AI-assisted portfolio look-through
Strataview looks through every fund you hold to show your true exposure by company, sector and region, so you can make better diversification decisions. AI agents find and read your fund documents for you, while every figure is calculated, not generated.
Shown with sample data. Account names and values are illustrative.
The problem
Two funds with different names can hold many of the same companies. When your investments are spread across several platforms, that overlap is hard to spot, and you can end up heavily concentrated in one company or sector without realising it. Working it out means going through factsheets and fund documents one at a time and piecing the picture together yourself.
How it works
Look-through
Strataview brings together the funds you hold across different platforms and looks through each one to the companies underneath. You can see how much of your money is really in any single company, sector or region, and where two or more funds are buying the same things. A list of fund names becomes a clear picture of what you actually own.
Fund research
Before you change anything, it helps to understand what you already hold. Ask Strataview a question about a fund, such as what its objective is or what it invests in, and get a clear answer drawn from the fund's own documents, with a citation for every point. If the documents don't answer the question, Strataview says so rather than guessing.
Answered from the fund's own KIID, through Strataview's document service in Claude.
Documents
Fund documents go out of date, and tracking down the latest version for every fund is tedious. Strataview's AI agent finds the latest documents for each fund from its ISIN and sets them aside for review, and nothing reaches the research library until it has been checked and approved. Preparing your holdings data is quicker too, with prep time cut by more than 80% in testing.
Trust
Every figure is calculated, not generated.
In investing, an answer you can't check isn't worth much. Every figure in Strataview comes from a calculation engine, not a language model. Every answer from fund research cites its source, and those citations are checked in code. Before any change ships, an automated evaluation scores answers on whether the documents support them, and a change that falls short is blocked.
Roadmap
Strataview is a working MVP, and the next stage is about making it more useful day to day.
Next
Cited answers move into Strataview itself, with citations you can click through to the right page of the source document.
Next
A value-over-time chart that separates new contributions from investment returns, so the figures show how the portfolio has actually performed.
Next
Regular checks spot when a fund publishes an updated document and queue it for review.
Build notes
Early on, I set one rule. The language model can explain what a fund does, but it never supplies a figure. Every number on screen, from exposure percentages to overlap between funds, comes from a calculation engine working on the underlying holdings data.
The reason is simple. A model can write a convincing sentence around a wrong number, and in investing a confident wrong number is worse than no answer. So the model narrates from the fund's own documents, and any figure it needs comes from the engine instead.
The rule has a cost. Some fund documents state figures of their own, such as a minimum exposure the fund commits to, and deciding which of those the model may quote has taken real care. I'd still make the same call. It's far easier to loosen a strict rule safely than to tighten a loose one after something has gone wrong.
About me
I'm Nnamdi Chukukere, a product leader with more than ten years in financial services and consulting, most recently in JPMorgan Chase's Asset & Wealth Management business. Strataview is my independent project. I built it because I wanted a clear view of my own investments that no product gave me, and to stay hands-on with how AI products are built and evaluated.
Connect on LinkedIn