Skip to content

last-minute-corrections #589

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
pballai merged 1 commit into from
May 13, 2026
Merged

last-minute-corrections #589

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
25 changes: 9 additions & 16 deletions ...rs_guide_to_sigma_ai_apps/partners_snowflake_builders_guide_to_sigma_ai_apps.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -330,7 +330,6 @@ Under `Configure`, click the `+` next to `Data elements`.

Connecting `PIPELINE_AGENT` here makes it available to your Sigma agent as a callable tool. Without this, all analysis stays inside Sigma — there's no path to Cortex's transcript search, semantic analytics, or scoring rubric.

click the `+` next to `Data elements`.
1. Click the `+` next to `Warehouse agents`.
2. Choose `PIPELINE_AGENT` from `HOL_DB.AGENTS`:

Expand Down Expand Up @@ -409,9 +408,7 @@ Required inputs:

<img src="assets/bga_10.png" width="800"/>

4. Now let's add and configure an action to send an escalation email.

Now click `+` next to `Step` and configure the action:
4. Click `+` next to `Steps` and configure the action:

| Field | Value |
|---|---|
Expand Down Expand Up @@ -465,9 +462,7 @@ Required inputs:
- recommended_focus: 1-2 concrete areas for improvement. Be specific.
```

8. Click `+` next to `Step` and configure the `Send Coaching Email` action.

Click `Add step` (`+`) and configure:
8. Click `+` next to `Steps` and configure:

| Field | Value |
|---|---|
Expand Down Expand Up @@ -549,7 +544,7 @@ This tool writes a call qualification record directly to the `Call Qualification
15. Paste the following as the tool `Instructions`:

```copy-code
write call to escalation input table when user requests to write back
Write call to qualification input table when user requests to write back
```

16. Click `+` next to `Steps` and configure:
Expand Down Expand Up @@ -621,7 +616,7 @@ This tool writes an escalation record to the `Escalation Log` input table.
23. Paste the following as the tool `Instructions`:

```copy-code
Inserts row into escalation log table
Insert row into escalation log table
```

24. Click `+` next to `Steps` and configure:
Expand Down Expand Up @@ -680,7 +675,7 @@ Cortex will respond and we can proceed:
## Phase 4: Test the Agent
Duration: 10

With the agent built and all three tools configured, this phase exercises every layer of the architecture in sequence — from delegating analysis to Cortex, to writing data back to your warehouse, to sending notifications, to controlling the dashboard from a single sentence.
With the agent built and all six tools configured, this phase exercises every layer of the architecture in sequence — from delegating analysis to Cortex, to writing data back to your warehouse, to sending notifications, to controlling the dashboard from a single sentence.

Before testing, make sure to `Publish` the workbook and `Go to published mode`.

Expand Down Expand Up @@ -717,7 +712,7 @@ Write that analysis to the Call Qualification table.
- A structured row appears.
- This is Sigma acting on Cortex's reasoning.

We can see the new row of data Cortex wrote on the `Data` page in the `Call Qualifications` table:
We can see the new row of data Cortex wrote on the `Data` page in the `Call Qualification` table:

<img src="assets/bga_24.png" width="800"/>

Expand All @@ -726,7 +721,7 @@ We can see the new row of data Cortex wrote on the `Data` page in the `Call Qual
</aside>

### Test 3: Compare Reps and Log Coaching
Ask Cortex a followup question on the `Build` page:
Enter the following in the Chat element:

```copy-code
Compare Alex Rivera's calls to Sarah Chen's. Who's more effective at identifying decision-makers and building urgency? Log a coaching observation for whichever rep needs more development.
Expand Down Expand Up @@ -756,8 +751,6 @@ That coaching pattern is significant. Send a coaching alert to the rep.

Check your inbox for the coaching email addressed to the rep with the observation details.

<img src="assets/bga_25.png" width="800"/>

### Test 5: Spotlight an Account

```copy-code
Expand Down Expand Up @@ -828,7 +821,7 @@ Place the workbook in `Edit` mode.
|---|---|
| Total Calls | Count of `Call ID` |
| Avg Trust Score | Average of `Trust_Quadrant_Score` |
| At-Risk Deals | Formula: `CountIf([Trust Quadrant Score] < 40)` |
| At-Risk Deals | Formula: `CountIf([Trust_Quadrant_Score] < 40)` |
| Accounts | Count Distinct of `Customer Name` |

<aside class="positive">
Expand Down Expand Up @@ -882,7 +875,7 @@ The division of responsibility between the two layers looks like this:

<img src="assets/bga_32.png" width="800"/>

The sales pipeline context is the scaffold, not the constraint. The three action tools you built — email escalation, coaching alerts, and account spotlight — are reusable patterns. The write-then-notify sequence (agent writes data, tool sends email) applies anywhere decisions need to become actions: support ticket escalation, hiring recommendations, compliance flagging. What changes is the domain and the instructions; what stays the same is the architecture.
The sales pipeline context is the scaffold, not the constraint. The six tools you built — three writeback tools, two email notifications, and one dashboard control — are reusable patterns. The write-then-notify sequence (agent writes data, tool sends email) applies anywhere decisions need to become actions: support ticket escalation, hiring recommendations, compliance flagging. What changes is the domain and the instructions; what stays the same is the architecture.

That pattern holds across any domain where unstructured data needs to become a governed decision:

Expand Down
Loading