150
+
151
+
with agents_client:
152
+
# your code to consume the client
153
+
pass
154
+
155
+
```
156
+
157
+
**Asynchronous Client:**
158
+
```python
107
159
import asyncio
160
+
import os
108
161
from azure.ai.agents.aio import AgentsClient
109
-
from azure.core.credentialsimportAzureKeyCredential
162
+
from azure.identity.aioimportDefaultAzureCredential
110
163
111
-
agent_client = AgentsClient(
112
-
endpoint=os.environ["PROJECT_ENDPOINT"],
113
-
credential=DefaultAzureCredential(),
114
-
)
164
+
asyncdefmain() -> None:
165
+
agents_client = AgentsClient(
166
+
endpoint=os.environ["PROJECT_ENDPOINT"],
167
+
credential=DefaultAzureCredential()
168
+
)
169
+
asyncwith agents_client:
170
+
# your code to consume the client
171
+
pass
172
+
173
+
if__name__=="__main__":
174
+
asyncio.run(main())
115
175
```
116
176
117
177
## Examples
@@ -311,6 +371,47 @@ with project_client:
311
371
312
372
<!-- END SNIPPET -->
313
373
374
+
### Create Agent with Deep Research
375
+
376
+
To enable your Agent to do a detailed research of a topic, use the `DeepResearchTool` along with a connection to a Bing Grounding resource.
377
+
This scenarios requires you to specify two model deployments. One is the generic chat model that does arbitration, and is
378
+
specified as usual when you call the `create_agent` method. The other is the Deep Research model, which is specified
# Create Agent with the Deep Research tool and process Agent run
395
+
with project_client:
396
+
397
+
with project_client.agents as agents_client:
398
+
399
+
# Create a new agent that has the Deep Research tool attached.
400
+
#NOTE: To add Deep Research to an existing agent, fetch it with `get_agent(agent_id)` and then,
401
+
# update the agent with the Deep Research tool.
402
+
agent = agents_client.create_agent(
403
+
model=os.environ["MODEL_DEPLOYMENT_NAME"],
404
+
name="my-agent",
405
+
instructions="You are a helpful Agent that assists in researching scientific topics.",
406
+
tools=deep_research_tool.definitions,
407
+
)
408
+
```
409
+
410
+
<!-- END SNIPPET -->
411
+
412
+
> **Limitation**: The Deep Research tool is currently recommended **only** in non-streaming scenarios.
413
+
> Using it with streaming can work, but it may occasionally time-out and is therefore not yet recommended.
414
+
314
415
### Create Agent with Azure AI Search
315
416
316
417
Azure AI Search is an enterprise search system for high-performance applications. It integrates with Azure OpenAI Service and Azure Machine Learning, offering advanced search technologies like vector search and full-text search. Ideal for knowledge base insights, information discovery, and automation. Creating an Agent with Azure AI Search requires an existing Azure AI Search Index. For more information and setup guides, see [Azure AI Search Tool Guide](https://learn.microsoft.com/azure/ai-services/agents/how-to/tools/azure-ai-search?tabs=azurecli%2Cpython&pivots=overview-azure-ai-search).
@@ -320,22 +421,24 @@ Here is an example to integrate Azure AI Search:
0 commit comments