summary: edgeful AI is a deep research tool for analyzing market reports and finding patterns across multiple data sources. overview of what it is, how it works, your monthly token allowance, and links to detailed guides.
new to edgeful AI? the question you ask shapes the answer you get. before you dive in, read best questions to ask edgeful AI — it covers the 3 question types the AI is built for, how to frame a precise prompt, and the "find commonalities" pattern that surfaces predictive setups.
video walkthrough
a 16-minute walkthrough of edgeful AI V1 with André — covers the Analyze Reports flow end-to-end: configuring reports, reading the merged data table, asking the AI good questions, and using the "find commonalities" prompt pattern.
what edgeful AI is
edgeful AI launched February 28, 2026. it's a research assistant with access to edgeful's report library, built for thorough, data-backed analysis that takes time to do manually.
think of it as a research analyst who's read every report on the platform and can pull any data point on demand — whose research is built on historical records. use it when you want to go deep, and the screener and WIP when you need to move fast — and on your data pages, the ask edgeful AI side panel now reads your live on-screen context, factoring what you're looking at into its answer.
edgeful AI reads the main reports on the platform — the supported list keeps growing.
how it works: analyze and build
edgeful AI started out answering questions from the platform's documentation. it now works with the actual report data, and the workspace has two tabs: analyze for research (free-form chat and Analyze Reports) and build, where you describe a custom report or custom indicator in plain English and it builds it. custom reports are included on every plan, and CPI, FOMC, GDP, and NFP release dates are built in.
reading your data: you can load real report tables for your ticker, your session, and your lookback period — and the AI reasons over that data directly. instead of explaining how the IB by Rejection report works, the AI can read your IB data and tell you which days the rejection setup actually played out.
this is what makes patterns surface that would be impractical to find by hand — the AI can scan months of data across multiple reports and pull out the commonalities much faster than you could on the report pages.
two ways to use it
you can work with edgeful AI in 2 ways: free-form chat or the structured Analyze Reports feature.
free-form chat is open-ended. ask any question about any report and the AI pulls data on demand. great for exploratory research and one-off questions.
Analyze Reports is a step-by-step wizard that configures a specific analysis — you choose the reports, ticker, session, and date range, then it returns a structured data table you can read and compare directly. this is also how you combine reports today — Analyze Reports lets you stack up to 3 reports in a single run so you can see cross-report patterns side by side. if you've been looking for a way to combine reports, this is it.
report templates and edgeful AI. you can't send a saved report template to the AI. templates store a configuration so you can reload it on the report page, nothing more. there are two ways to get that same setup in front of the AI: open the report with your template loaded and use the ask edgeful AI side panel, which picks up the report, ticker, session and date range you're already looking at; or re-enter those same settings in Analyze Reports. either way you're handing the AI the configuration, not the template itself.
chat features worth knowing
a few details that make the conversation flow smoother once you start asking questions:
context pills — every report you analyze shows up as a chip above the message box. these tell the AI which data tables are in scope. if a pill is missing, the AI can't see that report. click the × to drop a report, or reopen one from saved runs to re-attach it
follow-up suggestions — most responses end with a suggested next question. reply yes, run that or copy the prompt — multi-step research is built into how the AI responds
TL;DR summaries — long answers come with a TL;DR section at the bottom. skim it first if you're moving fast, then dig into the full breakdown when you have time
"find commonalities" prompts — asking the AI to find commonalities across a set of qualifying days (every double break day, every gap fill day, every IB single break day) returns a predictive checklist of features those days share. it's the most powerful prompt pattern in the analyze flow
live page context — ask from the side panel while you're on a report or a data page like the screener, what's in play, a discovery scan, or the algo template catalog, and the AI automatically factors in your current view — the page you're on and the rows, readings, or results on screen. a context chip in the message box (for example, wip · futures · new york) shows what it's picking up; remove it with its × for a general answer. opening the panel also shows your 3 most recent chats, with a searchable history of the rest
what edgeful AI remembers
edgeful AI keeps a lightweight memory that carries across chats, so you don't have to reintroduce yourself each time. it holds durable preferences — the tickers and sessions you trade, how you like answers formatted — and it's fully user-editable: you can view, change, or clear what it remembers whenever you want.
this is memory of your stated preferences, not access to your account, trades, or P&L. for the full picture, see how edgeful AI remembers you.
uploading your own files
beyond edgeful's own reports, you can attach your own files to a chat and ask the AI about them. click the attach file button — the paperclip icon in the message box, just to the left of analyze reports — and pick one or more files from your computer. drag-and-drop isn't supported yet. attaching files is a chat feature — the build tab you use for custom reports has no attach button, so bring the file into a chat and ask there.
supported formats are .pdf, .docx, .txt, .md, .csv, .xlsx, plus images (.jpg, .jpeg, .png), up to 10 files as context. attachments are scoped to the chat you add them in — the AI can read and reference a file throughout that conversation, but a new chat starts fresh. (that's about a single conversation's context — edgeful AI also keeps a lightweight memory of your preferences across chats; see how edgeful AI remembers you.)
it's how you bring your own data in alongside the reports — upload a CSV or Excel export of your trades and ask where you're leaking, drop in a chart screenshot and ask what it's showing, or attach your trading plan as a PDF or doc and pressure-test it against the data.
what it can and can't do
edgeful AI is built for deep research — not daily prep. the screener and WIP are your daily tools. for the full breakdown of what edgeful AI handles well and where its limits are, see what edgeful AI can and can't answer.
token allowance
your edgeful subscription includes 6 million tokens per month for use with edgeful AI — the same allowance on every plan. tokens are consumed each time you ask a question or run an Analyze Reports flow, and you can check your current usage right in the edgeful AI interface.
how usage is measured. it's not a count of questions — it's the size of what the AI reads and writes. a quick one-line question costs very little. an Analyze Reports run that loads 3 reports across a long lookback, plus a back-and-forth thread on top of it, costs a lot more, because the AI re-reads the loaded data on each follow-up. that's why two people asking the "same number" of questions can burn very different amounts.
when it resets. tokens reset on your subscription renewal date — not the start of the calendar month. so if your renewal is on the 21st, your tokens refresh sometime on the 21st.
how to conserve your tokens
a few habits stretch the allowance a long way:
load only the reports you need in an Analyze Reports run — every extra report and every extra month of lookback adds to what the AI re-reads on each reply
drop context pills you're done with — click the × on a report chip so the AI stops carrying that data on every follow-up
start a fresh chat for a new topic instead of piling unrelated questions onto one long thread — long threads cost more per message because the whole history rides along (that's the conversation's own context — your saved preferences carry across chats separately as memory)
reuse saved runs — opening a saved run to read it again doesn't re-spend tokens the way re-running the analysis does
ask focused questions — a precise prompt gets a precise answer; vague prompts lead to longer back-and-forths
edgeful AI is stuck on "loading"
if a response hangs on loading, don't keep resending the same question — each resend can spend more tokens. instead: give a heavy Analyze Reports run a moment to finish (big multi-report runs take longer), then refresh the page. if it's still stuck, start a new chat and ask again. a flaky connection or a very long thread is the usual cause.
if you're blocked well under your 6M monthly allowance — say the interface shows 1.4M used and it still won't run — start a new chat first. a single long-running conversation can hit its own size ceiling separate from your monthly tokens, and a fresh chat usually clears it. if a new chat still won't run, that's worth flagging — reach out through the chat widget so the team can check your account.
hit your monthly limit? there's no self-service top-up. reach out through the chat widget in edgeful and the support team can review your account and raise your allotment in some cases. full details in how to access edgeful AI.
plan access
edgeful AI is available on all plans — essential, pro, and all access. no upgrade required.
getting started
for a quick question, click ask edgeful AI in the top-right corner of any page to open the chat side panel. for the full experience — including the Analyze Reports flow — open the dedicated edgeful AI screen from the left sidebar (or click open in edgeful AI at the top of the side panel). on the full screen, look at the suggested prompts panel — they're designed to show you the kinds of questions the AI handles best. pick one and send it. from there, it's all follow-up questions.
common questions
what's the difference between the analyze and build tabs?
analyze is for research on edgeful's reports. build makes custom reports and indicators. more in the central FAQ.
how many reports can I combine in one analysis?
up to 3. more than that and the cross-report patterns get noisy.
why is the AI giving me generic answers?
most common cause is a missing context pill. open saved runs and click the report to re-attach it as a pill so the AI can read its data again.
can I refresh the data on a saved run?
no — saved runs are frozen snapshots. re-run the analysis to get today's data. see how to use saved runs.
can edgeful AI tell me what to trade today?
no — edgeful AI is a research tool, not a signal generator. for daily bias, use the screener and WIP. see what edgeful AI can and can't answer.
when do my tokens reset?
on your subscription renewal date — not a fixed time of day. there's no pre-set calendar-month reset.
I'm blocked but I haven't used all 6M tokens — why?
start a new chat. a single long conversation can hit its own size limit separate from your monthly allowance. if a fresh chat still won't run, reach out through the chat widget.
feedback and feature requests
edgeful AI is actively being improved. if there's a report you want added to Analyze Reports, a column missing from a data table, or feedback on the interface, send it through the chat support bubble — André's team reads it and uses it to prioritize what ships next.
want programmatic access to the data?
edgeful AI's Analyze Reports covers in-app combining today. if you're looking to pull report data into your own systems — bots, dashboards, custom analytics — that's what the edgeful API is for. it's live and included on every plan, so you already have access. generate a key in the API dashboard, then see the edgeful API — overview + walkthrough to get started.
the full API reference is live at edgeful.com/docs/welcome — quickstart, authentication, and the full endpoint list. you can also push any docs page straight into ChatGPT, Claude, or Cursor with the Open in menu in the top right corner of every page.
detailed guides