Ando has emerged from stealth with a team messaging product built around a simple premise: AI agents should participate in workplace conversations directly, rather than operate as tools that employees invoke and translate for everyone else.
The company says it has raised $20 million across pre-seed and seed funding from Accel, Index Ventures and Emergence. It is positioning its product as a replacement for Slack and similar internal communications platforms for organizations that expect agents to handle more execution and coordination work.
From bot integrations to agent participants
Founder Sara Du began pursuing the idea after working with companies building Model Context Protocol (MCP) servers in 2025. Customers wanted to use agents inside Slack, she told TechCrunch, but had to contend with moving messages, preserving context and managing token usage.
Her conclusion was broader than an integration problem. Existing messaging systems generally treat AI as an installed app or bot. Ando instead gives agents identities and inboxes, with access to channels, direct messages, group chats and transcribed live calls.
The intended outcome is to remove what Du calls “meat proxies”: employees who must collect an agent’s output and relay it to the people making decisions. In Ando’s model, an agent can follow relevant conversations, ask colleagues questions, surface work from another agent, or notify a human when it believes intervention is needed.
That changes the role of messaging from a notification layer into a shared operating environment for people and software workers.
The operational promise—and the harder governance question
For small teams, the appeal is straightforward. An agent able to read across conversations could identify duplicate work, connect people discussing the same issue and prepare context before a decision. Du described an example where an agent brought participants from two related channel conversations into a group chat, explained the context and suggested a decision.
If that behavior is reliable, it could reduce a familiar coordination tax: searching for prior decisions, relaying status between functions and discovering too late that parallel work is underway. Ando says it is working with customers in software, real estate and finance across 15 countries, although many of its current teams are small.
But letting agents enter conversations unprompted also raises operational requirements that go beyond chat design. Companies will need clear boundaries around which channels agents can access, what information they can retain or use, when they can contact employees, and which actions require human approval. In regulated or sensitive environments, auditability and permissions will likely matter as much as an agent’s ability to summarize or coordinate.
Token economics are another practical consideration. Du said some of the new funding will support token consumption. A platform whose value comes from continuously observing conversations will need to prove that its automation gains justify its ongoing model costs.
A difficult incumbent challenge
Ando is entering a market where the incumbents are already adapting. Slack has evolved Slackbot into an AI agent, while Microsoft has integrated Copilot with Teams and the broader Microsoft 365 environment. Jack Dorsey and Block also introduced Buzz, a group-chat product for teams and agents that is oriented more toward developers.
That means Ando must persuade companies to consider replacing a deeply embedded communications system, not merely adding an AI feature. Du acknowledged that early users often saw only a less-polished messaging app before recognizing the difference in how agents could operate.
The company’s bet is that legacy platforms will be constrained by architectures designed around human users, while an agent-native product can make AI participation feel fundamental rather than bolted on.
What to watch next
The key test is whether Ando can turn autonomous participation into trusted participation. Buyers should look for granular access controls, transparent agent activity, reliable context handling and measurable reductions in coordination work. They should also ask whether the product can coexist with established communications systems during adoption.
Ando’s $20 million gives it resources to improve the platform and support model usage. Its larger challenge is behavioral: proving that teams want agents not just answering in chat, but actively helping run it.
