Ando 欲挑战 Slack：让人类与 AI Agent 协同工作的团队消息应用

2025 年，当 Sara Du 帮助企业构建 MCP 服务器时，人们不断问她：怎样才能直接在 Slack 内使用 AI Agent？但其中存在一系列技术障碍：需要来回传递消息、为 Agent 提供正确的上下文，同时又不能耗尽 token 预算。

在深入研究这一问题后，Du 意识到，更深层的问题源于通信平台的设计方式。Du 告诉 TechCrunch：“我研究得越深入，就越觉得 Slack 和 Teams 是为一个正在远去的世界而构建的。Agent 被视为需要安装的应用，但它们正逐渐成为团队成员。”

Du 认为，解决方案可以是一套消除“肉身代理人”（meat proxies）的系统——这是她对人们不得不将 AI Agent 的工作转述给团队这一现象的称呼。她说：“人类实际上成了 Agent 与公司其他成员之间的信使。”

Du 表示，理想的系统应当让 Agent“参与到工作已经在其中得到协调的共享对话中”。“它们应该能够理解某项决策为何作出，向同事提问，在另一名 Agent 的工作基础上继续推进，并在需要判断时引入人类。”

她的创业公司 Ando 于周四结束隐身模式，推出了一款旨在实现这一目标的应用：一个同时面向人类员工和 AI 工作者的团队消息平台。

Ando 将自己定位为 Slack 或其他消息平台的完整替代品，面向那些使用 AI Agent 进行内部沟通的企业。该应用为 Agent 提供独立的身份和收件箱，让它们能够像人类一样自然地参与对话。

Du 表示，该应用具备频道、私信、群组对话，甚至还有可被 Agent 转录和查看的实时通话。Agent 可以浏览频道、自主选择加入哪些频道，甚至无需被 @ 提及也能加入对话。如果企业的 Agent 认为某位人类员工需要获悉某件事，它可以自行向其发送消息，而不必等待批准。

为继续开发该应用，Ando 于周四宣布已完成 2,000 万美元的 pre-seed 和 seed 融资，投资方包括 Accel、Index Ventures 和 Emergence。

Du 的想法并不新鲜。工作场所协作领域的大多数主要参与者都已加入对 AI Agent 的支持——例如，Slack 已将其原生机器人升级为 AI Agent，而微软也投入大量精力，将 Copilot 整合进 Teams，并使其能够连接 Office 365 套件的其他部分。

不过，鉴于 AI 正迅速进入工作场所，市场仍为挑战既有巨头的应用留出了空间。就在几个月前，Twitter 和 Block 联合创始人 Jack Dorsey 发布了 Buzz，这款消息应用将人类和 AI Agent 汇集在一起，但其目标用户更偏向开发者。

Du 也认为，在传统巨头仍试图弄清如何调整既有软件以适应这一趋势时，构建一个 AI Agent 原生的团队消息平台仍然具备足够优势。

无论如何，她似乎明白 Ando 面临的是一场艰难的战斗，并表示公司早期进展缓慢。她解释道：“我们早期展示给很多人时，他们不感兴趣是可以理解的。它在很多方面只是一个更粗糙的消息平台。人们甚至还没看到 Agent 有何不同，就会先卡在这一点上。”

但她表示，当越来越多客户开始长时间使用 Ando 时，她的团队知道自己找到了正确方向。通过 Ando，她开始看到 Agent 能做的事情超出人们预期：一个 Agent 可以发现不同频道中正在围绕同一问题展开的两场讨论，并且无需被要求，就能将所有人拉进一个群聊、解释背景，甚至提出决策建议。

她说：“这真的让我感到很惊喜。我觉得 Agent 在管理人方面可能比人类做得更好，因为它们能在更短时间内处理远多于人类的信息。”

她表示，Ando 目前正与来自 15 个国家的软件、房地产和金融行业客户合作，不过其当前服务的许多团队规模仍较小。她称，新获得的资金将用于招聘更多员工，以及消耗更多 token。

她说：“我认为，Agent 将让非常小的团队以过去需要数百人才能达到的规模运作。它们可以承担更多执行、研究和协调工作，而人类则能将更多时间投入判断、战略，以及决定下一步应该做什么。”

When Sara Du was helping companies build MCP servers in 2025, people kept asking her how they could use AI agents from within Slack itself. But there were a bunch of technical hurdles in the way: moving messages back and forth and getting agents the right context, all without burning through token budgets.

Upon digging into the problem, Du realized that there was a deeper issue that stemmed from how communication platforms were designed. “The deeper I went, the more I felt Slack and Teams were built for a world that was starting to pass us by,” Du told TechCrunch. “Agents were treated as apps you install even as they were becoming participants in the team.”

The solution, Du thought, could be a system that eliminates “meat proxies,” as Du calls the phenomenon of people having to relay AI agents’ work to their teams. “The human effectively becomes the messenger between the agent and the rest of the company,” she said.

An ideal system would instead let agents “participate in the shared conversations where work already gets coordinated,” Du said. “They should be able to understand why a decision was made, ask a colleague a question, build on another agent’s work, and bring in a human when judgment is needed.”

Her startup, Ando, on Thursday came out of stealth with an app that sets out to do exactly that: It’s a team messaging platform designed for both human and AI workers.

Ando bills itself as a full replacement for Slack or any other messaging platform that companies using AI agents might want for internal communication. The app gives agents their own identities and inboxes and lets them partake in conversations as naturally as people can.

The app has channels, DMs, group conversations, and even live calls that can be transcribed and viewed by an agent, Du said. Agents can browse channels, pick which ones to join, and can even join conversations without being tagged. If a company’s agent thinks a human worker needs to be notified of something, it can message them on its own rather than waiting for approvals.

To continue building the app, Ando on Thursday said it has raised $20 million in pre-seed and seed funding from investors including Accel, Index Ventures, and Emergence.

Du’s idea isn’t a new one. Most of the major players in the workplace collaboration space have already built in support for AI agents — Slack, for instance, has turned its native bot into an AI agent, and Microsoft has gone to great lengths to integrate Copilot into Teams in ways that connect to the rest of its Office 365 suite.

But there is still a market for apps to challenge the incumbents given how quickly AI is invading workplaces. Just a few months ago, Twitter and Block co-founder Jack Dorsey announced Buzz, which brings together people and AI agents into one messaging app, though it’s more targeted at developers.

Du also feels there’s still enough of an advantage in building an AI-agent-native team messaging platform as legacy incumbents try to figure out how to pivot their existing software.

Regardless, she seems to understand that Ando has an uphill battle ahead of it, and said that things were slow in the beginning. “A lot of people we showed it to early on were understandably unimpressed. It was, in many respects, just a jankier messaging platform. People would get stuck on that before they even got to what was different about the agents,” she explained.

But she said her team knew they were onto something when more and more customers started using Ando for longer periods. With Ando, she started to see that agents could do more than people expect: An agent could notice two conversations being held in different channels about the same problem, and without being asked to, could bring everyone together in a group chat, explain the context, and even suggest a decision.

“That was really delightful for me,” she said, “I felt agents could better manage people than humans can because they can process a lot more messages than a human can in a shorter span of time.”

She said Ando is currently working with customers in software, real estate, and finance across 15 countries, though many of the teams it serves right now are small. The fresh capital will be used to hire more people and burn through more tokens, she said.

“I think agents will let very small teams operate at a scale that previously required hundreds of people,” she said. “They can take on more of the execution, research, and coordination work, while humans spend more of their time on judgment, strategy, and deciding what should happen next.”