How a FolioLab scan works, step by step

October 11, 2026 · 6 min read

You type a ticker and tap Scan. A few minutes later you have a report with a 1 to 10 research score. This post walks through everything that happens in between: who reads what, who argues with whom, how the score is set, and where the limits are.

The short version

Reading the dataThree AI analysts each study one side of the stock.
Debating bull vs bearTwo AI researchers argue the case. A research manager rules.
Stress-testing the conclusionA strategist makes a plan. Three risk views go after it.
Writing the reportA head of research makes the final call, in plain English.

Every one of those roles is played by AI. No human analyst is involved at any stage.

Why a desk, and not one question to one AI

Ask a single AI model whether a stock looks good and you get one voice, with all of its blind spots. Investment research teams do not work that way. Analysts each cover their own ground, ideas get argued before anyone commits, a risk function pushes back, and one person owns the final call.

FolioLab runs that structure with AI. Each role has its own instructions and its own inputs, and every role builds on what the earlier ones wrote. A scan is dozens of AI calls made one after another, which is why it takes a few minutes rather than a few seconds.

1. Reading the data: three analysts, three lanes

The scan starts with three AI analysts. Each works in its own lane, pulls its own data, and writes a report that ends with its own lean: bullish, bearish, or somewhere in between.

The technical analyst

Reads five years of daily price and volume, and picks up to eight indicators from a standard set: the 50-day and 200-day moving averages, a 10-day exponential average, MACD, RSI, Bollinger Bands, average true range and a volume-weighted average. Its job is the shape of the chart: the trend, the momentum, where support and resistance sit, and how volatile the stock has been.

The fundamental analyst

Reads the business. Valuation (price to earnings, trailing and forward, the PEG ratio, price to book), margins, returns on equity and assets, debt and cash flow, plus the quarterly income statement, balance sheet and cash flow statement. Its question is whether the business looks strong or stretched at today's price.

The sentiment analyst

Reads the mood around the stock: the past week's company headlines and broader economic news, how the stock is being discussed on social platforms, and public economic data from the Federal Reserve Bank of St. Louis, such as interest rates, Treasury yields, inflation, jobs and growth. Behind the scenes this lane is two passes, one for the news and the economy and one for the social chatter. The app shows them together as one Sentiment signal.

These three reports are the evidence everything else builds on. In your report they become the Signals section, each with a confidence level that reflects how firm and specific that analyst's report is: a decisive report backed by numbers reads higher, a hedged one lower.

2. The bull and bear debate

Next, two AI researchers take the analysts' reports and argue. The bull researcher builds the strongest case for the stock; the bear researcher builds the strongest case against it. They go back and forth over several rounds, and each turn has to answer the other side's last argument, not just repeat its own.

When the debate ends, an AI research manager reads the whole exchange and rules on it, anywhere from bearish to bullish, with its reasons. This is the first moment the desk commits to a view, and the view comes out of the argument rather than from any single analyst.

The Debate section of your report shows the strongest version of each side, so you can weigh the arguments yourself.

3. The pressure test

A ruling is not the end. An AI strategist turns it into a concrete plan, and three AI risk views go after that plan: an aggressive view that asks whether it is too timid, a conservative view that asks what could go wrong, and a neutral view that weighs the two. They argue over several rounds, against the analysts' reports and against each other.

The point of this stage is to catch a conclusion that sounded convincing in the debate but does not hold up once someone looks hard at the downside.

4. The final call and the score

Last, an AI head of research reads the research manager's ruling, the strategist's plan and the whole risk debate, and makes the final call. That call sets the band of your score:

Bearish 1 to 2
Cautious 3 to 4
Neutral 5 to 6
Constructive 7 to 8
Bullish 9 to 10

The number inside the band is not a second opinion from the AI. It follows a fixed rule from the three analysts' leans, where each analyst counts as one vote for or against the stock, or half a vote if its report is mixed:

In a bullish or constructive band, the higher number needs the analysts' votes to add up to at least two in favor. Otherwise the score takes the lower one.
In a bearish or cautious band, the lower number needs at least two votes against. Otherwise the score takes the higher one.
In the neutral band, a net positive lean makes it a 6, otherwise a 5.

So a 10 needs a bullish final call and broad agreement underneath it, and a single enthusiastic analyst cannot push a score to the top on its own. The score describes where the desk's analysis landed, not an action to take.

The models behind the desk are Google's Gemini: a faster model does the reading and the arguing, and a more thorough reasoning model makes the two decisions that matter most, the research manager's ruling and the final call.

5. Writing the report

The desk's working notes are long and full of jargon. A final writing pass turns them into the report you read: a short summary, the three signals, the strongest bull and bear arguments, the investment thesis, and an outlook for the next 1 to 6 months and for 1 to 2 years and beyond.

Each piece is drafted and then checked against the desk's notes, so the plain-English version says what the desk said, in the same direction. Buy and sell language is rewritten as a research read, because FolioLab does research, not recommendations. The score itself is never rewritten: it comes straight from the desk's own output.

How long it takes, and what you see meanwhile

A scan typically takes a few minutes, up to about 10. You do not have to watch: leave the app, and a notification arrives with the score when the report is ready. While it runs, the app and your Lock Screen show the four stages. Those are an estimate based on elapsed time, not a live feed from the desk.

A scan uses one credit, charged only when the report is delivered, so a scan that fails costs nothing. Your first two scans are free, and you do not need an account.

What a scan is, and what it is not

Written entirely by AI. Every report is researched and written entirely by AI (large language models). No human analyst is involved, and every report says so at the bottom.
Not identical every time. These models are probabilistic, so two scans of the same stock can come out differently, and the market moves between scans too. That variation is normal for AI systems, not a malfunction.
A snapshot, not a live feed. Each report shows the date its data comes from.
Research, not advice. A scan does not know your goals, your portfolio or your timeline, and it never tells you to buy or sell. Use it as one input alongside your own research. This post, like every report, is research and education, not investment advice.

Want to see the desk at work on a stock you follow? Get FolioLab on the App Store. Your first two scans are free.