The Problem with 'Meat Proxies' in Modern Workflows

Traditional messaging platforms like Slack and Microsoft Teams were designed for human-to-human communication. When companies integrate AI agents into these environments, they often force a human to act as a "meat proxy"—manually relaying information, context, and outputs between the AI and the rest of the team. This creates a significant bottleneck, as the human becomes a messenger rather than a strategic decision-maker.

Sara Du, founder of Ando, argues that agents should not be treated as mere apps to be installed, but as active participants in shared conversations. By forcing agents to exist outside the core communication loop, legacy platforms fail to capture the full potential of AI-driven coordination.

Agent-Native Architecture

Ando is built from the ground up to treat AI agents as first-class citizens. Key architectural differences include:

  • Agent Identity and Autonomy: Agents possess their own inboxes and identities, allowing them to participate in channels, DMs, and group conversations as naturally as human employees.
  • Contextual Awareness: Agents can monitor live calls (via transcription) and browse channels to understand the "why" behind team decisions. They can join conversations proactively without needing to be tagged.
  • Proactive Coordination: Rather than waiting for human approval, agents can identify cross-channel redundancies. For example, if an agent detects two separate channels discussing the same issue, it can merge the conversations, provide context, and suggest a resolution without human intervention.

The Strategic Shift: From Execution to Judgment

Du posits that the goal of agent-native communication is to allow small teams to operate at the scale of much larger organizations. By offloading execution, research, and coordination to agents, human roles shift toward higher-level functions: judgment, strategy, and decision-making.

While Ando faces stiff competition from incumbents like Slack and Microsoft, as well as niche competitors like Buzz, the company is betting that building an agent-native foundation provides a structural advantage over legacy platforms attempting to retrofit AI into existing architectures. The platform is currently serving teams in software, real estate, and finance, with a focus on proving that agents can manage coordination tasks more efficiently than humans due to their ability to process vast amounts of information in real-time.