Ask questions about your own data

Everything so far has been preparation. Your numbers are in one place, in one base currency, and you know what not to paste into a chat box.

Now the useful part: an AI that can see the real portfolio rather than your description of it.

What changes

In Module 2 the core limitation was that a general assistant is answering blind. Ask “how is my portfolio doing?” and it can only speak in generalities.

An assistant connected to your actual data answers a different question entirely. It knows the holdings, the quantities, the cost basis, the currencies and today’s prices. So “how is my portfolio doing?” becomes answerable.

This is what Astra does inside Agni Folio — it reads your portfolio and answers from it.

Questions worth asking

Composition

What is my largest holding as a percentage of my total?
How much of my net worth is in a single sector?
What is my split between shares, property, cash and crypto?
Which currencies am I exposed to, and by how much?

Concentration is the most common blind spot, because it forms gradually. A holding that grows well becomes an outsized position without any decision being taken.

Performance

Which holdings have gained and lost the most since I bought them?
What is my overall return, and how much of it is currency movement?
Which positions have gone nowhere for years?

That last one matters. Losers get attention because they are painful; dead money sits quietly for a decade.

Cash flow

How much dividend income did I receive over the past year?
Which holdings actually pay me anything?
When do my fixed deposits mature?

Sense checks

Is anything about my allocation unusual for my age and timeline?
What have I not looked at in over a year?
If I needed 10,000 next month, where would it come least painfully from?

The prompt that finds problems

The single most useful thing to ask:

Looking at my actual holdings, what are the three things most worth my attention, and why?

It forces prioritisation across the whole picture rather than commentary on whatever you happened to mention.

What to still verify

A connected assistant removes the guessing about your numbers. It does not remove the limits from Module 2.

Still verify anything about tax or jurisdiction rules. Seeing your portfolio does not teach it this year’s rules where you live.

Still treat forecasts as arithmetic, not prediction. A projection is a calculation on assumptions you supplied.

Still check surprising claims against the underlying screen. If it says a holding is down 40%, look. Occasionally the data is stale or a corporate action was recorded oddly, and the assistant is faithfully reporting bad input.

It is not advice. It is analysis of your data. The decision remains yours.

Changes should require confirmation

If an assistant can also make changes — adding transactions, editing holdings — insist on confirmation before anything is written.

Well-built tools separate reading from writing. Reads happen freely; writes stop and ask. Agni Folio works this way: anything that would modify your portfolio appears as a confirmation card you approve or reject, and every action is logged.

This matters because language models occasionally misinterpret instructions. A misread question that only produces a wrong answer is an annoyance. One that silently edits your records is a real problem.

Practice step

Ask three questions about your own portfolio:

What is my largest single holding as a share of my total?
Which of my holdings has performed worst since I bought it?
What am I most exposed to that I have not thought about?

Most people are surprised by at least one answer. If nothing surprises you, your picture may still be incomplete — go back to Module 3 and check what is missing.

Next: Goals and financial independence — turning the picture into a plan.

Astra is built into Agni Folio and included free.