Can AI Review Your Investment Portfolio? What It Can and Can’t Tell You

Online Tech Tips Staff · Edited by Aseem Kishore
13 min read

Online Tech Tips is reader-supported. We may earn a commission when you buy through links on our site. Learn more.

You’ve got a brokerage account, a couple of ETFs, and a few stocks you picked because you liked the company. You also have a nagging feeling you’re lopsided somewhere. Then you remember ChatGPT is sitting right there in another tab. Can you just paste your holdings in and ask? Short answer: an AI portfolio review is useful for some jobs and risky for others.

AI handles the math well. It can work out position weights, flag when your top three holdings dominate, spot overlap between ETFs and stocks, and summarize public SEC filings in plain English. It can’t predict prices, and it doesn’t know your tax situation unless you tell it. Without live data, it can also state stale numbers with total confidence. (And to be clear, nothing here is financial, tax, or investment advice. It’s a guide to what AI tools do and how to use them carefully.)

AI Portfolio Review at a Glance

What You Want to KnowCan AI Handle It?Safest Way to Ask
What percent each holding isYes, reliablyType tickers and weights or share counts
Whether your top 3 positions dominateYes, reliablySame as above
Whether your ETFs overlap with your stocksYes, if it uses published fund holdingsName each fund and stock clearly
A plain-English summary of a 10-Q or 8-KYes, with a source linkAsk for the filing date and SEC link
Upcoming earnings or ex-dividend datesOnly with live, dated dataVerify against your brokerage or the company
Where a stock will go nextNoDon’t ask, and ignore it if offered
Your tax impact from sellingOnly with the details you supplyTalk to a tax professional

What Is an AI Portfolio Review?

An AI portfolio review is when you hand an AI tool a list of your investments and ask it to analyze them. That tool could be a general chatbot like ChatGPT or Claude. It could also be a dedicated app built around holdings data.

What you get back depends heavily on what you ask. Some questions have checkable answers. Others invite the AI to guess. Here’s the split:

  • Diagnostics (safe): Position weights, concentration, overlap, filing summaries
  • Lookups (verify first): Earnings dates, ex-dividend dates, current prices
  • Predictions (skip entirely): “Will this stock go up?” or “What should I buy?”

DIY investors get the most out of it. These are folks who manage their own brokerage accounts and want a quick sanity check on diversification. Diversification just means spreading money around so one bad bet doesn’t sink you. Investor.gov has a plain definition if you want the official version.

What AI Can Reliably Tell You About Your Portfolio

These are the jobs where AI earns its keep. Each one has an answer you can double-check with a calculator or a source document.

Position Weights and Top-3 Concentration

Give an AI a list like “VTI 35%, AAPL 12%, MSFT 10%, plus smaller positions.” It can tell you your top three holdings make up 57% of the portfolio. That’s plain arithmetic. You could do it in a spreadsheet, but AI does it in seconds and explains what it means.

If you give share counts instead of percentages, the AI needs current prices to work out weights. That’s where accuracy depends on its data source (more on that below).

ETF and Stock Overlap

This one surprises people. Say you own Apple shares directly and also hold a broad-market ETF. That ETF almost certainly holds Apple too. So your real Apple exposure is bigger than your Apple line item suggests.

AI can compare your stocks against a fund’s published holdings and point out duplicates—just make sure it cites the fund’s holdings ‘as-of’ date, since ETF disclosures lag live trading and look-through checks can miss derivatives, cash, or recent portfolio shifts. It’s like discovering you’ve been buying the same groceries at two different stores.

SEC Filing Summaries

Public companies file quarterly reports (10-Qs), annual reports (10-Ks), and event reports (8-Ks) with the SEC. They’re long and dry. AI is good at turning them into a few readable paragraphs.

The catch? Even an accurate summary can compress risk factors, footnotes, non-GAAP adjustments, or later amendments. Make sure the summary names the filing type, period, and filing date and links to the original document so you can read the relevant section on SEC EDGAR before making any decision.

Earnings and Ex-Dividend Dates

AI can list upcoming earnings reports or ex-dividend dates for your holdings. (The ex-dividend date is the cutoff for whether you get the next dividend.) This works only when the tool pulls from a live, dated market-data source. A chatbot working from memory can hand you last year’s date with a straight face.

The Three Big Blind Spots

Here’s where things go sideways. Know these three, and you’ll avoid most AI portfolio mistakes.

  • It Can’t Predict Returns: Chatbots generate fluent text based on patterns. A confident tone tells you nothing about accuracy. If an AI tells you where a stock is headed, treat that as a red flag. Nobody’s crystal ball works, and a chatbot’s is just better at sounding sure.
  • It Doesn’t Know Your Full Financial Picture: The AI sees only what you paste. It doesn’t know about your 401(k) or your spouse’s IRA. It doesn’t know your cost basis (what you originally paid), how long you’ve held each position, or your tax bracket. Any “sell this” suggestion ignores all of that unless you supply it.
  • It Can Hallucinate Stale Numbers: A chatbot without live market data can state an old price, market cap, or earnings date as if it’s current. No warning label separates a verified fact from a plausible guess. Always ask the AI where a number came from and what date it reflects. If it can’t say, check it yourself.

One more small gotcha: tickers can be ambiguous across exchanges, countries, and share classes. Add the full company name or exchange when there’s any doubt.

Is It Safe to Upload a Brokerage Screenshot or CSV to ChatGPT or Claude?

It can be, but only after some cleanup. Depending on your broker, a raw CSV export or screenshot can expose account numbers, names, addresses, contact details, tax-lot records, linked-account identifiers, and transaction IDs. None of that helps with a diversification check, so there’s no reason to share it.

Here’s the privacy routine I’d follow every time:

  • Turn Off Model Training First: In ChatGPT, go to Settings > Data controls and turn off Improve the model for everyone. OpenAI’s data controls documentation says this setting is on by default for personal Free, Plus, and Pro workspaces (managed ChatGPT Team/Business, Enterprise, Edu, and API workspaces exclude customer data from training by default). In Claude, look for a privacy or data-control option under Settings > Privacy that lets you opt out of model training. Per Anthropic’s current privacy documentation, opting out stops new chats from being used for future model training, though safety-flagged conversations and conversations you explicitly report may still be used for safety or model-improvement purposes.
  • Scrub the Entire File: Inspect every column and header—redact account numbers, identity and contact fields, tax-lot details, linked-account info, and transaction IDs so only ticker, shares or weight, and asset type remain.
  • Crop Screenshots: If you’re sharing an image, crop out the header with your account details before uploading.
  • Skip the File Entirely: This is the cleanest option. Just type “VTI 35%, AAPL 12%, MSFT 10%” into the chat. No file, nothing to scrub.
OpenAI's public help article on ChatGPT data controls, showing the section describing the "Improve the model for everyone" setting and the Settings > Data controls path

Two nuances are worth knowing. ChatGPT’s Temporary Chat mode keeps a conversation out of your history and out of model training. But OpenAI says those chats may still be retained for up to 30 days for safety purposes (OpenAI consumer data FAQ).

Opting out is prospective rather than retroactive: Anthropic notes that conversations already included in started or completed training runs aren’t removed when you turn off model improvement for future chats, and OpenAI’s retention and training depend on your workspace type and Data Controls at the time you send a prompt—which is why scrubbing sensitive fields before uploading still matters.

Anthropic's public Claude privacy help article explaining how to change the model-improvement setting under Settings > Privacy

How an AI Portfolio Review Works

Under the hood, there are two very different setups.

A general chatbot works from whatever you type plus its training data. Some versions can search the web. Still, you can’t always tell when a number came from a live source or from memory. It’s great for math on numbers you provide and for summarizing a document you paste in.

A grounded tool connects to an actual data source first. That might be your brokerage holdings, a live price feed, or the SEC’s filing database. The AI then answers questions about that data. Figures are less likely to be stale because they don’t come from the model’s memory.

Think of it this way. A general chatbot is a smart friend answering from memory. A grounded tool is that same friend reading off your actual statement.

Using a Tool Built Around Holdings Rather Than a Blank Chat Box

Disclosure: Online Tech Tips founder Aseem Kishore builds Position Tracker, the tool described in this section. Weigh what follows with that in mind.

If you’d rather not paste anything into a general chatbot, Position Tracker’s Portfolio Check calculator is built for that. According to the site, as of this writing, it lets you enter up to 15 tickers (with optional dollar values) in your browser and runs a single-stock and sector concentration check alongside SEC and Congress links, with no account required.

This description reflects the company’s own claims rather than independent testing, and features can change. Confirm the current details on the page before relying on it.

Position Tracker's free Portfolio Check tool showing the no-login ticker input and concentration check

There’s also a public SEC filings page that, as of this writing, shows AI summaries of public filings with links back to the original source documents. That source link is the part I’d look for in any tool, so verify it’s still there when you visit.

Position Tracker's public SEC Filings hub showing AI-analyzed filing highlights and free alert options

The full Position Tracker web app is described on its own site as connecting to your brokerage on a read-only basis, reportedly through providers such as Plaid or SnapTrade. That’s the company’s claim rather than something independently verified here, so confirm the current integration and scope on the site before connecting an account.

A genuinely read-only connection generally means it can see holdings but can’t place trades or move money, though the exact scope depends on what the provider authorizes. According to the site, the full app also provides daily holdings briefings and lets you ask questions about your portfolio. Like everything else here, it’s a diagnostic tool, not one that gives advice or trading signals.

A Quick Checklist for Judging Any AI Portfolio Tool

Before you connect anything, run through this list:

  • Read-Only Access Only: The connection should be read-only. It usually runs through OAuth, a login handoff that never shows the app your password. If it wants trading permissions, walk away.
  • No “Auto-Trading” Promises: Any tool claiming it’ll trade for you or beat the market is selling a fantasy.
  • Transparent Pricing: You should see what it costs, and what the free tier includes, before you sign up.
  • Grounded Data Sources: The tool should show where numbers come from and when. Pure chat guesses don’t count.
  • Source Links on Filings: Summaries should link back to the original SEC document.

Who Should Use an AI Portfolio Review?

  • DIY Investors With a Few Accounts: You want a fast concentration check without building a spreadsheet.
  • ETF Collectors: You own several funds and suspect they hold the same companies.
  • Busy Shareholders: A company you own just filed a 10-Q, and you want the highlights.
  • Privacy-Conscious Curious Types: You want to try it but refuse to link a brokerage login. Ticker weights or a no-signup calculator work fine.

It’s a poor fit if you want someone to tell you what to buy. Same goes for questions about taxes, estate planning, or your full household finances. That’s a job for a qualified professional.

Getting Started

Here’s the high-level path. Each step takes a minute or two.

  • Pick Your Entry Point: A general chatbot with typed ticker data, or a no-login calculator like Portfolio Check.
  • Lock Down Privacy (Chatbot Route): Turn off model training in ChatGPT or Claude before pasting anything.
  • Strip Identifying Info: Delete account numbers and names, or just type tickers and weights.
  • Ask for Diagnostics: Request weights, top-3 concentration, overlap, or a filing summary. Tell the AI not to predict prices or returns.
  • Verify Anything Time-Sensitive: Cross-check dates and figures against your brokerage or the original SEC filing.

Pricing

ChatGPT and Claude both offer free consumer tiers that can typically handle basic arithmetic like position-weight math, though exact features depend on the current plan and version in your region.

As of this writing, Position Tracker’s site lists a 14-day trial that doesn’t require a card up front, a Starter tier at $5/month, and a Pro tier at $10/month (figures in USD). These are current site-reported terms rather than independently verified pricing, and the company can change them, so confirm current prices and trial terms before signing up.

OptionWhat You GetPrice
ChatGPT Free / Claude FreeGeneral chat, manual data-sharing opt-out$0
Position Tracker Portfolio CheckBrowser-based concentration and overlap check, no account$0
Position Tracker read-only snapshotRead-only holdings snapshot via a brokerage link such as Plaid or SnapTrade$0 (as of this writing)
Position Tracker StarterDaily holdings briefings and portfolio Q&A$5/month (as of this writing); 14-day no-card trial
Position Tracker ProExpanded briefings and Q&A features$10/month (as of this writing); 14-day no-card trial

Pricing and trial terms above reflect Position Tracker’s own site at the time of writing and are subject to change. Confirm current details on the site before subscribing.

Alternatives

AlternativeBest ForKey Difference
Manual spreadsheetTotal privacyNo AI at all, so zero hallucination risk, but slower
ChatGPT or Claude with ticker-only inputFree one-off questionsFlexible, but you scrub data and verify dates yourself
Dedicated read-only portfolio toolsOngoing checks on live holdingsGrounded data, at the cost of an account link and possibly a fee

If you go the spreadsheet route, a big monitor or a second display helps. Juggling your brokerage tab and Google Sheets gets a lot less cramped.

Wrapping Up: The Bottom Line

So, can AI review your investment portfolio? For the math, yes, and it’s quick about it. Concentration, overlap, and filing summaries are solid uses. Price predictions and tax calls aren’t, no matter how confidently the chatbot phrases them. My take: start with typed ticker weights or a no-signup calculator. Only connect a brokerage if you want ongoing briefings, and only through a read-only link.

ProsCons
Fast, checkable math on weights and concentrationCan’t predict prices or returns
Catches hidden ETF overlap you’d likely missBlind to your taxes and cost basis unless you supply them
Turns dense SEC filings into readable summariesGeneral chatbots can state stale figures confidently
Free options exist that need no account at allPrivacy cleanup takes effort on every upload
StepActionApplies To
1Turn off model trainingChatGPT, Claude
2Scrub files or type tickers onlyAny chatbot
3Ask for diagnostics, never predictionsAny AI tool
4Verify dates against a live sourceAny AI tool
5Use read-only connections onlyDedicated apps

One last housekeeping tip: privacy settings can change after app updates. Re-check ChatGPT’s consumer privacy and Data controls settings along with your privacy settings in Claude every few months. If you connect a read-only tool and stop using it, revoke its access from your brokerage’s connected-apps page. Again, nothing here is financial, tax, or investment advice.