Verxio by Axio Lab

A self-improving AI agent that lives in Slack and learns how a company works, executes work across its tools, and automates real business tasks.

Comment

Inspiration

Most AI products are chatbots. Businesses don't need another chatbot. They need an AI employee that actually does work, remembers context, and gets better over time.

Today's best AI agents are locked behind developer tools, API keys, and complex setups that most businesses will never use.

We built Verxio to change that.

Verxio is an AI operating system for businesses that can learn how a company works and execute work across its tools. Instead of businesses buying separate AI tools for sales, finance, support, and operations, they would have one system that understands the company, coordinates specialized AI workers, and completes work end-to-end.

Verxio gives every business an AI agent that works where teams already communicate: WhatsApp, Telegram, Discord, Slack, and the web. It replies to customers, drafts reports, automates repetitive tasks, connects to over 1,000 business tools, can execute over 20,000 actions, runs scheduled workflows, and remembers everything it learns.

Unlike traditional AI assistants, Verxio improves over time. It builds reusable skills, remembers successful workflows, learns from past interactions, and becomes more valuable every day.

Businesses shouldn't have to hire developers just to use advanced AI agents. They should be able to sign up and start working with one immediately.

What it does

Every user gets a private workspace, over 1,000 tools and applications their agent can connect to, 20,000 actions their agent can execute, and the ability to interact with their agent via the web, desktop, WhatsApp, Telegram, Discord, or Slack.

Verxio can:

  1. Work across every channel. Continue the same conversation across WhatsApp, Telegram, Discord, and the web with shared context.

  2. Take real action. Use more than 400 built-in tools, browser automation, file management, scheduled jobs, and integrations with Gmail, Google Sheets, Xero, GitHub, HubSpot, Jira, Slack, and over 1,000 other applications through Composio.

  3. Learn continuously. Create new skills from experience, remember previous conversations, and improve workflows over time.

  4. Automate customer messaging with Pulse. Build visual workflows for Instagram, Messenger, and WhatsApp to respond to messages, qualify leads, tag conversations, and automatically route customers.

  5. Capture meetings with Notepad. Record meetings, generate transcripts, create AI summaries, and organize everything in searchable folders.

  6. Run scheduled work automatically. Execute recurring tasks and deliver results wherever they're needed.

  7. Keep every workspace isolated. Every customer receives dedicated memory, credentials, skills, and runtime for complete privacy and security.

Verxio is an AI employee that works inside your existing tools, connects to your business software, and becomes smarter every time you use it.

How we built it

Verxio is a hosted AI agent platform built on top of Hermes Agent, Nous Research’s open-source self-improving agent framework.

We built Verxio with Slack's MCP and Real-Time Search (RTS) API, and also the Qwen Cloud infrastructure for the agent's brain, and deployment.

Slack MCP gives the agent access to Slack as both a place to communicate and a tool it can act through. The real-time search API helps the agent discover and route the right connected-app actions across business tools.

Qwen Cloud powers the hosted model layer used for reasoning, planning, execution and deployment.

Hermes powers the agent runtime: memory, skills, tools, scheduling, delegation, messaging, and continuous improvement.

Core components:

  • verxio-api manages authentication, workspace provisioning, runtime orchestration, Pulse automations, Notepad, artifact management, Composio connections, and usage metering.

  • verxio-web provides the React interface for chat, settings, skills, automations, messaging, Pulse, Notepad, and connected apps.

  • verxio-desktop extends the web experience with native file access, terminal support, desktop bridge APIs, and system audio recording.

  • hermes-agent powers the private agent runtime: memory, skills, messaging, scheduling, delegation, tool execution, and self-improvement.

  • Slack MCP and the real-time search API connect the agent to Slack and external business tools so it can search, understand, and take action.

  • Qwen Cloud provides the hosted inference infrastructure for Verxio’s reasoning and automation layer.

Challenges we ran into

Scaling isolated runtimes

Every customer needs dedicated memory, credentials, and skills. We solved this by provisioning isolated Docker containers for every workspace and routing authenticated API and WebSocket traffic to the correct runtime.

Building one agent across many messaging platforms

Every platform has different APIs, authentication flows, and limitations. WhatsApp enforces reply windows. Instagram limits messaging capabilities. We built Pulse around a clear capability model so users always know what works on each platform.

Keeping clients synchronized

Hermes evolves rapidly. We introduced runtime version contracts that detect incompatibilities and notify users before mismatches become problems.

Managing two sources of state

Verxio stores product data while Hermes stores agent memory and learning. Separating those responsibilities cleanly required careful architecture and schema design.

Supporting every desktop platform

Recording system audio works differently across macOS, Windows, and Linux. We implemented platform-specific fallbacks to deliver a consistent experience everywhere.

Accomplishments that we're proud of

  1. Built a complete hosted platform with authentication, workspace provisioning, runtime orchestration, and production-ready infrastructure.

  2. Created one AI agent that follows users across WhatsApp, Telegram, Slack, Discord, desktop, and web without losing context.

  3. Built Pulse, a visual automation platform for Instagram, Messenger, and WhatsApp with AI-powered workflows.

  4. Launched Notepad with meeting recording, transcription, AI summaries, organization, and sharing. Connected more than 1,000 business applications through Composio.

  5. Shared one React codebase between desktop and web applications.

  6. Delivered a self-improving AI agent that genuinely becomes more capable through continued use.

What we learned

AI agents are only valuable if they're accessible.

Powerful runtimes alone are not enough. Businesses need authentication, secure infrastructure, isolated workspaces, polished interfaces, and integrations before they can trust AI in production.

Users also don't want another application to manage. They want AI inside the messaging platforms they already use every day.

Security cannot be optional. When AI has access to business tools, customer data, and credentials, every organization needs complete isolation.

Most importantly, learning changes everything. A chatbot starts every conversation from scratch. A self-improving agent builds knowledge, creates reusable skills, remembers what works, and becomes increasingly valuable over time.

What's next for Verxio

  1. Expand Pulse with TikTok, LinkedIn, broadcast campaigns, and advanced analytics.

  2. Launch shared workspaces with role-based access for teams.

  3. Introduce usage-based billing with hosted inference, credits, and spending controls.

  4. Add deeper CRM integrations and industry-specific skill packs.

  5. Complete production hardening with signed desktop applications, stronger security, and enterprise-grade infrastructure.

  6. Ship prebuilt AI employees for customer support, e-commerce, sales, operations, content creation, and other common business workflows.

Built With

Share this project:

Updates