Case study · AI Financial Advisor

Ask your money a question.

For years, every naira, dollar and investment I hold has been recorded in a careful ledger. The AI Advisor turns that record into something I can talk to. It answers real questions with real numbers from my own data, not generic advice.

The purpose

Data you can't question is only half useful.

The Networth app gathers everything: savings accounts, fixed deposits, stocks, business stakes, property, loans and what I owe, across Naira, Canadian dollars and US dollars. Every change is recorded as a transaction, so the history is complete.

A dashboard can only answer the questions it was designed for. Real financial questions are cross-cutting. They join several asset types, currencies and dates at once:

"Which investments mature this quarter, and would the proceeds be better used paying down a liability?"

No single chart answers that. A person going through the records could, slowly. The advisor does it in seconds, and shows its working.

The gathered database

Ten kinds of record, one picture.

How it works

From ledger to conversation.

  1. 01

    A question

    Asked in plain English, like you'd ask an accountant.

  2. 02

    A full snapshot

    Every conversation starts with the current picture: every account, holding, liability and exchange rate, with totals converted to one currency.

  3. 03

    Tools for depth

    When the snapshot isn't enough, the model calls tools: net-worth history, recent transactions, investments about to mature.

  4. 04

    The outside world

    For rates, regulation or market news, it searches 21 trusted Nigerian, Canadian and global finance sources, and nothing else.

  5. 05

    An answer with numbers

    Specific figures, risks and next steps. Short by default; detailed only when asked.

The beauty of it

Intelligence on top of a well-kept record.

  • Grounded, not guessed

    Answers come from my actual records, fetched by tools, not from what a model thinks a typical person might have.

  • Every currency, one view

    Naira savings, Canadian investments and US-dollar stocks are compared fairly, because the stored exchange rates do the converting.

  • Aware of time

    It knows what's maturing, what moved last month and how net worth has trended. A spreadsheet only knows today.

  • Connected to the world, carefully

    News and rates come from a short list of regulators and serious publications, like the CBN, the Bank of Canada and Reuters, not the open web.

  • Concise by design

    Three to six sentences unless I ask for more. Useful advice is short.

  • Private by design

    It lives inside my Networth app, behind my own sign-in. There are no public endpoints and no separate service to secure.

Questions it can answer

The kind of thing I actually ask.

Illustrative questions, and the data each one draws on.

"Which investments mature in the next 60 days, and what should I do with the proceeds?"

Maturing investments · liabilities · savings

"How has my net worth moved over the last six months, and what drove it?"

Net-worth history · recent transactions

"How exposed am I to the Naira compared with the Canadian dollar?"

Every asset class · exchange rates

"What did the latest CBN rate decision mean for my fixed deposits?"

Web search (CBN, BusinessDay, Nairametrics) · investments

Honest limits

  • Only as good as the ledger. If a record is missing or out of date, so is the answer.
  • Exchange rates are as of their last update, not live to the second.
  • An analyst, not a licensed advisor. It helps me think. It doesn't replace professional financial advice.
  • Built for one person's records. It isn't a multi-user product.

Built with

  • Claude Sonnet (Anthropic)
  • Tool calling
  • Server-sent events
  • Django
  • PostgreSQL
  • Tavily search
  • Heroku

The advisor runs inside the main Django site, using the same models and database as the Networth app, so it always sees the same records, and the same sign-in protects both.