Overview

This release delivers significant stability improvements across xdge, an upgraded notification system, smoother workflows, improved meeting reliability, and UI refinements across all screen sizes. We addressed key issues impacting Clips, workflows, and meeting bots, while introducing enhancements that make xdge faster, clearer, and more intuitive to use.

🔔 New Features

Enhanced Notification System

Real-Time Bell Notifications

Stay on top of workspace activity without switching contexts.
The in-app notification center now provides clearer, faster alerts:

• Critical system events appear instantly in the notification bell
• Team activity notifications keep you updated on workspace actions
• Redesigned formatting makes every notification easier to read

Trial Expiration Email Alerts

Workspace admins now automatically receive an email when a 14-day trial expires, ensuring full visibility into subscription status and next steps.

✨ Improved User Experience

Automatic Clip Access

Once a Clip recording completes, xdge now:

• Opens a new tab automatically
• Navigates directly to the newly created Clip
• Eliminates manual searching for new recordings

Upgraded Workflow Search (Slack Activity & Mentions)

Workflow search now delivers faster and more accurate results:

• Surfaces Slack user activity and mentions in seconds
• Provides richer context in search results
• Improves reliability for daily and weekly update workflows

Overall UI Stability

Numerous interface-level regressions have been addressed, including tooltip behavior, label formatting, icon loading, click interaction inconsistencies, connector page redirect loops, and alert errors. The application interface now behaves consistently across views, improving clarity and usability throughout the platform.

🐞 Bug Fixes

User Interface & Design

Layout and responsiveness issues across different screen sizes have been fully resolved. The welcome page now displays correctly on large screens with properly aligned buttons, Collections no longer appear bloated or squeezed, and spacing across responsive breakpoints has been standardized. Oversized text in select views has been corrected, and general layout regressions have been eliminated. The sign-up experience has also been refreshed with an updated design and improved copy, creating a smoother onboarding flow.

Meetings

Meeting automation is now significantly more reliable. Zoom meeting bots consistently join scheduled meetings following recent API changes. Authentication errors when opening meeting links from Slack have been resolved, missing transcripts are now properly captured in production, and scheduled meetings are now correctly displayed in the Upcoming Meetings view. Together, these fixes restore end-to-end reliability for meeting recording and playback.

Clips

Clip recording and playback stability have been restored. Recordings now initiate correctly when the record button is pressed, clip processing completes reliably in production, and corrupt playback issues have been resolved. Thumbnail handling has been improved so placeholders no longer break the interface, and loading states across Clip tiles now display consistently.

Workflows

Workflow execution, notifications, and reporting are now significantly more reliable. Slack-based workflows no longer fail during runtime, and duplicate Slack and email notifications have been eliminated. Workflow reports load instantly without requiring manual refresh. Jira-related workflow errors caused by NullPointer exceptions have been resolved, ensuring consistent data handling across steps. Error messaging now clearly indicates when required applications are not connected, reducing confusion during setup. Deep Research reports are completed successfully, restoring end-to-end reliability for complex workflows.

🧹 Deprecations & Removals

Deep Research has been renamed to Workflows. All existing links automatically redirect to /workflows, but users should update bookmarks and shared links when convenient.

Free trials have been removed for new signups. Existing customers are not affected.

2025 marked a turning point for the company. We evolved from an AI workplace search tool into a complete knowledge and automation platform, and reintroduced ourselves as xdge.

Throughout the year, we focused on delivering the following concrete outcomes for our customers:

• Enterprise automation & actions – turning research and insights into executable workflows that transform ideas into actions. Orchestration, actions, code-generation, safe-mode previews & chat-based flow builder.
• Speed that feels instant – search and assist fast enough to be used continuously throughout the day. A massively multi-threaded & optimized enterprise RAG stack.
• Platform depth & breadth – ensuring users see value through the versatility & robustness of capabilities: Automations & actions, Skills Library, Salesforce & Jira query languages, web search, & the Microsoft stack.
• Meeting intelligence without friction – capturing and operationalizing meeting knowledge without disrupting conversations. Meetings & Clips for synchronous and asynchronous note-taking.
• The rebrand from Ayraa to xdge AI - a new name & brand that goes beyond being just an AI assistant to being an end-to-end enterprise platform.

Below is a recap of what we shipped in 2025, organized around these themes. Together, these releases transformed xdge into a platform teams rely on every day to search, reason, and act.

Enterprise Automations & Actions 🤖

The most crucial product theme of 2025 was automation, not automation that suggests actions, but automation that does the work.

Workflows 2.0: Automatic Tool Creation (Flagship)

Our most significant product achievement of 2025 was the launch of Workflows 2.0.

This marked our transition from AI that answers questions to AI that builds and runs tools on your behalf.

Over the summer, our Workflows evolved dramatically. By September, Workflows 2.0 launched, allowing multi-step automations that can not only gather information but also perform updates and create records in other apps. We introduced a safe execution mode (“dry run”) so you can preview what a workflow will do before it makes any changes, giving you confidence and control. This paved the way for true “hands-off” automation of business processes.

Workflows 2.0 is now the foundation of the xdge platform and the lens through which future features are built.

Real Workflow Examples

Throughout 2025, we built and demoed powerful use cases that replace manual work:

• Slack → Jira Automation: Using our workflow engine, we set up an automated process that listens for bug discussion in Slack and creates Jira tickets autonomously. No more forgetting to file a ticket – the system recognizes when a message describes a bug or task and logs it for you in Jira, complete with details.
• Salesforce Data Sync: We deployed a workflow that monitors deal updates in Slack or email threads and updates Salesforce records accordingly. It even automatically moves opportunities through the pipeline stages. This kept CRM data up to date without having sales reps spend time on data entry.
• Daily Team Report Generation: We launched an automation in November that pulls together activity from Jira and Slack to produce a daily progress report. It compiles what got done each day, what’s stuck, and key achievements, then shares it to a team Slack channel each morning. Managers no longer need to chase updates or assemble reports manually – they wake up to a ready-made digest of yesterday’s work.
• Inbox Zero for Gmail: One of our new workflows (readied in December) uses AI to continuously triage your email. It analyzes incoming Gmail messages and auto-labels them into categories like Important, Needs Response, FYI/Notifications, or Spam. Frequent contacts and urgent content get flagged so you can focus, while routine notifications are tucked away. This helps busy professionals maintain inbox zero with minimal effort, as the AI skim-filters the noise and spotlights what matters.

“Replit for Work” – Custom Tools on the Fly

A highlight of our automation journey was unveiling dynamic tool creation in June, think of it as a “Replit for work”. This technology enables our AI to generate custom mini-apps or connectors on the fly within a workflow. If your process needs to interact with a system that isn’t pre-integrated, the AI can spin up a temporary tool or code to bridge the gap. This is cutting-edge agentic AI, and it means our platform can adapt to handle novel tasks without waiting for our engineers to build a custom integration. It’s a glimpse of how future-proof and flexible automation can be.

Speed That Feels Instant 🚀

In 2025, speed became a top-priority product. Our goal was simple: make xdge fast enough that it disappears into the background of daily work.

Sub-Second Search

We began the year by overhauling our core search engine. In January, we launched an improved Recent Search mode that delivers near-instant answers for the last 90 days of data. By combining semantic retrieval with optimized keyword indexing, search now returns results in seconds across email, Slack, documents, and meetings.

We also introduced more intelligent ranking using Reciprocal Rank Fusion (RRF), ensuring the most relevant results consistently surface first. This reduced noise and made search behavior feel intuitive and predictable.

Faster App Performance

Beyond search, we aggressively optimized platform performance. By early Q1, page load times dropped by 43%, from ~3.5 seconds to under 2 seconds. Additional improvements later in the year, including smarter caching and progressive loading, made navigating Search, Chat, and Workflows feel fluid and responsive, on par with top consumer apps.

Faster AI Responses

Speed mattered just as much for AI assistance. By mid-year, we optimized our AI orchestration pipeline and model selection, dramatically reducing response times. Complex cross-tool questions like “What’s the status of Project X?” now return coherent answers in seconds, enabling a truly conversational workflow that keeps pace with how people think.

Instant Indexing

In September, we eliminated one of the most significant onboarding pain points: waiting for data to index. With pre-crawling, new integrations (Google Drive, Box, Confluence, etc.) now index the most recent 30 days of content within minutes. Teams can start searching immediately, rather than waiting hours or days.

Together, these enhancements made xdge incredibly responsive. Speed isn’t just a nicety; it drives usage. With sub-second searches and immediate answers, our platform became something teams rely on dozens of times a day to find information and get work done without interruption.

Platform Depth & Breadth 🏭

In 2025, xdge AI became more versatile and powerful

Skills Library & Templates

Perhaps our strongest initiative for rapid time-to-value was launching the xdge Skills Library. This is a public, searchable gallery (at skills.xdge.ai) of pre-built workflow templates, prompts, and “skills” curated by our team. We spent months developing and refining templates for everyday use cases in Sales, Engineering, Customer Success, and more. On November 17, we officially launched the library with an initial batch of certified workflows, including a Slack-to-Jira issue creator, the Gmail Inbox Zero triage bot, a Salesforce opportunity updater, and a daily team activity report generator. Instead of starting from scratch, new users can browse the Skills Library and deploy a ready-made workflow in minutes. It’s like having an app store for work automation, one that dramatically cuts down the time and expertise needed to get up and running. For example, a sales manager curious about xdge can grab a “Salesforce Pipeline Health Check” skill from the library and see it working with their data on day 1, without any complex setup. This library continues to grow, and it embodies our commitment to out-of-the-box value.

Salesforce & Jira: Live Business Querying with SOQL and JQL

Early in the year, we shipped Salesforce Integration v2, powered by native SOQL-based querying. Users can now ask natural-language questions about Salesforce data and get live answers, without building reports or dashboards.

Examples:

• “Which enterprise deals over $100k are at risk this quarter?”
• “What opportunities have been stuck in negotiation for more than 30 days?”

We extended the same capability to engineering teams with Jira JQL-powered querying, enabling questions like:

• “Which high-priority bugs have been open for more than two sprints?”
• “What tickets were created after last week’s incident?”

Behind the scenes, xdge translates intent into optimized SOQL and JQL, executes queries safely, and returns structured results instantly. Salesforce and Jira became conversational systems, not static databases.

Web Search, Built In

In 2025, xdge expanded beyond internal knowledge by introducing native web search, fully integrated into Assist, Deep Research, and Workflows.

Users can now combine internal company context with real-time external information, for example:

• Market research blending internal sales notes with live competitor updates
• Vendor evaluations enriched with current documentation and pricing
• Research workflows that continuously monitor the web for changes

Web search in xdge is context-aware, citable, and composable into workflows, making xdge a true research and execution layer, not just a workplace search tool.

Microsoft Ecosystem Integration

We also completed full integration with the Microsoft ecosystem, a critical milestone for enterprise adoption.

xdge now supports:

• Microsoft Outlook Email
• Microsoft Outlook Calendar
• Microsoft OneDrive
• Microsoft SharePoint
• Microsoft Teams meetings
• Microsoft Teams chat

Teams can search and summarize emails, generate follow-ups from meetings, trigger workflows from calendar events, and combine Microsoft data with Salesforce, Jira, Slack, and Drive, all in one place.

For Microsoft-first organizations, this unlocked xdge as a unified intelligence layer across their entire stack.

Meeting Intelligence Without Friction 📈

Meetings contain some of the most valuable information in any organization, but capturing and using that information is usually painful. This year, we made meeting intelligence a seamless part of xdge without friction or third-party bots, so you can focus on the conversation while the AI handles the rest.

First-Party Meeting Infrastructure

At the start of 2025, we made a big architectural change: we brought our meeting transcription bots in-house. Our Meetings v2 update replaced third-party services with xdge’s own proprietary bot. When you schedule a meeting (on Zoom, Google Meet, or Microsoft Teams), our bot joins to record, transcribe, and summarize, now with greater reliability and security since it’s all under our roof.

By November, we had fully migrated all meeting recordings to our own infrastructure, squashing a host of lingering issues (especially around MS Teams) and improving accuracy in speaker identification and note-taking. This gave us more control to innovate faster on meeting features without being held back by external dependencies.

Lower Costs, More Data

By owning the full meeting pipeline and optimizing our tech, we slashed the cost of transcribing meetings by 80% (from about $1 per hour of audio to just $0.20). These savings enabled us to consider offering transcriptions at disruptively low pricing. Our vision is to remove any barrier to recording and analyzing meetings – you shouldn’t have to think twice about hitting “Record” on an important call. We’d rather you capture everything (with consent) and let the AI surface the highlights, action items, and insights for you later. This strategy of potentially making text transcripts free (while charging for premium video/audio features) also drives virality: it encourages sharing meeting notes and AI summaries widely within and even outside your organization.

Botless Recording with Clips

We tackled the common concern: not everyone is comfortable with a bot in their meeting. In February, we pioneered a botless meeting recorder using a browser extension.

This approach captures audio clips (and even screen-shares) directly from your browser during a meeting, so there’s no visible bot participant.

We evolved this into a full feature called Clips.

By August, Clips graduated from QA into production as a lightweight way to record any conversation or demo, not just formal meetings. With one click, you can capture your screen and microphone, and our system will transcribe and analyze the recording. It’s like a stealth meeting recorder or a next-gen “Loom,” right inside xdge. Clips produces transcripts in as little as 20–30 seconds after you finish recording, and it provides AI-generated summaries and even a chat interface to ask questions about the recording.

This was a game-changer for users who want to save call notes, create walkthrough videos, or share knowledge asynchronously without scheduling a meeting.

Chat With Meetings & AI Coaching

Throughout the year, we improved the quality of meeting intelligence. Our transcription accuracy and speaker detection improved, and we enhanced the summaries to include key points, decisions made, and follow-up actions.

By July, we introduced chat with your meetings, the ability to ask our AI Assistant questions about a specific meeting transcript or recording. Instead of re-reading notes, you can query, “What did we say about Project Alpha’s deadline?” or “Who was tasked with the budget review?” and get an instant answer pulled from the meeting content. This turns your past meetings into a searchable knowledge base.

Later in the year, we launched AI Coaching, delivering role-specific feedback after meetings, for sales, engineering, leadership, HR, and Customer Success, grounded in real conversation data.

These AI Coaching prompts analyze your meeting transcripts and give you tailored feedback and tips. Unlike sales-only tools that coach reps on calls, xdge’s coaching covers multiple roles: sales teams get feedback based on proven frameworks like SPIN Selling and Challenger Sale, Customer Success managers receive tips on relationship building and retention, engineering leads get communication and delegation guidance, HR leaders see suggestions rooted in Nonviolent Communication and even Radical Candor, and executives get coaching aligned with leadership models like GROW. This all happens automatically after your meetings; it’s like having a specialized coach in every team meeting, offering pointers on what went well and what could be improved, customized to each discipline. And because it’s based on actual conversation data, the feedback is concrete and immediately applicable.

In short, xdge turned meetings into structured, usable knowledge. Instead of conversations disappearing into calendar history, xdge makes them part of your day-to-day workflows.

The Rebrand: from Ayraa to xdge

In June 2025, we introduced xdge, a new name that reflects a clearer, more ambitious direction for the platform.

xdge represents a shift toward a reasoning-enabled, agentic system built to sit at the edge of enterprise intelligence. It’s designed to connect knowledge to execution through search, meetings, and workflows, and to serve as a durable infrastructure that teams can rely on every day.

Alongside the rebrand, we launched a refreshed website focused on clarity. Core pages like Workflows, Search, and Meetings were redesigned with direct, action-oriented language, while the homepage and “Why xdge?” page clearly show how the platform’s core products work together. A late-November refresh sharpened the message further with modern visuals and positioning aligned to real customer pain points.

We also invested heavily in educational resources to support adoption. Throughout Q4, we released end-to-end product demo videos showcasing real workflows and use cases, published technical deep dives and white papers for champions and decision-makers, and added a continuously playing platform demo so visitors can see xdge in action immediately.

Through these efforts, we drastically shortened the time between signing up and seeing results. Whether it’s deploying a pre-built workflow from the Skills Library, experiencing instant search, or watching a demo tailored to a specific use case, teams can move past setup quickly and start benefiting from xdge right away. Our goal is for teams of any size, from a 5-person startup to a 5,000-person enterprise, to feel value on Day One and grow value in the days and weeks that follow.

Enterprise & Security Readiness 🔐

In 2025, we strengthened xdge’s enterprise security and compliance foundation to support broader adoption across regulated organizations.

We achieved SOC 2 Type II certification, validating that our security controls are consistently enforced over time, and completed CASA v2 compliance, ensuring our Google Workspace integrations meet strict data protection standards. We also published a formal Security Whitepaper outlining our security architecture, encryption practices, and data handling model.

These milestones ensure xdge can be deployed with confidence in enterprise environments.

Looking Ahead 🔮

In 2026, we’re shifting from “capability expansion” to outcome depth. We’ve built a powerful platform—now we’re going to take the most important use cases and go deep enough that teams get turnkey solutions they don’t need to build in-house.

Here are the outcomes we’ll be focused on next:

Turnkey, end-to-end workflows for high-value use cases
We’ll take our most-used workflows and harden them into complete solutions—fully integrated, fully automated, and ready to deploy.

• Inbox Zero, built inside-out: not just triage, but a complete system for prioritization, follow-ups, routing, and closure—so your inbox stays clean without constant manual effort.
• End-to-end sales automation: a tightly integrated motion that keeps your CRM and pipeline updated automatically. Opportunities get checked and progressed, tasks are created when deadlines approach, nudges are sent at the right time, Slack/email chatter gets captured into the opportunity record, and follow-up emails are drafted based on meeting next steps and live deal context. The goal: sales ops that runs itself, without every team assembling custom workflows.
• RFP response automation: we’ll automate the workflow of responding to RFPs—pulling the right internal context, drafting responses, and packaging outputs so teams can move faster without reinventing the process each time.

Expanding into small business and solopreneur workflows
We’re increasingly seeing demand beyond large enterprises—especially from small teams and operators who need automation even more because they don’t have staff bandwidth. We’ll build for use cases like staying on top of inbound customer messages (e.g., Facebook Messenger and SMS-style text inquiries), responding quickly, tracking follow-ups, and keeping customer communication organized across channels.

Connector scale: from dozens to thousands
Today, we support dozens of connectors. In 2026, we’re aiming to reach hundreds to thousands—potentially through partners—so xdge can plug into far more systems without waiting on bespoke integrations.

Tools: expanding the Skills Library beyond workflows and prompts
We’ll be launching Tools as a new capability that extends the Skills Library from workflows and prompts into certified, reusable action primitives—specific operations inside connectors that can be reliably called in automation.

For example, beyond searching and retrieving content from Google Docs or Microsoft Teams, tools will enable actions like:

• Creating a Google Calendar event with attendees
• Posting a message to a Microsoft Teams channel or thread

The key shift: instead of “vibe coding” connector logic on the fly via the “Replit for Work” approach, we’ll generate and ship these capabilities in advance in a certified, production-ready form. Over time, this becomes hundreds (and eventually thousands) of tools that dramatically reduce the time it takes to build and run automations.

That’s the focus for 2026: fewer DIY builds, more turnkey outcomes—powered by deeper workflows, broader connectors, and a certified tools layer that makes automation dramatically easier to deploy at scale.

Overview
This release delivers significant stability improvements across xdge, an upgraded notification system, smoother workflows, improved meeting reliability, and UI refinements across all screen sizes. We worked with key issues impacting Clips, workflows, and meeting bots, while introducing enhancements that make xdge faster, clearer, and more intuitive to use.

🔔 New Features

Enhanced Notification System

Real-Time Bell Notifications

Stay on top of activity without checking multiple places.
The new in-app notification center provides clearer, faster alerts:

• Critical system events now appear instantly in the bell icon
• Team activity notifications keep you updated on workspace actions
• Redesigned formatting makes every notification easier to read

Trial Expiration Email Alerts

Workspace admins now automatically receive an email when a 14-day trial expires, ensuring complete visibility into subscription status and next steps.

✨ Improved User Experience

Automatic Clip Access

Once a Clip recording is complete, xdge now:

• Opens a new tab automatically
• Navigates directly to the newly created Clip
  No more hunting for your recording — you’re taken straight to your content.

Upgraded Workflow Search (Slack Activity & Mentions)

We’ve added more powerful search capabilities that:

• Surface Slack user activity and mentions in seconds
• Deliver more accurate, context-rich results
• Improve reliability for daily/weekly update workflows

🐞 Bug Fixes

User Interface & Design

Screen Responsiveness

We resolved critical layout issues affecting 13” and 24” displays:

• Fixed welcome page layout and button alignment on 24-inch screens
• Corrected bloated/squeezed elements in Collections
• Standardized component spacing across breakpoints
• Fixed oversized “All Apps” text in certain views

Updated Sign-Up Screen

A refreshed look and improved copy create a smoother, more welcoming onboarding experience.

Notifications

Faster, Clearer Notification Display

• Fixed delayed delivery — notifications now appear instantly
• Corrected formatting so workflow-related details are highlighted properly

Meetings

Meeting Bot Reliability

• Fixed bots not joining scheduled Zoom meetings due to recent Zoom API changes
• Resolved “Not found” errors when opening Meetings via Slack DM links while logged out
• Ensured missing transcripts are now captured properly in Production
• Fixed scheduled Zoom meetings not appearing in the Upcoming Meetings tab

Clips

Visual & Loading Fixes

• Replaced broken thumbnail placeholders with a proper icon
• Thumbnails now display correctly even if generation fails
• Improved consistency across Clip tiles and loading states

Workflows

Execution, Errors & Reporting

• Fixed Slack workflow runtime failures
• Resolved duplicate Slack + Email notifications for workflow actions
• Reports now appear instantly without refreshing the tab
• Corrected NullPointerException errors for Slack → Jira workflows
• Added clearer error messages when required apps aren’t connected
• Deep Research reports now complete successfully

Across many IT and operations teams, there is a recurring moment at the end of each month when the momentum of day-to-day work pauses and gives way to a very different class of obligation, not a technical outage, not a crisis demanding immediate resolution, but the slow and persistent burden of assembling the month-end report. What begins as an apparently simple requirement to consolidate weekly activity logs quickly mutates into a time-consuming exercise: hunting down the correct versions of spreadsheets, reconciling columns that do not perfectly match, filtering out anomalies that quietly creep into the data, constructing pivot tables with precisely the layout leadership expects, and finally shaping the results into something polished enough to present. It is a process so deeply entrenched in organizational routine that teams no longer question its existence; they perform it because it has always been done, not because it has ever been efficient.

The tension at the heart of this ritual is the mismatch between its significance and its mechanics. Reporting is essential for understanding operational health and performance, yet most of the effort required to produce it is mechanical rather than analytical. As organizations expand, adding more departments, projects, initiatives, and layers of activity, the volume of data grows, but the reporting method remains trapped in a fixed, manually driven paradigm. The result is a system that scales in workload but not in intelligence. Teams spend more hours performing the same low-leverage tasks without producing additional insight; the friction increases, but the value does not.

You can see the structural flaw most clearly when leadership asks what seems like a harmless follow-up question: “Can we break this down by team?” or “Can we show projects separately?” or “Can we isolate incidents from project work?” What appears to be a request for a simple alternate view often requires a complete dataset rebuild, triggering a cascade of redundant labor. Manual reporting, by its nature, locks teams into a backward-looking, fragile workflow. It forces analysts to invest their highest-quality cognitive effort not in interpreting what the data means, but in assembling, cleaning, and normalizing the data just so it can be interpreted later. It fosters accuracy but discourages curiosity. It redirects technical leadership away from strategy and toward clerical cleanup.

Automation breaks this pattern not simply by accelerating the process, but by fundamentally altering its structure. When built correctly, an automated reporting pipeline replaces improvisation with determinism. It eliminates whole classes of errors that manual workflows inevitably produce, such as schema drift, inconsistent labeling, misaligned ranges, accidental omissions, and restores clarity to a process that has slowly accumulated complexity through years of minor ad-hoc adjustments. Once the mechanical layer disappears, teams are free to focus on the interpretive and strategic insights that reporting was always meant to support.

To understand the extent of this transformation, it helps to examine how a well-designed, AI-driven reporting workflow operates when engineered from first principles.

How the Automated Workflow Works

We designed the workflow as a sequence of logical, dependable stages, each one intentionally removing a category of manual effort that previously consumed hours of attention and introduced a wide range of failure modes.

1. Data Gathering from Source Files

The process begins with automated source discovery: the system searches a designated repository, such as Google Drive, for the weekly Excel files that make up the month's activity (e.g., “August Week 5 Tasks”). Instead of relying on humans to find and open the correct files, the workflow performs deterministic file matching, followed by schema validation that ensures all expected fields, Business Unit, Department, IT Group, Project Code, Priority, Status, and others, are present and correctly named. This prevents silent inconsistencies, ensuring the workflow operates only on structurally sound inputs.

2. Extraction and Consolidation

Once validated, the files are ingested into the system, which extracts the relevant fields and consolidates them into a single unified dataset. During this step, the workflow applies cleansing rules that remove empty rows, zero-value entries, malformed records, and error states. It performs type consistency checks, deduplication, and column-order normalization, so downstream analysis runs on a clean, coherent foundation. What typically requires manual copying, filtering, and reformatting is completed deterministically in seconds, producing a dataset free from structural anomalies.

3. Automated Summary Tables

With the master dataset established, the system generates several structured summaries that present the data from different analytical angles:

• By Business Unit: showing the distribution of IT effort across organizational domains like Finance, Operations, and Customer Service.
• By Department / Team: revealing the internal allocation of resources across functional groups.
• By Activity Type: breaking work into meaningful categories such as project tasks, support, maintenance, training, and compliance. A semantic normalization layer resolves inconsistent naming, treating “Project Work,” “Projects,” and “Proj Work” as the same category, ensuring accuracy even when individual contributors use different labels.
• IT Group Workload: summarizing hours by IT group and by individual, highlighting capacity load, bottlenecks, and utilization imbalances.

These tables, which typically require a full day or two of pivoting and re-pivoting, are now generated automatically and consistently.

4. Project Portfolio Analysis

The workflow then analyzes the project landscape in greater depth. It groups projects by status (Completed, In Progress, Pending), ranks the top 10 by hours invested, and calculates indicators such as completion rates or approximate progress positioning. This provides a transparent view of where the month’s project time went and whether high-effort initiatives are progressing proportionally.

5. KPI Calculations

Beyond simple rollups, the system computes key performance indicators that convert raw logs into operational signals: project-to-BAU ratios, incident percentages, category averages, team capacity utilization, and time-series variance metrics. These KPIs allow leadership to see whether the organization is moving toward more strategic work or being pulled back into reactive tasks.

6. AI-Generated Insights and Recommendations

After constructing the numerical scaffolding, the system moves into interpretation. It identifies anomalies, detects emerging patterns, and produces narrative insights that would typically require an analyst’s attention and judgment. If, for example, one business unit shows an unexpected surge in support hours, the AI highlights it and may infer likely drivers. If incident load becomes disproportionately high relative to previous months, it may suggest opportunities for automation or problem-management interventions. These insights turn the report into a strategic asset rather than a passive summary.

7. Professional Report & Delivery

Finally, the workflow compiles all tables, visualizations, KPIs, and insights into a well-structured, executive-ready document. It presents the information in a consistent, readable format, Business Unit Distribution, Activity Analysis, Top Projects, KPIs, Strategic Recommendations, and automatically saves the report to shared storage, optionally posting a digest to Slack. No manual formatting, exporting, or stitching together of multiple documents is required. Everything is produced in one automated pass.

Each of these steps runs in the background with a single trigger, requiring no supervision, no manual adjustments, and no reassembly of spreadsheets when leadership requests another slice of the data.

The Broader Impact of Automation

The broader significance of this approach is that it does more than eliminate repetitive work; it reframes how organizations understand their operational reality. When the burden of cleaning, merging, and validating data disappears, teams can focus on interpreting patterns, why one group’s workload increased, why a particular category of incidents spiked, which projects are consuming disproportionate time, where capacity is strained, and where efficiencies can be introduced. Insights that were once buried beneath hours of clerical work now surface naturally as part of the workflow.

For organizations that have relied on manual reporting for years, the improvement feels almost disproportionate to the change in process. What once demanded an entire afternoon now completes itself in the background, quietly and reliably. The immediate gains are time savings and reduced cognitive load, but the long-term impact is the return of analytical clarity: the ability to think strategically rather than mechanically.

Overview

This release introduces a fully unified notification center, major improvements to workflow reliability, faster Slack activity search powered by OpenSearch, and a cleaner trial/onboarding experience. We focused on strengthening platform stability, improving meeting and workflow performance, and elevating the UI across every touchpoint. With dozens of fixes and refinements, this release significantly streamlines how users navigate, search, collaborate, and automate their daily tasks.

🔍 New Features

14-Day Trial Expiry Notifications

You now receive clear email notifications when your 14-day trial ends. Messaging is correctly routed only to workspace owners, eliminating confusion for team members. The updated emails provide clarity on workspace status, renewal options, and next steps.

OpenSearch-Powered Slack Activity Search

Slack activity search is now dramatically faster. Fetch times have been reduced from nearly one hour to under 30 seconds, enabling more reliable workflows, usage analysis, and report generation.

Clips Permission Alerts

Clips now detects when camera or microphone permissions are blocked and surfaces an immediate warning dialog. This helps users resolve recording issues proactively and prevents silent failures during clip creation.

🛠️ Major Improvements

Unified Notification Experience

The bell icon now displays workflow reports, deep research updates, shared clips and meetings, team activity, collections alerts, and system notifications all in one place. Outdated notifications have been removed, ensuring users see only timely, relevant activity.

AI Transparency in Result Limitations

The AI now notifies users when a tool returns capped or partial results. This provides clearer expectations, reduces confusion, and improves trust in data completeness.

🐞 Bug Fixes

Meetings & Communication

• Scheduled Zoom meetings now appear consistently in the Upcoming Meetings tab
• Only one bot joins Google Meet, resolving duplicate-bot issues
• Meeting pages now use standardized spacing and font styles

Onboarding & Trial Experience

• Trial expiry emails now route to the correct recipients
• The upgrade modal’s close button behaves reliably
• Trial expiry pop-up has improved icons, spacing, and readable action buttons
• The sign-up screen now has refined spacing, colors, and imagery

Search

• Search no longer stalls after onboarding or connector setup
• Calendar and Gmail-linked search scenarios now return consistent results

Workflows

• Duplicate workflow notifications are eliminated
• Salesforce workflow code generation is now reliable
• Jira workflow reports no longer get stuck on “Fetching details”
• Workflow duplication errors have been fully resolved
• Newly created workflows and workflow reports appear instantly without refresh
• Slack workflow reports now show proper user names, not raw IDs
• Workflow scheduler time dropdown renders correctly across all screen sizes

Deep Research

• Deep research reports now complete consistently without errors
• Assistant failures in QA environments have been resolved

Clips

• Optimized thumbnails that now show a proper placeholder
• Search results use the correct icon instead of default or incorrect logos
• Clip timestamps now display in local timezone (UTC)

User Interface

• Follow-up queries are visually centered
• Email templates now have consistent branding, spacing, and readable formatting

Slack Integration

• Slack bot messages produce properly aligned bullet points

⚡ Performance Enhancements

Query & Workflow Stability

• Workflow tiles and workflow reports now appear immediately after generation
• Search and Assist performance has improved, especially after new connector onboarding
• Internal routing and display pipelines are more responsive across the UI

Monitoring Upgrades

• Additional logging and visibility have been added for search and workflow performance diagnostics

🔒 Security Updates

Access Control Improvements

• Trial-related modals and system notifications now respect updated admin-only visibility
• Authentication issues with Slack bot automations have been fixed
• Token handling across automated workflows is now more secure

🧹 Deprecations & Removals

Legacy Notification Behavior

• Outdated, older notifications are no longer shown in the bell center
• Redundant or stale system prompts have been removed for clarity

🔧 Infrastructure & Operations

System Stability Improvements

• Build stability has improved across frontend pipelines
• Logging for trial expiration, workflow creation, and Slack activity retrieval has been expanded
• Automation reliability has been strengthened for on-call jobs and background workers

Overview

We're excited to share the newest updates to your platform. This release introduces powerful real-time Web Search, expanded Google Apps write capabilities, enhanced GoLinks, and major improvements across Assist, Search, and Workflow management.

This release brings significant advancements to help your teams research faster, automate more confidently, and collaborate more effectively:

• Web Search across the entire platform
• Verified Google Calendar, Drive, and Gmail write scopes
• Streamlined workspace creation for sales and growth teams
• More reliable GoLinks and Assist responses
• Faster, more accurate Slack search and reference handling

🔍 New Features

Web Search & External Search

Real-Time Web Research Across the Platform

Bring the power of the internet into your workflows, deep research, and Assist queries, seamlessly blended with your internal workspace data.

What's included:

• Direct web search inside Assist, Deep Research, and Workflows
• Intelligent source filtering between apps, collections, files, and the web
• Proper formatted references and source attributions
• Time-based filtering for the most current results

Required Action:
Enable Web Search in your workspace settings, then use it in Assist or Workflows to expand your research beyond internal knowledge.

Integration Enhancements

Google App Verification & Write Scopes

We’ve added verified support for Google Calendar, Drive, and Gmail write scopes, unlocking deeper, more powerful automation.

Capabilities:

• Create and update calendar events
• Write to Drive folders and files
• Send emails and automate Gmail actions

Required Action:
Reconnect your Google account to enable new write-level permissions.

Workspace Management

Streamlined Workspace Creation for Sales Teams

Sales and growth teams can now spin up new prospect workspaces instantly, without requiring manual onboarding from prospects.

What’s new:

• Direct workspace creation
• Automatic Pro trial activation
• Proper admin assignments during setup

Required Action:
Use the Workspace Creation panel to set up new trial environments for prospects in just a few seconds.

🚀 Major Improvements

GoLinks Enhancements

Smoother, Smarter Link Management

GoLinks now offers a more intuitive editing and sharing experience.

Improvements:

• Edit modals pre-fill with existing link data
• Accurate sharing of feedback while creating or editing
• Clear duplicate prevention with actionable messaging
• Sharing status displays correctly throughout the process

Assist Experience

More Intelligent and Reliable Responses

Assist now responds with more accuracy and cleaner formatting.

Updates:

• App-specific filtering now correctly limits the response sources
• Eliminated JSON formatting glitches
• Restored proper formatting for reasoning steps

Search & Discovery

Faster, More Accurate Information Retrieval

Search and discovery across your workspace just became significantly more reliable.

Enhancements:

• Better semantic Slack search, including threaded conversations
• Eliminated UI overlap and animation issues
• Cleaned and consistent reference formatting for Slack messages

🐞 Bug Fixes

User Interface

• Fixed dropdown text overflow when multiple apps/collections are selected
• Resolved notification dot appearing incorrectly during time filter selection
• Corrected font inconsistencies in web search buttons
• Addressed search tab animation glitches

Data Access & Sharing

• Fixed “Access Denied” errors when chatting with shared reports
• Corrected Gmail OAuth errors inside workflows
• Resolved shared data visibility inconsistencies

When teams start working seriously with AI, familiar patterns emerge.

One prompt gets written to generate product copy. Another way to summarize feedback. Another way to rewrite a brief. Soon, these prompts are scattered, shared in Slack threads, buried in Notion, stashed in Apple Notes. People tweak them, lose track of them, and forget where the best version lives.

And as your team leans on AI more, the gap becomes obvious.

Because what you’re building isn’t just a collection of prompts. It’s a body of skills.

We call it the Skills Library because it’s about more than storing prompts; it’s about turning them into durable, reusable knowledge.

A single prompt might be a sentence in a textbox, a clever experiment, or a helpful shortcut. But when that prompt gets refined, reused, and shared across a team, it becomes something more. It becomes a repeatable process. A standard. A skill.

The Skills Library is where this evolution occurs, where prompts evolve into structured, versioned, team-ready tools. Like code or design systems, they stop being one-off inputs and become part of how your organization actually works.

Why We Built It

A good prompt doesn’t just work once; it becomes part of how your team gets work done. You refine it over time, adapt it to different inputs, and reach for it again when the same task comes up. Eventually, it becomes something worth sharing, not just because it saves time, but because it consistently outperforms a blank page.

The problem is that most teams store these in the wrong places. Note apps, Slack threads, and Notion pages are none of them built for structured reuse. They don’t version your work, they don’t support execution, and they don’t let others build on what you've already figured out.

The Skills Library changes that. It gives teams a way to manage AI logic like tangible operational assets, searchable, shareable, versioned, and ready to run. What starts as a well-crafted prompt becomes something much more: a reusable, living component in the way your team works.

What Lives in the Skills Library

AI-native work tends to organize itself around a few recurring formats.
We’ve formalized those into four types of skills:

1. Prompts
   These are lightweight, flexible inputs, often with placeholders. They’re ideal for tasks that vary slightly each time but follow a familiar structure. For example: “Write a note to [Name] summarizing [Project] for [Quarter].”
2. Workflows
   More rigid and repeatable. Workflows are designed for consistency—run the same way every time, often on a schedule. They're perfect for tasks like daily digests, weekly summaries, or standardized reports.
3. Agents
   These are predefined AI behaviors with a distinct tone, knowledge base, or style but without a specific task until runtime. Think of them as experts waiting for direction. You don’t have to tell them how to work, just what to work on.
4. Tasks
   These go beyond generating text. They perform actions—create a document, update a Jira ticket, send an email. They bridge the gap between language and execution, automating the steps that typically follow a prompt.

Templates for each of these types are available out of the box, certified by xdge AI and built for everyday use cases. But the real power comes when your team starts building its own: evolving internal knowledge into a library of living, executable tools.

From Knowledge to Interface

What’s powerful about the Skills Library isn’t just what it stores, it’s how it works.

It turns knowledge into UI.

You can search by department, tag by use case, or browse by category. You can click to preview, click to edit, and click to run. The same way IT teams manage software rules, or data teams manage queries, AI-native teams can now manage real, executable work.

• Need a prompt for summarizing churn feedback? Search → Select → Run.
• Want to schedule a workflow that generates weekly product updates? Tag it → Schedule it → Done.
• Sharing a marketing agent with a new team member? Copy the link and everything’s preconfigured.

This is how prompts become usable infrastructure. You’re not only storing ideas, you’re deploying them.

At first glance, the Skills Library might look like a utility, a more innovative way to organize prompts and workflows.

But in practice, it becomes infrastructure. It helps teams scale their use of AI not through experimentation alone, but through repeatable systems that improve over time. A good prompt doesn’t stay in one person’s notes. It becomes a skill that the entire team can use and build on.

Out of the box, the Skills Library includes templates for various everyday tasks, including research, content generation, analysis, triage, and more. However, over time, your team develops its own stack, a set of living, evolving tools that reflect the way you work.

The Knowledge Layer Teams Rely On

As AI embeds itself deeper into work, a new layer is forming, one that sits above your data and above your tools. This is the knowledge layer: the part of the system that encodes how work gets done.

The Skills Library is the interface to that layer. It turns tacit know-how into something searchable, executable, and reliable.

Most teams won’t describe it this way. They’ll just notice work moving faster. Fewer things slip through the cracks. Good ideas spread. The same task takes half the time.

That’s what it looks like when knowledge becomes a system, work stops being reinvented, and simply gets done.

Certified Skills, at Scale

The Skills Library isn’t just a tool, it’s a growing ecosystem. We’re launching with dozens of certified Skills available today, and soon there will be hundreds more. Each one is reviewed and standardized by xdge AI so teams can trust what they’re using. Whether it’s release notes, sales outreach, invoice workflows, or call summaries, Skills come pre-built, certified, and ready to apply across the org.

If you’re a knowledge worker, a manager, or really just anyone trying to keep your head above water with technology right now, you probably know this feeling. The workplace is flooded with single-purpose AI tools. They each promise to revolutionize productivity, but they don’t really connect. Instead, you end up with fragmentation, not efficiency.

That’s why we built xdge. It’s spelled X D-GE, but pronounced “edge”—as in having a competitive edge. That name reflects our mission: to unify knowledge and automation so every employee can stay productive, connected, and focused on what matters most.

Our Mission: Productivity First

From the very beginning, we designed xdge with a strong B2B focus. This isn’t consumer AI bolted onto business software—it’s purpose-built to connect employees directly to company knowledge, the tools they use every day, and the automated workflows that keep work moving.

Our vision follows four stages:

1. Search – making scattered information instantly accessible.
2. Knowledge Management – organizing it into something coherent and usable.
3. Workflows and Automation – enabling people to act intelligently with that knowledge.
4. Productivity – delivering measurable results across the organization.

A Toolkit That Works Where You Work

We designed xdge to feel less like a single tool and more like an operating system for knowledge work. That means meeting employees where they already are—not forcing them into another dashboard.

With meeting bots, clips for capturing audio, video, and screen shares, Go links for fast navigation, Slack integrations, browser extensions, curated knowledge collections, and unified search and chat across all apps, xdge captures and connects knowledge wherever it happens, in any format.

To lower the barrier even further, we created pre-packaged departmental solutions that work right out of the box: pipeline analysis for sales, inbox zero automation for email, Jira compliance for engineering, Slack recaps, and even auto-generated daily to-do lists.

Statefulness: Going Beyond Stateless Automation

At the core of xdge is the way we merge knowledge and automation. Traditional automation platforms are stateless—they run a script and forget. We built xdge workflows to be stateful. They draw from organizational knowledge before making a move.

For example, instead of sending a canned reply to an incoming email, xdge can check if the sender is on your calendar, review the project status in Jira, confirm billing history, and then craft a smarter, more relevant response. It’s a leap from rigid, one-size-fits-all automation to context-aware decision-making.

Workflows Built for Simplicity, Intelligence, and Trust

We made workflows simple enough to build in natural language—no flowcharts or coding required. To expand flexibility, we introduced Vibe Coding, which lets the AI generate tools on the fly within safe, pre-authorized boundaries. And with Safe Mode, you always see exactly what a workflow will do before it runs. That way, automation is powerful, but oversight stays firmly in human hands.

Competing Across the Landscape

We know the market is crowded. That’s why we deliberately designed xdge to close the gaps left by four categories of tools:

• Enterprise knowledge platforms like Glean or Copilot are strong in search, but we go further by pairing bots, clips, and a hybrid API + crawl retrieval approach that delivers fresher data at lower cost.
• Automation platforms like Zapier or Workato provide orchestration, but they’re stateless and developer-heavy. xdge matches that depth while making workflow creation accessible in natural language, powered by stateful knowledge.
• B2C AI assistants like ChatGPT Enterprise are generalists. We built xdge enterprise-first, with certified, auditable workflows that connect directly into business systems.
• Specialized point solutions may shine in one task but don’t talk to each other. xdge eliminates that overhead by giving you one connected platform.

Engineering for Cost Efficiency

Power only matters if it scales affordably. That’s why we engineered xdge to be 2–3x more cost-efficient than alternatives like ChatGPT Teams.

Our hybrid API + crawl retrieval indexes what matters—structured data via APIs in real time, and static content through scheduled crawls. This dramatically reduces infrastructure costs. On top of that, xdge is LLM-agnostic, dynamically choosing the right model for each step of a workflow, balancing cost, speed, and performance.

The Bigger Picture

Our pitch is simple: unify search, knowledge, workflows, and automation into one cohesive platform that’s safe, scalable, user-friendly, and cost-effective.

We built xdge to replace the fragmentation of dozens of disconnected apps with a single system that captures, organizes, automates, and delivers measurable productivity.

The question for enterprises isn’t whether fragmentation is costly—we all know it is. The question is: will you unify?