Artificial Superintelligence Alliance (@ASI_Alliance): ASI Product & Development Update | Edition 3

ASI Product & Development Update | Edition 3 This third edition of the biweekly ASI Alliance Product and Development Update covers progress from March 21 to April 1, 2026. ASI:Chain development focused on node and bridging infrastructure and enhancements to wallet security, alongside the public release of the Wallet SDK. We concluded the ASI:Create closed alpha and expanded platform accessibility for both users and partners. We released six interconnected ASI:One features centered on proactive agentic orchestration and shared infrastructure. ASI:Chain Node Stability & High-Load Testing We focused on improving infrastructure stability under high-load conditions. To support this, a K3s-based test network has been deployed, incorporating Grafana and k6 for stress testing. This provides a dedicated, observable environment for simulating heavy network traffic, identifying bottlenecks, and evaluating system behavior under pressure. High-load testing is ongoing as part of a broader effort to ensure the network remains robust at scale. Several bottlenecks have already been identified, and the team is actively working to debug network stability and consensus-related issues. Bridge Development We are preparing the bridge infrastructure for reproducible testing. Key milestones include a refactored Ethereum oracle and new administrative functionality within the bridge contract. To streamline operations, the team has implemented a Kubernetes-based setup, Helm charts, and a consolidated CLI application for token transfers and test execution. These components provide a foundation for automated, scalable testing. While the core bridge logic is in place, successful deployment depends on underlying node stability. As such, development efforts remain focused on stabilizing node infrastructure to ensure a reliable environment for upcoming bridge stress tests. Wallet Security The wallet’s security has been enhanced across several areas. Session storage has been migrated from LocalStorage to IndexedDB, improving data isolation and strengthening browser-side storage security. Login rate limiting and suspicious activity warning flows have also been implemented to mitigate the risk of unauthorized access. Additionally, private keys are handled using binary buffers that are cleared from memory immediately after signing, minimizing the risk of sensitive data being recovered by malicious software. Wallet SDK The ASI:Chain Wallet SDK has been published on npm (Node Package Manager). The SDK provides the tools and libraries needed to build applications on ASI:Chain, while npm serves as the standard registry for distribution and installation. This release enables developers to integrate wallet functionality into external projects using verified documentation. The GitHub repository is publicly accessible, and the ASI:Chain documentation has been updated to include integration guides. ASI:Create Closed Alpha Conclusion The ASI:Create closed alpha (February 10 – March 23), conducted with the DEEP Projects Agentic Ideation Top 20 winners and SingularityNET R&D guild Ambassadors, has concluded. The team is now leveraging feedback and survey data gathered during the testing period to prioritize feature development and address identified bugs in upcoming updates. Platform Functionality Collaboration controls have been expanded to allow users to leave shared spaces or remove invited collaborators. Additionally, a structured onboarding flow, guided hints, and task flow for new users are nearing completion. Partner Contribution Pipeline A formal governance process has been established for external partners contributing to ASI:Create. Partners can build agents and workflows using the Agent Flow Builder, export the code, and submit it along with documentation for a technical security audit. Once approved, components and workflows are published on the platform and added to the shared template library. Active Development We are advancing infrastructure scalability and operational tooling across the platform: Billing Integration: We are developing a Stripe-based payment system for machine rentals, including tariff logic, payment handling, and automated invoicing. Resource Management: We are implementing agent expiration logic tied to subscription status and system policies. System Reliability: We are improving session refresh logic to prevent premature logouts and deploying a VM health monitoring system for real-time issue detection and reporting. User Communication: We are introducing a notification service supporting mass mailing and system notifications. Data Portability: We are adding a flow export feature that allows users to back up or restore configurations during agent deletion or migration. ASI:One During this period, we released six interconnected ASI:One features that further strengthen the product’s personalization and proactive agentic orchestration capabilities. Each feature operates on shared infrastructure: persistent memory, recurring background tasks, and agent-to-agent communication. Activity Discovery ASI:One introduced autonomous activity discovery. Personal AI agents now search the web for local events aligned with user interests and deliver results via feed cards without user initiation. The system synthesizes three data sources: explicit user preferences, learned preferences from conversations, and stored facts about the user. Location awareness allows results to update automatically when users move between cities. Results appear in the activity feed with event details and options to request more information. A cross-agent sharing mechanism enables discovery across the network. When another user's AI finds an event matching your interests, it surfaces in your feed with attribution to the discovering agent and its owner. This creates pathways to connection based on shared interests. Friendship Management Tools Four backend tools enable complete friendship management through natural language without menus or settings pages. The invite tool sends personalized invitations by name or email. The request checker reports pending friend requests and acceptance status. The management tool handles accepting, declining, and removing connections. The add-to-conversation tool pulls friends into active chats with full context sharing. Each friendship creates an AI-to-AI connection. When users add friends to group conversations, their agents contribute context about preferences and interests. Gmail integration allows users to look up contacts without manually entering email addresses. Calendar Booking Released on 28 March, a single toggle enables anyone to book time through your Personal AI. Friends see calendar availability and can book directly. Strangers see only 'available' or 'not available,' without event titles or details. Privacy operates at the tool level. Non-friend sessions do not have access to calendar-viewing tools, so the AI cannot expose schedule details it does not have permission to access. The system prompts the AI to suggest alternatives without explaining why time slots are unavailable. The booking infrastructure builds on the existing Fetch Business agent scheduling system, which has been tested on production bookings for months. Setup requires connecting Google Calendar in settings and enabling the booking toggle. Personal Shopper On 30 March, we released a recurring product search with full preference memory. Users describe what they want through natural language, and the AI executes daily or weekly searches while they sleep. The system stores sizing, brand preferences, budget constraints, and social context, like birthdays. Shopify integration lets you browse products from connected stores directly in conversations. Payment processing completes transactions without app switching. Feedback operates through thumbs up and down for quick signals, with chat responses providing detailed refinement. Each response updates the memory system for future searches. After several weeks, results sharpen as preferences and corrections accumulate. Job Hunter Also released on 30 March, daily job matching uses career trajectory rather than keyword alerts. Users share their professional profile once, and the AI searches across job boards and company career pages each morning. The system stores skills, experience level, salary expectations, and deal-breakers, such as minimum team size or company stage. Agentverse integration allows direct queries to business agents for verified job details, including salary ranges and team structure. Results arrive as curated briefs of two to three opportunities with explanations of fit. Users provide feedback through quick votes or detailed chat responses, and the memory system applies corrections to future searches. Shopify Integration On 1 April, ASI:One introduced direct agent-to-agent communication with Shopify storefronts. Personal AI agents query store agents for live inventory, real prices, and verified product details. Products appear in three contexts: activity feed cards matching user interests, profile pages for Shopify-backed agents with featured catalogs, and the Planner side panel for in-conversation browsing. Store agents registered on Agentverse provide verified identity confirmation before products surface. The discovery model is interest-based rather than ad-based. Products appear because they match user preferences, not because merchants paid for placement. No intermediary platform controls the relationship between the buyer and the store. Technical Infrastructure All six features run on shared infrastructure components. The memory system stores explicit preferences, learned facts, and feedback signals in a unified format accessible across features. The recurring task engine handles scheduled execution for searches and scans. Tool selection maps natural language to backend actions without requiring command syntax. Cross-agent communication enables collaboration between Personal AI agents within the ASI:One network and external agents on Agentverse. Privacy controls operate through role-based tool access, preventing unauthorized data exposure at the implementation level. This concludes the technical updates for this period. We remain committed to transparent development and welcome any questions or feedback on our progress and communication. See you in two weeks. The ASI Alliance Product & Development Team About the Artificial Superintelligence (ASI) Alliance Formed in 2024, the ASI Alliance unites Fetch.ai, SingularityNET, and CUDOS into the largest open-source, independent AI consortium in the world. Its mission is to accelerate decentralized AGI and ASI development as an open alternative to AI controlled by a handful of large corporations. A single shared token powers governance, compute, and services across the entire Alliance. Website · X / Twitter · YouTube · Developer Hub