Getting a market report from your AI is one of the best use cases you will find. Done well, it can save you hours every week. But there is a catch most people run into: ask the same question on different days and you might get a chart once, a table the next time, and just a bare number after that. A good answer is not the same thing as a consistent report, and if you are handing this to someone else, consistency is the part that actually matters.
Watch the walkthrough
The catch: a good answer still needs to be a trustworthy one
None of this works if the data behind the answer cannot be trusted, and it is worth being honest about what happens when it cannot.
With no data provider behind it, an AI has no access to verified pricing and no concept of which series is even valid for your market. Rather than refusing to answer, it is likely to give you something plausible sounding using whatever it can find, which is a worse outcome than no answer at all, because nothing about it signals that anything is wrong.
With a weaker or competing connection, you may still get an answer, but with thinner coverage, no independent benchmarking behind it, and sometimes no real data at all, just another chatbot relaying a question and handing back whatever that second system was willing to share. Your AI ends up working with less context than it should have.
A report is only worth building once if the numbers inside it are ones you would be comfortable defending.
Walkthrough: build a reusable market report prompt
What you will need
- An AI subscription that includes MCP connections
- Expana IQ Connect
- A connection to your own data (optional, but recommended)
Step 1. Open a chat with Expana IQ Connect enabled.
Step 2. Ask your AI to draft the report prompt for you. Tell it what the finished report needs to include, not how to phrase the request.
Try this
“Write me a prompt I can reuse to build a comprehensive one-pager for a commodity market. It should include a summary of three to five relevant series and their price history, news and commentary from Expana, and present every series and article as a link back to the Expana platform.”
Step 3. Review the draft before you use it. It will typically be more thorough than anything you would have written yourself, covering which tools to use, what order to present things in and how to handle links. Adjust anything you would want done differently. It is your prompt.
Step 4. Run it on a real commodity to generate your first report.
Step 5. Test it in a brand new chat on a different commodity. This confirms the prompt reliably produces a report in the same format and level of detail, rather than only working for the one case it was drafted against.
Step 6. Check the report is defensible. Look for direct links back to source for every series, time based fields rather than just a current price (such as one month, three month and one year change), and a driver or commentary section tied to a real, citable Expana article.
Step 7. Export it in the format you distribute. PDF, Word document or email: this step is not platform specific and depends on the tools you already use.
You will know it is working when
The assistant answers in plain language, names the specific Expana series or article it pulled from, ideally as a link back to the Expana platform, and running the prompt again on a different commodity produces the same structure and level of detail, not something different each time.
Why this is worth doing once, properly
Ask the same simple question fresh every time and you leave the format up to chance. Build the reusable prompt once, and every report that comes out the other side looks the way you decided it should, on whichever commodity you point it at.
Written by Michael Forman