xdge | Blog

Release notes for xdge upgrades, as well as news, vision, and thoughts throughout our journey.

xdge | Blog

Ditch the Meetings and Effortlessly Capture & Rewind Your Day

Ditch the Meetings and Effortlessly Capture & Rewind Your Day

The universal problem

If you’ve ever thought, “This meeting could’ve been an email” - you’re not alone. The truth is, meetings, emails, & chats all come with their own frustrations.

Not everything that ends up as a meeting needs to be one. In practice, most meetings aren’t for real-time collaboration or decision-making; they’re for updating others, sharing context, or explaining something once so that a group of people can hear it simultaneously. These kinds of meetings persist not because they’re efficient, but because they’re familiar. People need to express their thoughts out loud. Others need to feel included. You can’t replace this kind of communication with a document or a Slack message, because reading a paragraph of text doesn’t carry the same clarity, tone, or nuance as hearing someone explain it directly. And long Slack threads sprawl - then become impossible to search later. At the same time, synchronizing everyone’s schedules to have that conversation in real time is often unnecessary overhead.

In short -

  1. People hate being stuck in meetings.
  2. People also hate long Slack threads.
  3. Search across threads & meetings is broken.

Introducing Clips

The goal of Clips is to address this gap.

With Clips, you capture your thoughts with a single click - audio, video, or screen-share - with no time limits. When you’re done, share a secure perma-link; recipients don’t need apps or signups. They can play in high definition, watch at up to 2× speed, pop it into picture-in-picture to multitask, or skip the video entirely and interact with the automatically generated transcript. That transcript includes speaker labels, timestamps, and an embedded chat interface where you can ask questions about what was said.

For example, you can ask Clips, “What were the key takeaways?” or “Who was responsible for the next steps?” and it will return an answer derived directly from the transcript, along with references to the sections it pulled from.

So the output of a single recording becomes more than just a video; it becomes a transcript you can read, a summary you can scan, a set of action items, and a searchable record you can come back to later.

With Clips, you can:

🎥 Record once, broadcast asynchronously, and skip the meeting.

🤖 Send an AI notetaker when you can’t avoid a meeting.

📝 Receive auto-generated summaries and action items.

💬 Chat with transcripts enhanced by speaker detection.

🔎 Search for answers across clips & workspace apps.

When Live Meeting Is Unavoidable

For the conversations that truly need to happen, Clips syncs with your calendar and can send an AI notetaker to the meeting on your behalf to capture notes and key takeaways, ensuring you never miss out.

Unlike other tools that rely solely on bots to attend meetings, xdge offers two methods for capturing conversations, tailored to the situation. Inside the Meetings app, you can choose Join on My Behalf, where a visible bot joins the call for you, captures everything, and delivers a full transcript and summary afterward - ideal when you can’t attend in person.

If you’re already present, you don’t need the bot at all. Use Clips, from the Clips tab or directly in-browser, to capture from your own device. This is invisible to others and avoids bot friction, while still giving you the same high-quality transcript and searchable notes. You control how context is captured, without the awkward workarounds.

One Search Bar for Your Work

Your recorded clips and meeting notes don’t live in a silo. They sit alongside the rest of your workplace knowledge so you can use a single, unified search & answer bar across chat threads, meeting transcripts, recorded clips, and other workspace apps. The result is faster retrieval of answers - whether they came from a clip you recorded yesterday or a thread from last quarter.

A New Way of Working

This enables a different way of working: instead of coordinating everyone to be available at the same time, you capture your thinking once and allow others to consume it whenever it fits into their schedule. This is more than saving time - though it does that. It’s about decoupling communication from presence, conveying what needs to be said without requiring others to be physically present to hear it. The value here compounds.

A five-minute explanation, recorded once, can be reused, referenced, or watched by ten people without the speaker needing to repeat themselves. If someone joins the team two weeks later, they can still catch up just by watching the same link. If someone forgets what was discussed, they can search the transcript. If someone wasn’t paying full attention, they can jump back to the relevant part. A live meeting doesn’t have these properties; once it’s over, it’s mostly gone, except for whatever someone remembered to write down.

The Transcript at the Center

The transcript is not a secondary feature. It’s the center of how Clips works. Every recording is automatically transcribed and divided by speaker.

You can scroll through it like a chat log. Click a sentence to jump to that point in the video. Copy-paste a quote. Filter by speaker or ask questions directly, such as “What did Jane say about the timeline?” or “List all the objections made during the discussion.” This turns recordings into something closer to a structured memory than a passive playback.

Sharing Without Friction

Because every recording is hosted via a simple link, sharing is effortless. You don’t need to export files, attach them to messages, or upload anything manually. The person you send it to can open the link and immediately view the video, scan the transcript, or ask the AI a question - no account required. This makes it suitable for internal team updates as well as external communication, customer walkthroughs, design feedback, onboarding guides, and support responses, all of which benefit from clear explanations, asynchronous access, and persistent, searchable output.

Personalized Coaching

Clips doesn’t just capture context - it optionally helps you improve how you deliver it. Every recording comes with built-in coaching insights on:

  • Speaking time - spot imbalances.
  • Clarity & pacing - see if you’re too fast, slow, or filler-heavy.
  • Interruptions - track overlaps and cut-ins.
  • Message impact - check if your key points land.

These reports are invaluable for sales calls, internal presentations, and updates where clarity matters most. And because they’re automatic, you get actionable feedback with zero setup - turning every clip into both a communication asset and a coaching tool.

The End of Wasteful Meetings

The goal isn’t to eliminate meetings for the sake of efficiency. It’s to reserve live attention for when it’s genuinely needed and use async communication everywhere else. Many recurring or update-style meetings can be replaced outright. Others can be shortened or front-loaded by sharing a recording before the call, so that live time is used only for questions or decisions.

A Compounding Archive of Knowledge

Finally, your thoughts and meeting notes are now stored alongside all other workplace knowledge. The result? You get a single, unified search & answer bar for your chat threads, meeting transcripts, recorded clips, and all other workspace apps!

When you record something once and make it easy to revisit, extract from, and share, you’re not just avoiding a meeting; you’re creating an asset. You’re capturing not just what was said, but how it was said - and making it available in a way that written notes or secondhand summaries can’t match. This isn’t about replacing meetings with video. It’s about treating explanations, decisions, and shared context as durable knowledge.

So, if you’ve ever thought, “This meeting could’ve been an email,” you were probably right. But it would’ve been a bad one. It should instead have been a Clip.

Intro Video

Learn more at xdge.ai

Release Notes: August 1-18, 2025

Release Notes: August 1-18, 2025

Overview


This release introduces significant enhancements to our AI-powered workplace search and automation platform. Key highlights include the launch of Chat for Meetings, and Workflow improved Deep Research Assistant reasoning interface, enhanced notification control settings.


🔍 New Features

Chat for Meetings

You can now have interactive conversations with your meeting transcripts and recordings. Ask questions about specific meetings, get summaries, extract action items, or dive deep into discussion topics. This feature works seamlessly with both live and recorded sessions.

Per-User Notification Settings

Take control of your notification preferences with granular settings for email, Slack, and in-app notifications. Customize which features send you updates and choose your preferred delivery methods to reduce notification fatigue while staying informed about what matters most.

Enhanced Virtual Notetaker for Ad Hoc Meetings

Our virtual notetaker now supports ad hoc meetings, automatically capturing and transcribing impromptu discussions when invited to your meeting links.

Advanced Channel Search with At xdge

The Slack bot now supports querying open channel discussions, analyzing the last 100 messages or 48 hours of conversation (whichever is smaller), including threaded discussions for comprehensive context.

Slack Integration for Deep Research Workflows

Deep Research Workflow reports now include rich summaries and headlines directly in Slack DMs, reducing the need to switch between applications to review your automated research results.


🛠️Major Improvements


Redesigned Assist & Deep Research Assistant Reasoning Interface
We've completely enhanced the reasoning steps display for both Assist and Deep Research Assistant:

  • Progressive disclosure: Reasoning steps now appear one at a time for a more natural experience
  • Enhanced visual design: Better icons, typography, and layout, matching our latest design system
  • Detailed app insights: See exactly which applications were searched and how many results were found
  • Improved readability: Better text formatting and container layouts prevent overflow issues

Accelerated Follow-up Query Processing
Implemented parallel processing with Cerebras integration for follow-up queries:

  • Intent checking: The system automatically determines if your follow-up needs previous context
  • Faster processing: Parallel execution reduces wait times by 5-7 seconds
  • Smarter context handling: Better decision-making about when to use conversation history vs. new searches

Workflow Error Recovery & Debugging
When Deep Research Workflows encounter issues, you can now:

  • Chat with execution logs: Understand what went wrong and get suggestions for improvement
  • View detailed progression: See which steps succeeded and which failed
  • Get actionable feedback: Receive guidance on how to improve future workflows

Performance Enhancements


Search Result Optimization
Implemented dynamic result reranking with visual feedback:

  • Progressive loading: Results appear quickly, then reorder based on AI relevance scoring
  • Shimmer animations: Smooth visual transitions during result reorganization
  • Improved accuracy: Better result ranking puts the most relevant content first

Response Time Improvements

  • Reduced Time-to-First-Token (TTFT) for common queries from 10+ seconds to 3-4 seconds
  • Enhanced compression matrix algorithms for faster content processing

System Reliability

  • Enhanced auto-routing logic to prevent conflicting response streams
  • Better resource allocation prevents delays during high-usage periods

🐞 Bug Fixes


Media & File Handling

  • Resolved OneDrive file picker detection problems that prevented proper file uploads
  • Corrected transcription generation for Zoom-scheduled meetings when leaving while recording

Search & Discovery

  • Fixed At-xdge search failures due to RRF combined scoring inconsistencies

User Experience Improvements


Workflow & Navigation

  • Smart default pages: Workflow Template tab now opens by default when no existing workflows are present
  • Gradual step revelation: Thinking steps appear progressively rather than all at once
  • Improved URL formats: Consistent link structures across meetings, workflows, and reports
  • Better time filtering: Recent mode now correctly filters content from the current timeframe

Content Display

  • Enhanced formatting consistency: Uniform styling across chat interfaces and follow-up queries
  • Proper text wrapping: Eliminated content overflow issues in reasoning containers
  • Clear status indicators: Better visibility of "Generating response" and other processing states
  • Improved placeholder text: Context-appropriate hints (e.g., "Ask anything about this transcript" for meetings)

These release notes cover all major changes deployed between August 1 and 18, 2025. For technical support or questions about these updates, please contact our system administrator or support team.

Release Notes: July 19 - July 30, 2025

Release Notes: July 19 - July 30, 2025

Overview

This release delivers significant enhancements to our AI-powered workplace productivity platform, with major improvements to Deep Research Assist, meeting functionality, and overall system performance. Key highlights include the introduction of advanced follow-up processing, enhanced meeting bot capabilities, user notification controls, and numerous UI/UX refinements across the platform.


🔍 New Features

Advanced Follow-Up Processing

Enhanced AI Response Speed and Accuracy

  • Implemented parallel processing for follow-up queries using Cerebras acceleration
  • New intent checking system determines when to use previous context vs. perform fresh searches
  • Significantly reduced response times by eliminating 5-7 second bottlenecks in follow-up interactions

Chat with Meetings

Interactive Meeting Analysis

  • Added the ability to chat directly with meeting transcripts and recordings
  • Ask questions about specific meetings and get AI-powered responses
  • Integrated across both frontend and backend systems for seamless user experience

User Notification Controls

Granular Notification Management

  • New per-user settings to control notifications via Slack, email, and other channels
  • Customize notification preferences for different features and apps
  • Reduces notification fatigue while maintaining important updates

Advanced Workflow Testing

Test Features

  • Added comprehensive testing interface for Deep Research Workflows
  • Users can chat with execution logs to identify what worked or failed
  • Provides detailed feedback for workflow optimization before scheduling

🛠️Major Improvements

Deep Research Assist Enhancements

Comprehensive UI and Functionality Overhaul

  • New tabbed interface with Answer, References, and Reasoning tabs
  • Progressive loading indicators showing real-time processing steps
  • Stop button functionality to halt ongoing research processes
  • Browser notifications with direct links to completed reports
  • Improved formatting with better handling of tables, lists, and text structure
  • Follow-up capabilities allowing continued conversation with research reports

Meeting Bot Reliability

Enhanced Meeting Integration

  • Fixed critical issues with bots not joining scheduled and ad-hoc meetings
  • Resolved Zoom-specific connection problems affecting transcript generation
  • Improved meeting URL handling and error page resolution
  • Enhanced notetaker functionality for various meeting platforms

Search and Assist Performance

Speed and Reliability Improvements

  • Integrated Cerebras acceleration for faster query processing
  • Reduced Time to First Token across search and assist functions
  • Enhanced query routing between standard search and deep research modes
  • Improved handling of complex queries and large datasets

⚡Performance Enhancements

AI Processing Speed

  • Cerebras Integration: Implemented Cerebras acceleration for Stage 1 prompts across search and assist functions
  • Parallel Processing: New follow-up flow processes multiple operations simultaneously
  • Reduced Latency: Significant improvements in query response times

System Reliability

  • Fallback Systems: Added Gemini backup for token limit scenarios in operations
  • Error Handling: Enhanced error recovery and processing continuation
  • Resource Optimization: Improved handling of large datasets and complex queries

🐞Bug Fixes

Meeting Functionality

  • Fixed Zoom transcript issues: Resolved problems where meeting tiles disappeared and transcripts weren't generated after leaving recordings
  • Corrected bot joining failures: Addressed Recall API errors preventing bots from joining meetings
  • Resolved extension conflicts: Fixed side panel failures when users weren't logged into LinkedIn
  • Meeting URL errors: Eliminated error pages for valid meeting URLs

Collection Management

  • Item count synchronization: Fixed delays in collection folder item counts after adding or removing files
  • Reference handling: Corrected issues with collection references not opening specific notes/cards

Pricing and Display

  • Updated pricing information: Corrected upgrade plan displays showing outdated $9/mo instead of current $15/mo pricing
  • UI consistency: Resolved various display and formatting inconsistencies across components

Deep Research Assist

  • Email notifications: Fixed broken images and formatting in DRA email notifications
  • Reference links: Resolved duplicate and incorrectly formatted reference links
  • Processing states: Fixed issues with blank UIs and interrupted streaming responses
  • Tab functionality: Corrected tab state management and content display problems

🪄UI/UX Improvements

Deep Research Interface

  • Progressive disclosure: Step-by-step revelation of reasoning processes instead of bulk loading
  • Better loading states: Contextual messages and shimmer effects during processing
  • State preservation: Deep Research responses maintain state when switching between apps
  • Enhanced accessibility: Improved keyboard navigation and screen reader support

Reference and Citation Display

  • Clickable references: Made reference cards interactive in assist responses
  • Improved formatting: Better display of citations, links, and source materials
  • Consistent styling: Unified appearance across different content types and sources

These release notes cover all major changes deployed between July 19 and 30, 2025. For technical support or questions about these updates, please contact our system administrator or support team.

Release Notes: July 9 - 18, 2025

Release Notes: July 9 - 18, 2025

Overview

In this release, we've enhanced search and assistant capabilities with faster response times, resolved critical meeting bot issues, and improved overall system reliability. Key improvements include optimized AI prompt processing, better context understanding, and enhanced user experience across all major features.

🛠️ Major Improvements

Search and Assistant Performance

We've implemented comprehensive performance optimizations that significantly reduce response times:

  • Faster Query Processing: Optimized prompt templates reduce token usage by 70-80% while maintaining accuracy
  • Parallel Processing: Search operations now run simultaneously, eliminating sequential delays
  • Streamlined AI Responses: Simplified output structures provide faster generation and parsing

Meeting Bot Enhancements

Enhanced meeting bot functionality with improved user recognition and profile management:

  • Better User Identification: Meeting bot now properly displays participant names and information
  • Improved Profile Display: Enhanced thumbnail and profile image handling across Google Meet and Microsoft Teams

🐞 Bug Fixes

Meeting and Collaboration Features

  • Fixed recurring Day0 meeting screen that appeared daily despite proper bot configuration
  • Resolved meeting scheduler visibility issues where meeting organizers weren't appearing in participant lists
  • Corrected meeting bot profile image display problems in both Google Meet and Microsoft Teams

User Interface and Experience

  • Fixed dropdown menu overflow issues where the three-dot menu would move outside container bounds during page scrolling
  • Resolved search query text overlap with the clear button when using long search queries
  • Improved overall UI component positioning and responsiveness

Deep Research Workflows

  • Fixed timezone display issues in workflow reports - timestamps now show in user's local timezone instead of UTC
  • Resolved Salesforce workflow failures that prevented data retrieval
  • Fixed missing references in Google Drive reports when searching for specific file titles

At-Ayraa Assistant Features

  • Improved context understanding when using channel names and thread references
  • Fixed issues where internal Slack naming conventions appeared in user responses
  • Enhanced accuracy when interpreting queries with special characters like hashtags

System Reliability

  • Resolved duplicate subscription creation issues for free trial users
  • Fixed Box file picker network errors during document selection
  • Improved error handling and fallback mechanisms for AI model failures

⚡Performance Enhancements

AI and Search Optimization

  • 500ms Faster Search: Embedding generation now starts in parallel with initial query processing
  • Reduced Token Usage: New prompt structures decrease output tokens by up to 95% for certain operations
  • Optimized Response Generation: Simplified JSON structures eliminate parsing overhead
  • Enhanced Auto-Routing: Decision processes now run in parallel to avoid blocking fast responses

Infrastructure Improvements

  • Optimized disk space usage and automated cleanup processes
  • Enhanced logging efficiency to reduce unnecessary system overhead
  • Improved cost monitoring with integrated AI and AWS expense tracking

🔒Security Updates

  • Enhanced model fallback mechanisms ensure continued service availability during AI provider outages
  • Improved error handling prevents exposure of internal system information in user-facing responses

🧹Deprecations & Removals

No features were deprecated or removed in this release. All existing functionality remains available with improved performance and reliability.

Release Notes: July 2 - July 8, 2025

Release Notes: July 2 - July 8, 2025

Overview

This release introduces our groundbreaking Deep Research Workflows feature alongside significant improvements to search performance, bug fixes for core functionality, and enhanced integrations with Microsoft Outlook. We've focused on streamlining user experiences while adding powerful new automation capabilities.


🔍 New Features

Deep Research Workflows

Deep Research Workflows revolutionize how you extract insights from your workplace data. This powerful new feature allows you to create, schedule, and run sophisticated research workflows that automatically generate comprehensive reports from your integrated apps. With Deep Research Workflows, you can:

  • Set specific instructions and let AI handle complex multi-step research tasks
  • Schedule workflows to run automatically at your preferred times
  • Customize visibility settings for seamless team collaboration
  • Select specific app connectors to include in your research

The feature represents a significant advancement in research capabilities, allowing you to focus on deep work while automating routine information gathering. Deep Research Workflows brings our vision full circle - creating a truly smart enterprise where your entire operation runs beautifully on "auto-pilot".

Enhanced Microsoft Outlook Integration

  • Outlook Email Support: Create automated workflows that analyze and report on your email communications, including filtering messages, identifying key conversations, and generating summaries
  • Outlook Calendar Integration: Build workflows that automatically track meeting schedules, analyze calendar patterns, and generate attendance reports
  • Improved File Picker: Enhanced support for Box and OneDrive file uploads with better error handling and processing

🛠️ Major Improvements

Assist Performance Enhancements

  • Two-Stage Prompting Architecture: Completely redesigned the Assist pipeline to eliminate latency bottlenecks. Search results now appear faster by running AI classification and workspace search simultaneously
  • Streamlined Query Processing: Removed collection-only search flows from default Assist mode to improve response times and reduce processing overhead
  • Optimized Search Routing: Disabled infrequently-used JQL/SOQL interpretation in default Assist behavior, focusing resources on most common use cases

Enhanced Search Intelligence

  • Improved Context Understanding: DRW now retrieves surrounding message context for better thread relevance detection, reducing false positives in automated workflows
  • Better Keyword Handling: Enhanced search logic to properly prioritize all keywords in queries, not just the most common terms

🐞 Bug Fixes

Authentication & Account Management

  • Fixed Email Sign-up Flow: Resolved issue where users were logged out after completing email verification and profile setup
  • Account Deletion Visibility: Corrected regression where personal accounts under tenant management incorrectly showed delete options

AI & Search Functionality

  • Assist Query Reliability: Fixed issues with Built-in Collection (BIC) queries getting stuck or showing incomplete responses when switching between AI models
  • Confidence Score Accuracy: Corrected display of confidence scores for GPT-powered queries
  • Search Result Consistency: Improved keyword search accuracy, particularly for less common terms and specialized vocabulary

Collections & Workspace Management

  • Real-time Folder Updates: Fixed collection folder visibility issues that required manual cache clearing to reflect changes across users
  • Workspace Deletion: Resolved blocking issue preventing workspace deletion that was affecting new user onboarding

Meeting & Communication Features

  • Meeting Summary Formatting: Standardized font sizes in automated meeting summary emails for consistent presentation
  • Slack Integration Stability: Fixed multiple notification issues including duplicate thread summaries and excessive discover alerts
  • Calendar Search Accuracy: Resolved parametric search issues for Outlook Calendar queries with date and time filters

⚡Performance Enhancements

Query Processing Optimization

  • Enhanced Monitoring: Added comprehensive logging for search and assist query processing times to enable better performance tracking
  • Reduced Response Times: Two-stage prompting architecture cuts average query response time by eliminating blocking AI evaluation steps
  • Streamlined Data Flow: Optimized assistant routing to focus on high-impact features while reducing system overhead

Infrastructure Improvements

  • Build Stability: Resolved frequent frontend build failures that were impacting development velocity
  • Cost Monitoring: Improved tracking and reporting of AI usage costs across different features and use cases

🔒Security Updates

Access Control Improvements

  • Production Environment Security: Removed overly permissive visibility settings from production test environments
  • Enhanced Token Management: Fixed Slack bot automation authentication issues to prevent unauthorized access

🧹 Deprecations & Removals

Feature Streamlining

  • Collection-Only Search: Removed dedicated collection-only search flow from default Assist mode (still available via collection filters)
  • JQL/SOQL Interpretation: Disabled automatic JQL/SOQL query interpretation in default Assist behavior to improve performance

🔧Infrastructure & Operations

System Improvements

  • Enhanced Cost Tracking: Implemented separate API keys for different AI use cases to enable better cost analysis and usage monitoring
  • Improved Automation: Fixed on-call job scheduling and Slack bot automation reliability
  • Development Tools: Updated feedback collection systems and resolved Jenkins plugin compatibility issues

Release Notes: June 24 - July 1, 2025

Release Notes: June 24 - July 1, 2025

Overview

This major release brings groundbreaking research capabilities with Deep Research Workflows, comprehensive Outlook Email support, and significant improvements to search accuracy and AI performance. We've also completed our brand refresh to xdge and upgraded our AI models for better responses across the platform.


🔍 New Features

Deep Research Workflows:

Revolutionary Research Automation

Deep Research Workflows revolutionizes how you extract insights from your workplace data. This powerful new feature allows you to create, schedule, and run sophisticated research workflows that automatically generate comprehensive reports from your integrated apps.

Key capabilities:

  • Automated Research Tasks - Set specific instructions and let AI handle complex multi-step research processes
  • Smart Scheduling - Configure workflows to run automatically at your preferred times
  • Team Collaboration - Customize visibility settings for seamless team sharing
  • Selective Data Sources - Choose specific app connectors to include in your research
  • Comprehensive Reports - Generate everything from release notes and JIRA activity reports to sales analytics and workspace recaps

Deep Research Workflows represents a significant advancement in enterprise research capabilities, allowing you to focus on strategic work while automating routine information gathering.

Full Outlook Email Support

Complete integration with Outlook Email is now available, enabling you to search, analyze, and get insights from your email communications alongside other workplace data.

Enhanced Meeting Infrastructure

New meeting server capabilities provide improved stability and performance for all meeting-related features.


🛠️ Major Improvements

Brand Refresh to xdge

We've completed our comprehensive rebrand from Ayraa to xdge across the entire platform, including:

  • Updated branding throughout the user interface
  • Refreshed email templates and communications
  • Consistent naming across all touchpoints

Search Intelligence Enhancements

  • Relevance Improvements - Significantly better search result accuracy and ranking
  • Recency Bias Optimization - Smarter prioritization of recent content in search results
  • Deep Search vs Standard Search - Clearer differentiation between search modes for optimal results

Enhanced Deep Research Workflows Features

  • Improved Thread Selection - Better AI judgment for identifying relevant discussion threads
  • JIRA Integration Tools - New history tracking capabilities for project management insights
  • Slack Summary Enhancements - More accurate and comprehensive workspace activity summaries

🐞 Bug Fixes

File Management & Uploads

  • Fixed image file upload issues when using Box or OneDrive file pickers
  • Resolved inconsistencies between desktop and cloud storage uploads

Search & Query Reliability

  • Fixed search failures when keywords clearly existed in content
  • Resolved calendar query issues for meeting searches
  • Corrected Outlook single keyword search problems
  • Fixed collection queries that were incorrectly failing

Notification & Communication Issues

  • Eliminated duplicate thread notification problems
  • Fixed thread summarization nudging occurring multiple times
  • Resolved missing references in Deep Research Workflows and Assist

User Interface & Authentication

  • Corrected case sensitivity issues in email matching
  • Fixed Microsoft Account login problems
  • Standardized text casing throughout the interface
  • Resolved user management display inconsistencies

Data Integration & Reporting

  • Fixed Google Calendar report data display issues
  • Corrected missing daily meetings in calendar reports
  • Resolved Salesforce integration problems for specific customers
  • Fixed formatting issues in Google Drive reports

⚡ Performance Enhancements

Cost Optimization

  • Resolved OpenAI API cost spike issues through improved query optimization
  • Enhanced efficiency in AI model usage across the platform

Response Quality

  • Fixed overly verbose AI responses that were exposing unnecessary reasoning steps
  • Improved response formatting consistency across different query types
  • Optimized Google Calendar automation query performance

System Reliability

  • Enhanced automation scan accuracy for various app integrations
  • Improved weekend processing reliability for scheduled workflows
  • Better handling of time-based queries and filters

🔒 Security Updates

  • Enhanced authentication flows for Microsoft Account integration
  • Improved data handling for multi-tenant environments

Welcome to Playground: Unlocking Agentic Workflows Without Limits

Welcome to Playground: Unlocking Agentic Workflows Without Limits

“What if your workplace agent could write its own code and run it on the fly across your tools?”

That question is no longer a hypothetical.

Over the past few weeks, our Head of Engineering, Sid, quietly built a system that takes agentic workflows to their logical conclusion, one built to adapt beyond brittle templates, pre-coded functions, or tightly scoped toolchains. It creates its own integrations on the fly. It runs its own logic. It fixes itself when it fails. And it keeps you in control the entire time.


The Problem: Prewritten Agents, Prebaked Limits

Workplace AI today is largely constrained by the same pattern:

  • Predefined tools and scripts
  • Static integrations (MCPs, function calls)
  • Narrow, brittle workflows

If the system hasn’t been explicitly told how to do something, it simply can’t.

For example:

“Who moved the most JIRA tickets last month by status?”
“Summarize customer feedback across Intercom, Gmail, and Slack.”
“Find hiring actions from meeting notes that were never followed up on.”

Sounds doable, but unless someone’s already coded that exact workflow, you’re out of luck.


The Breakthrough: Code-Generating, Self-Healing Agents

Sid’s breakthrough changes all of that.

At xdge, we’ve built an agentic execution engine with just two core superpowers:

1. Code Generation on Demand

The agent writes custom code for your apps and your ask — in real-time. No need for pre-integrated tools.

2. Prompt Generation as a Skill

It can generate and use its own prompts. Think of this as "reasoning about reasoning," the AI deciding how to talk to itself to get the job done.

And here’s the kicker: all generated code runs safely inside a temporary, self-destructing container. Secrets never leave your environment, only placeholders are used during code generation, with real credentials securely injected at runtime.

This guarantees total security and full control in every execution.


How It Works

Every time you issue a task, say:

  • “Plan my day using emails, JIRA, Slack, and my calendar”
  • “Track stalled deals with no activity in 14+ days”
  • “Write tickets from product feedback in Slack threads”

… the system does the following:

  1. Breaks the task into atomic steps
  2. Writes fresh code for each step based on app APIs
  3. Runs it securely in a sandboxed environment
  4. Validates the output
  5. If it fails → debugs itself and retries (up to 5 times)
  6. Chains the outputs into a clean, cohesive result
⚠️ Write access? Before executing any write action (e.g., send message, update task), it pauses, shows you exactly what will happen, and waits for your approval.

A Glimpse into Execution

In the demo below, the system was asked to:

“Analyze my Slack, Gmail, JIRA, and Google Calendar data and generate an action plan for today.”

It:

  • Queried all tools (with no pre-built functions)
  • Extracted priorities, deadlines, blockers
  • Picked up the user had a flight scheduled (!)
  • Created a time-blocked plan that included email replies, JIRA priorities, meeting prep, and the flight itself

The output? A fully customized, realistic, live-action schedule. Built and executed in minutes, with zero prewritten templates.


Why This Matters


Playground enables autonomous execution that goes far beyond generating suggestions, it writes real code, runs it safely in a controlled environment, and validates every output before taking the next step. This is true end-to-end reasoning that is not just understanding intent, but completing the task across tools with full context awareness and built-in self-correction. It’s built to be flexible enough to adapt to any stack, secure enough to run in real-world environments, and extensible enough to handle use cases no one has anticipated or built for yet. It’s like if ChatGPT had hands.


In short, this is what happens when generative AI stops talking and starts doing.


The Future That Works for You

We believe this is the natural next and perhaps final step in the evolution of workplace AI. One where tools go beyond assisting, they take meaningful action. And one where the agent builds capabilities dynamically, expanding what’s possible as it works.

This is xdge playground, the future of work, now writing itself.


Get Early Access

We're rolling this out in beta soon and if you want to be among the first to try it:


▶️ Watch the full demo here : https://www.youtube.com/watch?v=N51QF22Vs0c
💬 Comment “xdge” or DM us on social media for early access

Release Notes: June 11 - June 23, 2025

Release Notes: June 11 - June 23, 2025

Overview

This release represents a major milestone in our platform evolution, introducing powerful new integration capabilities and significantly enhancing our research and search functionality. We've added comprehensive Outlook Email support, expanded file management with OneDrive integration, and laid the groundwork for next-generation Deep Research Workflows. Additionally, we've upgraded our AI engine and implemented numerous quality-of-life improvements based on customer feedback.


🔍 New Features

Deep Research Workflows

Deep Research Workflows revolutionizes how you extract insights from your workplace data. This powerful new feature allows you to create, schedule, and run sophisticated research workflows that automatically generate comprehensive reports from your integrated apps. With DRW, you can:

  • Set specific instructions and let AI handle complex multi-step research tasks
  • Schedule workflows to run automatically at your preferred times
  • Customize visibility settings for seamless team collaboration
  • Select specific app connectors to include in your research

From automated release notes and JIRA activity reports to sales analytics and workspace recaps, DRW enables workflows you never thought possible.

Outlook Email Integration

We've introduced comprehensive Outlook Email support as a standalone connector, expanding your ability to search, analyze, and get insights from your email communications:

  • Complete Email Search: Search across all your Outlook emails with powerful filtering and relevance ranking
  • AI-Powered Email Assistance: Get intelligent answers and summaries from your email content
  • Dedicated Email Interface: Browse and manage email results in a dedicated tab separate from calendar events
  • Advanced Integration: Full connector management with enable/disable controls and connection status monitoring

OneDrive File Management

Enhanced file picker functionality now includes native OneDrive support:

  • Seamless File Access: Browse and select OneDrive files directly from the assistant chat interface
  • Unified File Experience: OneDrive integration works alongside existing Box support for comprehensive cloud storage access
  • AI-Powered File Analysis: Ask questions and get insights from OneDrive documents through our assistant

Connector Architecture Enhancement

We've restructured our Microsoft Office integration for better flexibility and control:

  • Separated Connectors: Outlook Email and Outlook Calendar are now independent connectors
  • Granular Control: Enable or disable email and calendar functionality separately based on your needs
  • Improved Organization: Cleaner interface with dedicated tabs and management options for each service

🛠️ Major Improvements

Enhanced Search Intelligence

  • Dual Search Modes: Choose between lightning-fast regular search (sub-3 second response) or deep search mode with AI-powered relevance ranking for maximum precision
  • Smarter Result Ranking: Implemented recency bias that prioritizes newer content when multiple results have similar relevance scores
  • Improved Query Handling: Better processing of natural language queries including time-based filters like "on Sunday"

AI Engine Upgrade

  • Claude 4 Integration: Upgraded from Claude 3.7 to Claude 4 across the entire platform for enhanced intelligence and more accurate responses
  • Refined AI Responses: Reduced verbosity and eliminated logical step leakage in general queries for cleaner, more focused answers

Deep Research Tool Enhancements

  • JIRA Audit Capabilities: Added comprehensive JIRA change tracking that monitors ticket movements between states during specified time periods
  • Enhanced Reporting: Improved formatting and reliability across Google Calendar, Google Drive, and Salesforce research tools
  • Better Reference Management: Resolved issues with missing citations and references in research outputs

User Experience Improvements

  • Mobile User Guidance: Implemented smart redirection that guides mobile users to use desktop browsers for optimal experience
  • Responsive Design Updates: Enhanced signup and onboarding flows to work seamlessly across different screen sizes
  • Improved Workflow Creation: Streamlined the workflow setup process with better error handling and user guidance

🐞 Bug Fixes

Search and Assistant Fixes

  • Fixed "Anytime" Filter Issues: Resolved problems where queries working in Recent Mode failed in Anytime filter
  • Improved Calendar Query Processing: Fixed handling of specific calendar queries and meeting searches
  • Enhanced Query Accuracy: Resolved cases where searches failed despite exact keyword matches in content

Integration and Connector Fixes

  • Outlook Email Display: Fixed issues where email cards appeared incorrectly under calendar tabs
  • App Visibility: Resolved problems with connected apps not appearing in dashboard and integration pages
  • Authentication Improvements: Fixed Microsoft account login issues and connection status reporting

Workflow and Research Fixes

  • DRW Editor Reliability: Fixed keystroke handling and cursor positioning issues in the workflow instruction editor
  • Report Generation: Resolved failures in Google Calendar and Salesforce research workflows
  • Notification Accuracy: Fixed duplicate discover notifications for inactive threads

⚡ Performance Enhancements

Search Performance

  • Faster Regular Search: Optimized standard search to deliver results in under 3 seconds consistently
  • Improved Search Relevance: Enhanced algorithm provides more accurate and contextually relevant results

System Stability

  • Platform Reliability: Addressed multiple critical stability issues for improved overall system performance
  • Connection Handling: Enhanced connector reliability and reduced authentication-related failures

Security Updates

Authentication Improvements

  • Enhanced Login Security: Improved Microsoft account integration with better error handling and security validation
  • Email Security: Strengthened email-based authentication processes to prevent case sensitivity issues

🧹 Deprecations & Removals

Branding Updates

  • Platform Rebranding: Completed transition from "Ayraa" to "xdge" across all user interfaces, email templates, and system messaging
  • Terminology Updates: Updated platform terminology for consistency and clarity throughout the user experience

This release significantly expands our platform capabilities while maintaining our commitment to reliability and user experience. The addition of Outlook Email integration and OneDrive support, combined with the foundational work for Deep Research Workflows, positions us to deliver even more powerful insights and automation capabilities in upcoming releases.

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to xdge | Blog.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.