Workplace chat was designed around people sending messages to other people, with software tools sitting quietly in the background. Ando AI believes that structure becomes increasingly awkward once autonomous agents start performing meaningful parts of everyday work.
The San Francisco startup emerged from stealth on September 24 with $20 million in pre-seed and seed financing. Accel, Index Ventures, and Emergence backed the company as it develops a messaging platform built equally for humans and AI agents.
Founder Sara Du arrived at the idea while helping companies build Model Context Protocol servers during 2025. Customers repeatedly wanted their agents operating inside workplace messaging platforms, but maintaining context across conversations created significant technical problems.
Du eventually concluded that connecting increasingly capable agents to traditional communication software would not solve the deeper problem. Existing platforms generally treat agents as installed applications rather than participants that can independently follow conversations and coordinate work.
Ando AI takes the opposite approach by giving agents their own identities, inboxes, permissions, and access to shared conversations. The platform includes channels, direct messages, group discussions, and live conversations designed for both human and artificial team members.
Agents can browse conversations, identify relevant discussions, and communicate with colleagues without waiting for someone to copy information between systems. That model could reduce the growing problem of employees manually relaying outputs between AI agents and their wider teams.
The company already works with customers across software, finance, and real estate businesses spanning 15 countries. Many are currently small teams, giving Ando an early testing ground for organizations where agents handle substantial portions of daily work.
Ando AI Treats Agents Like Team Members, Not Bots
Traditional workplace bots usually respond when someone summons them, sends a command, or asks a specific question. Ando wants agents to understand enough shared context to recognize when their involvement could actually help a conversation.
An agent might notice separate channels discussing different parts of the same business problem, for example. It could bring relevant employees together, explain the overlapping context, and suggest possible next steps without waiting for direct instructions.
That ability changes workplace AI from a tool employees repeatedly consult into something closer to an active collaborator. Ando AI wants agents following organizational discussions closely enough to understand how decisions developed and when human judgment becomes necessary.
The company’s website describes its product as team coordination software built around humans and machines working inside one shared environment. Ando says agents should carry context, connect information across conversations, and help keep work progressing between different teams.
Index Ventures describes Ando’s architecture as three connected layers covering agent execution, human messaging, and organizational memory. The system manages agent permissions while structuring conversations so earlier decisions and relevant context remain easier for both humans and machines to retrieve.
The platform is also designed to remain independent of any single AI provider or agent ecosystem. Companies can bring agents they already use, including systems built around Codex, Claude, and other agent frameworks.
That flexibility could become important as businesses increasingly run different agents for research, coding, finance, customer service, and internal operations. Teams may eventually need one communication layer where those systems can coordinate rather than several disconnected agent interfaces.
Live calls form another part of the approach because important workplace decisions rarely happen only through written messages. Ando can transcribe discussions so authorized agents can review conversations and incorporate relevant information into their subsequent work.
The broader goal is to prevent valuable context from becoming trapped inside individual employees, channels, and disconnected agent sessions. An agent that understands earlier conversations could potentially complete assignments without repeatedly asking workers to explain decisions already made elsewhere.
However, giving agents greater access also creates important questions involving permissions, privacy, security, and unwanted intervention. Companies will need reliable controls determining which discussions an agent can read, which actions it can take, and when humans must approve decisions.
Ando says its agent execution layer addresses permissions alongside memory and communication behavior. How effectively those controls perform across larger organizations will become increasingly important as the startup moves beyond smaller early customers.
Ando AI Faces Slack and Teams in an Agentic Workplace
Building better technology may only be part of Ando AI’s challenge because workplace messaging is an established market. Slack and Microsoft Teams already sit deeply inside businesses where years of discussions, applications, workflows, and integrations have accumulated.
Those incumbents are also developing their own agent experiences rather than ignoring the shift toward autonomous software. Slack, for example, allows users to add agents to channels, communicate through direct messages, and manage agent sessions from its dedicated Agents and tools interface.
Microsoft has similarly been pushing Copilot and agent capabilities throughout Teams and the wider Microsoft 365 environment. That means Ando must convince businesses that rebuilding messaging around agents provides enough value to justify changing established workplace habits.
Du acknowledges that the earliest versions of the product did not immediately impress potential customers. Without experiencing how agents behaved inside longer-running team conversations, users could initially see little beyond another less-developed workplace messaging application.
Longer deployments began revealing the difference, particularly when agents connected conversations and volunteered useful context without explicit instructions. Those moments strengthened the company’s belief that autonomous coordination could become a defining feature rather than another chatbot capability.
The $20 million investment will help Ando AI expand its team and support the computing costs created by running agents continuously. Those expenses could become substantial when multiple agents monitor conversations, retrieve context, and perform work across many simultaneous projects.
The company is also entering a category attracting other startups that believe agent-heavy workplaces need different communication systems. Jack Dorsey’s Buzz, for example, has pursued a related concept with team communication involving humans and AI agents, particularly for technical users.
Ando’s advantage, if its strategy works, could come from designing the entire communication experience around agents from the beginning. Established platforms must introduce autonomous participation without disrupting products already supporting millions of traditional workplace conversations.
Still, replacing Slack or Teams remains a difficult proposition because messaging software becomes valuable partly through network effects. A new platform must persuade entire teams to move, integrate their existing tools, transfer established habits, and trust a relatively young company with sensitive conversations.
Ando therefore needs to show benefits that go considerably beyond adding an intelligent assistant to workplace chat. Its strongest argument is that agents should understand the same conversations where humans already coordinate work instead of operating through isolated windows.
The idea also fits a broader shift across enterprise software as AI agents take responsibility for more execution and research. Ema, which we covered earlier, is similarly developing agent teams that automate processes across HR, IT, and finance.
What distinguishes Ando AI is its focus on the communication layer connecting those increasingly capable digital workers with people. Du believes small teams could eventually operate at scales that previously required much larger organizations because agents can absorb substantial coordination and execution work.
Whether businesses want agents participating that actively in internal conversations remains an open question. Some employees may appreciate agents connecting information quickly, while others could find proactive automated participation distracting without careful controls.
For now, Ando’s funding gives it resources to test that idea against some of the largest names in workplace software. Its bigger bet is that tomorrow’s team chat will involve far more than humans messaging each other while AI waits for someone to ask for help.
