a 16-minute walkthrough of Analyze Reports with André — from picking your first report through stacking multiple reports, reading the merged data table, asking the AI follow-up questions, and using the "find commonalities" pattern to surface predictive setups.

edgeful AI has 2 ways to work with data. you can ask questions in free-form chat, or you can use the Analyze Reports feature — a structured tool that runs a configured analysis on specific reports and returns a clean data table that gets attached to the chat as context.

the key difference: free-form chat is open-ended. Analyze Reports is precise. you tell it exactly which reports, which ticker, which session, and what lookback period — and it returns structured output the AI can then reason over directly.

this is the V1 evolution of edgeful AI. V0 worked off the platform's documentation and could answer expert-level questions about reports and features. V1 actually connects the report data into the chat — so instead of describing how the IB report works, the AI can read the IB data table for your ticker and find patterns inside it.

Analyze Reports lives inside the edgeful AI interface. open edgeful AI from the left sidebar, then click the Analyze Reports button. a step-by-step wizard opens to walk you through the configuration.

the wizard starts by asking you to configure your first report. you'll set 4 things:

you can also select a subreport variant where applicable — for example, choosing between IB Standard and IB by Retracement for an Initial Balance analysis.

most subreport columns are already merged into the standard report. you don't need to load 50 individual reports to get the data — the standard variant pulls in size, weekday, rejection, close, and other column families automatically.

in practice this means for IB you'll usually see only 2 options in the dropdown — IB Standard and IB by Retracement — because everything else (by size, by rejection, by close, by weekday) is already baked into the standard variant's data table. the same merging applies across other report families.

whatever customizations you have set on the report in your edgeful dashboard come into Analyze Reports with you. if you've got a 15-minute IB configured on your dashboard (unusual but valid), the wizard will use 15 minutes. double-check your customizations before clicking Analyze — they directly shape the output.

customizations are locked once the analysis runs. if you want to change a setting after seeing the results, you have to start a new analysis — there's no edit-in-place. you can view the customizations that produced a given run by clicking through to the configuration panel in the data table, but you can't edit them there.

take your time on step 1. the precision of your configuration directly determines the usefulness of the output.

after configuring your first report, you can click Add Report to stack a second report on top of the first — and then a third if needed. each additional report goes through the same configuration: report type, ticker, session, subreport if applicable. the date range stays shared across all of them.

the maximum is 3 reports per analysis. this limit keeps the output focused — when you're looking at 3 reports side by side, the cross-report patterns are clear. stacking more tends to add noise rather than clarity.

you don't have to add more than 1 report. a single-report analysis is completely valid — especially when you want a thorough read on one specific setup.

once your reports are configured, click Analyze. edgeful AI processes the configuration and returns a structured data table — the key numbers from each report laid out clearly so you can read and compare them without digging through individual report pages.

the data table sits open on the right side of the chat. you can keep it open while you ask questions, or close it for more chat space. keeping it open is generally the better habit — being able to see the column names while you type makes it much easier to phrase precise questions.

three columns appear on every report you pick:

past those three, the rest of the columns are specific to whichever reports are in the run. an IB standard analysis will show IB size, IB by rejection, IB by close, weekday breakdown, and more in one table — all of the IB column families merged in.

when you stack 2 or 3 reports, each report's columns are colour-coded so it's easy to tell at a glance which numbers come from which report.

some reports only apply to specific market conditions. if you pull the outside days report, you'll see dashes (—) on every row that wasn't an outside day. same for gap fill rows on flat-open days. that's normal — the dashes just mean "this row doesn't qualify for this report." the days that do qualify have full data.

every report you've analyzed shows up as a chip / pill above your message box. these pills are what tell the AI which data tables are currently in context for the conversation.

if you accidentally drop a report mid-conversation and re-add it, expect the AI's continuity to take a hit — it's safest to re-ask the question after the pill is back in place rather than assume the prior turn carries over.

precision comes from matching the AI's vocabulary to the column names in your table. the closer your wording is to the actual labels, the less the AI has to guess.

a quick example. both of these ask the same thing:

both will produce a usable answer. the second one is more reliable, especially on edge cases.

at the end of most responses the AI suggests a follow-up question. things like "want me to break down those Tuesday double breaks by how they resolve?" or "want me to check the opening candle direction on those 12 days?" — the AI is reading what's still unanswered in the data and pointing at the next logical question.

you can reply with yes, run that analysis or copy the suggested question verbatim. either works. follow-ups are how multi-step research happens — one analysis seeds the next.

when an answer runs long, edgeful AI appends a TL;DR section at the end with the short version. if you're skimming, jump straight to it. the full breakdown is there above it when you want to dig in.

one of the most useful prompt patterns for the analyze flow is asking the AI to find commonalities across a set of days. examples:

the prompt is intentionally open. the AI scans every row that meets the condition, then looks across the rest of the columns (size, weekday, high/low formed first, opening candle direction, etc.) to surface which features show up most often on those days. the output is usually a ranked "predictive checklist" — a short list of features that historically appear when the condition is true.

it's the closest thing edgeful AI offers to letting you act like a quant without being one. you give it the outcome you care about, and it finds the conditions that lead to it.

once you've analyzed a set of reports, a great opening prompt is simply: what questions can I ask you about these two reports? — the AI knows what's in the table and will suggest the most productive research angles. use that list as a menu for the rest of the conversation.

the real power of stacking reports is finding confluence — where multiple independent reports point to the same thing.

for example, running an IB Standard analysis and a gap analysis together for NQ in the NY session gives you 2 separate data points on the same ticker's directional tendency. if both reports are showing a bullish lean, that's a stronger read than either one alone.

a few combinations that work well together:

the point isn't to find 3 reports that all say the same thing. it's to have 3 independent reads, then assess where they agree and where they diverge — that's the actual edge.

the Analyze Reports feature covers the main reports on the platform — the picker keeps expanding, so here's the current list:

if the report you want isn't in the supported list, you can still ask about it in free-form chat — the AI has access to edgeful's full report library through the chat interface. and if there's a report (or a specific column) you'd like added to Analyze Reports, let support know — the feature is actively being expanded based on member requests.

you don't pick "by rejection" as a separate report — the by-rejection breakdown is already merged into the standard ORB data table, along with the other subreport angles (by size, by weekday, by close). load ORB, then ask the AI for it directly — something like break down the ORB by rejection: when the high prints first, how often does the low break first? the same approach works for IB by rejection.

most queries return in seconds, but a few patterns are worth knowing:

have feedback on the interface or the answers? use the chat support bubble — André's team reads it and uses it to prioritize improvements.