Skip to content

Update deployment instructions for AI models #749

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
tiagov8 merged 1 commit into from
Jun 11, 2026
+32 −28
Merged

Update deployment instructions for AI models #749

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
60 changes: 32 additions & 28 deletions content/docs/EagleEye/deploy-ai-model.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@ If you plan to use **Non-AI Tags** only, you can skip this guide.

At the moment, EagleEye supports **chat models only**, so make sure you choose a compatible chat model.

**EagleEye supported models include:**
✅ **EagleEye supported models include:**

* `gpt-4.1-mini (recommended)`
* `gpt-4o-mini`
Expand All @@ -32,36 +32,39 @@ At the moment, EagleEye supports **chat models only**, so make sure you choose a
* `gpt-4.1`, `gpt-4o`
* `gpt-3.5-turbo`

**Do not use:**
⚠️ **Do not use:**

* Reasoning models such as `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-5-pro`, `gpt-5-codex`, `gpt-5.4-mini`, `o1`, `o1-mini`, `o3-mini`, `o4-mini`
* Non-OpenAI Foundry catalog models such as Llama, Mistral, Grok, DeepSeek, Cohere, Kimi, `gpt-oss-*`, and Phi
* Non-chat models such as embeddings, image models, audio models, speech models, and legacy completion models

## Step 1 - Deploy an AI chat model in Azure AI Foundry

You have two options. Pick the one that best fits your workflow.
You have 2 options. Pick the one that best fits your workflow.

### Option A - Deploy via Azure AI Foundry portal

1. Go to your EagleEye resource group used during Step 3 installation (e.g., `Northwind.EagleEye.RG`).
2. Find your `openai-shared-xxx` resource and open it.
1. Go to your EagleEye resource group used during Step 3 installation (e.g., `Northwind.EagleEye.RG`)
2. Find your `openai-shared-xxx` resource and open it
![](/EagleEye/Guides/OpenAI-Shared.png)
3. Click **Go to Azure AI Foundry portal**.
3. Click **Go to Azure AI Foundry portal**
![](/EagleEye/Guides/Foundry-Portal-Link.png)
4. In Foundry portal, click **Model catalog**.
4. In Foundry portal, click **Model catalog**
![](/EagleEye/Guides/Model-Catalog.png)
5. Pick a compatible chat model > Use this model
![](/EagleEye/Guides/GPT-4.1-Mini.png)
6. On the deployment screen, set:
* **Deployment name** to exactly `eagleeye`
* **Deployment type** to `Global Standard` if available. If not, choose another compatible deployment type with available quota.
* **Deployment type** to `Global Standard` if available. If not, choose another compatible deployment type with available quota
* **Tokens per Minute Rate Limit** to at least `100`
![](/EagleEye/Guides/Model-Config.png)
7. Click **Deploy** and wait for the deployment to become active.
7. Click **Deploy** and wait for the deployment to become active

> **Important:**
> The deployment name must be exactly `eagleeye`. If you use a different name, EagleEye will fail to find the model at runtime.
<OutlineBox
content={<>
❗️ **Important:** The deployment name must be exactly `eagleeye`. If you use a different name, EagleEye will fail to find the model at runtime.
</>}
/>

### Option B - Deploy via AI agent

Expand All @@ -76,14 +79,17 @@ Create the deployment with the name exactly `eagleeye`.

The agent should:

1. Ask for the resource group you used during Step 3 installation (e.g., `Northwind.EagleEye.RG`).
2. Find the `openai-shared-xxx` Azure OpenAI resource in that resource group.
3. Show which compatible non-reasoning chat models have available quota.
4. Ask which model and capacity you want to deploy.
5. Create the deployment with the exact name `eagleeye`.
1. Ask for the resource group you used during Step 3 installation (e.g., `Northwind.EagleEye.RG`)
2. Find the `openai-shared-xxx` Azure OpenAI resource in that resource group
3. Show which compatible non-reasoning chat models have available quota
4. Ask which model and capacity you want to deploy
5. Create the deployment with the exact name `eagleeye`

> **Tip:**
> Review the commands your agent proposes before approving them. AI agents can still make mistakes.
<OutlineBox
content={<>
💡 **Tip:** Review the commands your agent proposes before approving them. AI agents can still make mistakes.
</>}
/>

## Step 2 - Verify AI email analysis works

Expand All @@ -99,19 +105,15 @@ Use EagleEye's built-in test flow to verify the model deployment without sending

4\. In the **AI decision prompt** field, enter a simple test prompt such as:

```text
Check if the email has been checked by someone by looking for keywords like "Checked by xx", "Checked by xx & xx", or similar variations.
```
> Check if the email has been checked by someone by looking for keywords like "Checked by xx", "Checked by xx & xx", or similar variations.

5\. Paste sample email content such as:

```plain
(Checked by Adam)

Hi team, I just checked the invoice from Acme Corp and everything looks good.
Regards,
Hark
```
> (Checked by Adam)
>
> Hi team, I just checked the invoice from Acme Corp and everything looks good.
> Regards,
> Hark

6\. Click **Test**.

Expand All @@ -127,6 +129,8 @@ Hark
</>}
/>

---

## References

* [Step 3 - Install from Azure Marketplace](./step-3-install-from-marketplace)
Expand Down
Loading