Changelog

​

A dedicated Skills section, faster answers, and expert-interview transcripts

• Skills — A dedicated Skills section where you can browse and search the library, build your own reusable skill from a research workflow and starter prompt, and launch any skill on demand in Chat. The ready-made set also grows with new deal and private-markets playbooks — buyer lists, quality-of-earnings, merger and IPO models, credit screens, and diligence trackers.

• Search across your full chat history — Searching your past conversations now matches the full text of each chat, not just titles, and shows highlighted snippets so you can jump back to the right thread fast.
• M&A activity rolled up to the ultimate parent — View a company’s deal activity rolled up across its subsidiaries to the ultimate parent, with clearer acquirer and target detail.
• Clearer earnings reports — In generated earnings reports, GAAP, adjusted (non-GAAP), and consensus EPS are now separated and labeled on the correct basis, so every figure reads unambiguously.

• Very Quick mode — A new fastest-answer mode in Chat for instant, to-the-point responses, alongside Quick, Agent, and Deep Research.
• One-click prompt Improve — Turn a rough question into a polished, well-scoped prompt with a single click before you send it.

• Expert interview transcripts — Search and read Third Bridge expert-interview transcripts alongside your research, and ask the assistant about them with citations back to the source.

• No duplicate alert emails — Fixed scheduled report and grid automations that could occasionally send the same alert email more than once.
• Reliable delivery of long reports — Longer scheduled reports now finish and arrive by email reliably, and a delivery that fails now shows an accurate failed status instead of appearing successful.
• Cleaner report output — Fixed reports that could show filing or table boilerplate in place of the company name, display a literal chart placeholder instead of the chart, or mislabel totals on earnings bridge charts.
• Forward guidance preserved — Fixed earnings reports that could rewrite full-year or forward guidance to match the latest quarter; guidance now appears exactly as reported.
• Accurate filings for cross-listed companies — Fixed a case where pulling filings for Canadian or other cross-listed tickers could surface a different company’s documents.
• Uploaded documents appear right away — Fixed a delay where a newly uploaded document could be missing from search results for a short time after upload.

​

Smarter Document Search, text-message alerts, and follow-ups that keep your files

• Text-message alerts — Opt in to receive your most important alerts by text — market and price moves, account and security notices, and a heads-up when a scheduled run fails — so you stay informed away from your screen. Reply STOP to unsubscribe or HELP for help at any time.

• More powerful Document Search — Filter any search to a specific time window — this month, year-to-date, the trailing twelve months, or a custom range — across every source, from news and filings to transcripts, press releases, research, and your Data Room. Choose how you search with four modes — Keyword, Semantic, Ask AI, and Deep Dive — and load more results when you want to dig deeper.
• Document types in Grids — Grids now label each source by type — earnings transcripts, SEC and SEDAR filings, press releases, and investor-relations events — and let you choose which document types feed each company.
• Citations that link to the source — Citations in generated reports now open the underlying document directly — the filing, transcript, or press release behind the number — instead of showing an unlinked reference.
• Search and filter your agent runs — In Agent Studio, search across your past runs and filter them by workspace to find earlier work faster.
• Teach AllMind your templates in the background — Teaching AllMind one of your own report templates now runs as a background task with live progress, so you can keep working while it learns.
• Completion notices for manual runs — Running or test-running an automation by hand now drops a completion notice in your inbox when it finishes, just like a scheduled run.

• Follow-ups remember your files — When you ask a follow-up question, AllMind now keeps the files already in the conversation — including ones it just produced, such as a PowerPoint, Word, or Excel file — so you can refine or build on them without attaching anything again.

• Upcoming triggers appear instantly — After you create or edit an event-based automation, its next upcoming trigger now shows up right away instead of briefly reading “no upcoming triggers.”
• More reliable notifications — Clicking a browser notification now reliably opens the right page, and out-of-date notification registrations are cleaned up automatically so your alerts keep arriving.

​

Agent workspaces, a notification center, and scheduled AI research

• Agent workspaces — Run several research agents at once in a dedicated workspace, follow each run live as it works, and pick up your runs and history from any device.
• Notification center — A new in-app inbox tells you the moment a scheduled task, report, grid, or research run finishes — or someone shares work with you — with unread badges, one-tap mark-as-read, and optional browser push alerts.
• Scheduled AI research — Set an AI research task to run on a schedule — a quick answer, a full agent run, or deep research — drawing on the sources you choose and delivered to you as a finished PDF or Word document.

• A refreshed interface — A cleaner, faster design across the company and equities pages, Chat, the navigation sidebar, and news.

• More accurate earnings figures — Earnings reports now cross-check more reported figures — adjusted EPS, EBIT, and marketing and selling costs — against the company’s official filings to catch and correct misstatements, and no longer treat broker notes written before the results as support for post-earnings conclusions.

• Reliable multi-company reports — Reports built from multi-company and sector templates now research the right companies, and a single slow section no longer fails the whole report.
• Earnings-reaction and commodity data restored — Earnings-reaction analysis works again — with per-quarter EPS and revenue surprises alongside the one-day and one-week price moves — and live commodity prices for gold, oil, natural gas, and copper are back.
• Large research answers no longer cut off — Very large responses with big tables or long output now come through complete.
• Automations no longer get stuck — A scheduled task that could previously hang in a running state and stop firing now recovers on its own.

​

Voice in Chat, sharper research, and ETF coverage

• Voice dictation in Chat — Speak your question instead of typing it: tap the microphone in Chat and your words are transcribed into the message box in real time, ready to refine and send.

• Export reports to Word — A finished research report can now be downloaded as an editable Word document, not just a PDF, with its citations and formatting intact — so you can revise it directly in your own work.
• Deep research survives a dropped connection — A long research run no longer breaks if your network briefly drops; it keeps working in the background and picks up right where it left off when you return, so you never lose the answer.

• More thorough, more accurate reports — Research reports now back far more of their content with in-line citations and leave fewer sections lightly sourced, pull fresher figures from broker research and earnings calls, and correct how key return and profitability measures are calculated.
• More analytical tools in research — When building its analysis, the assistant can now run correlation and risk comparisons between securities and against the market, weigh relative performance and seasonality, and draw on a wider set of economic data.

• ETF and index coverage — Look up detailed ETF data — fund profile (expense ratio, assets under management, net asset value, sector exposure), full holdings, and country weightings — and have the assistant track ETFs and the constituents of an index.

• Reports that could fail to generate — Fixed an issue where a research report could occasionally fail partway through because an underlying data source was temporarily unavailable.
• News analysis — Resolved issues that could affect the quality and reliability of news analysis.
• Dashboard flicker — Fixed a full-page flicker that could appear when moving between dashboard pages, and trimmed background refreshes for a steadier, smoother experience.

​

Market-wide M&A activity, deeper research data, and sharper citations

• M&A Activity — M&A is now its own top-level section with a market-wide feed of the latest deals — filter by timeframe, status, and deal size, and search any company’s full deal history as buyer or target — plus a new M&A panel on your home dashboard for deals at a glance.
• Document search — Switching between search styles when researching filings and documents now keeps your query and current results in place instead of clearing them, so you can re-run the same search a different way without retyping.

• More accurate citations — When the assistant reports a figure such as revenue, EPS, or margin, it now cites the primary source — a company’s own press release, earnings presentation, or filing — rather than second-hand coverage, backs each distinct fact with its own source, and no longer shows stray citation markers that don’t link anywhere.

• M&A deal data in research — Ask the assistant about any company’s M&A history — completed, pending, withdrawn, even rumored deals — with deal value, the stake acquired, cash-versus-stock terms, the acquirer and target, the advisors on each side, and a scan of the latest deals across the market.
• Supply-chain coverage in research — Ask the assistant to map a company’s suppliers, customers, competitors, and partners — joint ventures, distribution, licensing, manufacturing, R&D alliances, and equity stakes — including customer-concentration detail and side-by-side peer comparison.

• M&A deal filters — Fixed an issue where filtering M&A deals by more than one status at once (for example completed, pending, and withdrawn together) returned no results.
• Presentation generation — Fixed an error that could stop a presentation from generating when a specific slide layout was requested; presentations now generate reliably whichever layout you choose.

​

Supply-chain networks, M&A activity, and a much bigger AI research toolkit

• Supply Chain — Map any company’s customers, suppliers, and partners in an interactive relationship network, then drill into each connection in a sortable table — straight from the company’s profile.
• M&A Activity — Follow a company’s mergers and acquisitions in one place: deals where it was the target or the acquiror, the advisors behind them, and how deal activity has trended year over year.

• New analyst skill library — Ask the assistant to run full professional analyses end to end — DCF and LBO models, trading comps, due diligence, investment memos and pitch books, sector deep-dives (banks, biotech, energy, semiconductors, REITs, insurance, and more), and scenario, stress, and Monte-Carlo work — powered by a large new library of built-in research skills you can browse and apply.

• Deeper analyst coverage — Company estimates now carry richer analyst consensus ratings, price targets, and recent firm rating revisions, so you can see how sell-side conviction is shifting at a glance.

• Document search — Fixed distracting loading spinners and pop-ups that could appear when searching for text inside filings and documents.

​

Performer screening grids, file-aware Chat, and faster news

• Performer screening grid templates — New Top Performers and Bottom Performers checklist templates let you score and rank a basket of stocks in a single grid, with ready-made columns for valuation, analyst coverage, relative performance, seasonality, estimate revisions, insider and institutional activity, and dividend and debt trends.
• Files and images in Chat — Attach documents and images directly to a chat and ask questions about them in your conversation.

• More grid templates — Added new dividend and fundamentals column templates and refined the grid-building experience, so more analyses are one click away.
• Improved document search — A redesigned search panel with sharper keyword highlighting and in-line PDF viewing makes it faster to find the exact passage inside your documents.

• Deeper, faster news — More thorough and quicker news search, plus one-step recent-news recaps for any company.

• Reliable cells on large grids — Cells in very large grids no longer come back blank under load and now fill reliably.
• Cells from transcripts and filings — Fixed grid cells that draw on earnings-call transcripts and filings occasionally returning blank; they now populate.
• Charts and exports — Resolved issues with chart generation and exporting reports to documents.
• Events, saved items, and logos — Fixed display problems affecting live events, saved items, and company logos, along with general session stability.

​

A major company-financials upgrade — plus cloud sources for grids, steadier automations, and cleaner reports

• Standardized and as-reported financials — Switch any company’s income statement, balance sheet, and cash flow between a standardized view that lines up comparably across companies and the company’s own figures exactly as filed — both in the data viewer and when you ask the assistant.
• Adjusted (non-GAAP) metrics — A new Adjusted Metrics tab brings a company’s management-defined figures — adjusted revenue, EBITDA, EPS, FFO, and more — alongside the core statements, and the assistant can pull them on request.
• Add grid documents from your cloud — Attach source documents to a grid straight from Google Drive, OneDrive / SharePoint, Gmail, and Outlook, in addition to uploading from your computer.

• Redesigned financial charts — Chart any financial line item or business segment with switchable bar, stacked-bar, line, and area views, on-chart CAGR and total-change stats, and one-click export to a branded image.
• More complete earnings calendar — Scheduled events now surface even when they are filed under a company’s other share class or a cross-listing, so dual-class and cross-listed names no longer slip through.
• Larger event automations — Track up to 200 companies in a single event-triggered automation, raised from 15.
• Clearer automation status — An active event automation now tells you when no matching event is scheduled in the next 90 days, with a one-click option to run a one-off report instead.

• Steadier long research sessions — The assistant now holds much more of a long, multi-step research task in active memory, so it stays on track through complex work instead of losing its place and repeating earlier steps.

• Faster, more accurate financials — Financial statements, ratios, valuation multiples, and segment detail across the platform now draw on a faster, more accurate data source.
• Operating KPIs and segment breakdowns — Pull a company’s key operating metrics — subscriber counts, assets under management, store counts, and other published KPIs — alongside revenue, EBIT, EBITDA, and gross profit broken out by business line and geography.

• Correct quarter-by-quarter financials — Quarterly statements, ratios, and segment breakdowns now show each quarter in its own column instead of merging quarters of the same fiscal year, so quarter-over-quarter trends read correctly.
• Truer technical-indicator charts — Technical indicators no longer drop valid zero readings or plot missing data as a misleading zero, and multi-line indicators now show partial data while still warming up.
• More reliable multi-company automations — Automations that watch several companies at once no longer occasionally drop pending alerts.
• Cleaner finished reports — Generated reports no longer show stray calculation code, scratchpad notes, or placeholder charts, and they reliably keep their real charts and diagrams through the quality-review pass.
• Reliable final answers — Fixed a case where the assistant could finish its work but occasionally fail to deliver its written answer.

​

Data Rooms, Chat, and Agent Studio

• Data Rooms — Bring a set of files, filings, or transcripts together and ask questions across all of them, with citations.
• Chat — Ask a research question and get a fast, cited answer to shape your next step.
• Agent Studio — Turn a repeatable research process into an agent that runs on its own.