Skip to content
Strahl API

Tracing a financial claim across sources

A financial research assistant reviews four Microsoft earnings calls and writes a memo about how the AI investment story changed from FY23 Q4 through FY24 Q4. The memo combines growth figures, capital-spending measures, and management’s capacity commentary. Which source passages influenced that synthesis?

This authored example uses historical investor materials and a memo written as of July 30, 2024. The transcript contains prepared remarks from the first two calls and the full FY24 Q3 and Q4 calls, including Q&A.

Select the synthesis

Section titled “Select the synthesis”

Start with this passage in the final memo:

AI moved from a small contributor to Azure growth to a material growth driver whose near-term expansion was constrained by available infrastructure

The target is in assistant message 11:

{
  "ref": { "message_index": 11, "field": "content" },
  "span": { "start": 91, "end": 238 }
}

This selection investigates a statement that combines material from different quarters. Inspect the returned excerpts to see which parts of the source packet influenced the summary.

Run the example

Section titled “Run the example”

Download the transcript to inspect the conversation, or the prepared request to run the selection above. Save the request as microsoft-ai-investment-request.json.

Terminal window
curl --fail-with-body --compressed https://strahl.ai/api/v1/attributions \
  -H "Authorization: Bearer $STRAHL_API_KEY" \
  -H 'Content-Type: application/json' \
  --data-binary @microsoft-ai-investment-request.json

Send the same request to /api/v1/attributions/scores to inspect individual source spans. The API reference describes both response shapes.

Response

Section titled “Response”

We show only the top five ranked excerpts here. Highlighted text marks the specific influential phrases within each excerpt.

The original user request ranks highest because it sets the AI growth and capacity themes summarized in the selected sentence. This is consistent with the assistant following the assignment. The earnings-call excerpts also highlight Microsoft’s shift from talking about AI to applying it at scale.

In the final earnings call (message 10), an analyst describes Azure demand as “once again greater than available capacity.” That passage connects directly to the assistant’s summary: “near-term expansion was constrained by available infrastructure.”

Stronger color means greater influence within this response.

#1 · Message 1 · contentRelative score: 1.000

Prepare a research memo as of July 30, 2024 on how Microsoft's AI investment story evolved from FY23 Q4 through FY24 Q4. Use the FY23 Q4 and FY24 Q2 prepared remarks and the full FY24 Q3 and Q4 calls. Explain how AI's contribution to Azure growth, capital spending, and capacity constraints changed. Distinguish cash purchases of property and equipment from capital expenditures including finance leases. Assess whether management described a demand problem or a capacity problem, without treating all Azure growth as AI or management's outlook as an accomplished result. Keep the memo to about one page and identify the relevant calls.

#2 · Message 6 · contentRelative score: 0.371

It was a record quarter, driven by the continued strength of the Microsoft Cloud, which surpassed $33 billion in revenue, up 24%. We’ve moved from talking about AI to applying AI at scale.

#3 · Message 4 · contentRelative score: 0.298

SECTION: FY23 Q4 prepared remarks BRETT IVERSEN:Good afternoon and thank you for joining us today. On the call with me are Satya Nadella, chairman and chief executive officer, Amy Hood, chief financial officer, Alice Jolla, chief accounting officer, and Keith Dolliver, deputy general counsel.

#4 · Message 4 · contentRelative score: 0.270

Now, I’ll highlight examples of our progress, starting with infrastructure.

#5 · Message 10 · contentRelative score: 0.240

Amy, with Azure demand once again greater than available capacity, I appreciate the capex investments in the build out and acceleration you expect in the back half. But as we think about cloud capacity and AI services, specifically, can you talk about both the near-term and long-term strategy around the AI partnerships that you’re signing with the likes of Oracle and CoreWeave, for example?

Source packet

Section titled “Source packet”
Transcript message Investor material
4 FY23 Q4 prepared remarks
6 FY24 Q2 prepared remarks
8 FY24 Q3 earnings call and Q&A
10 FY24 Q4 earnings call and Q&A