0:00
Ando just came out of stealth with $20 million to build a Slack replacement where AI agents are actual members of the team, not bots you install.
0:15
Welcome to Startup Bios, I'm Brian McCullough.
0:18
Like and subscribe to hear about the newest startups coming out of stealth and raising new rounds almost every day.
0:23
Sara Du spent a good chunk of 2025 helping companies build MCP servers, which are the connectors that let AI agents plug into other software. And she says one question kept coming up over and over.
0:36
People wanted to know how they could use their agents from inside Slack.
0:41
That sounds simple enough, but it turned out there were a bunch of technical headaches in the way, like shuttling messages back and forth, getting the agent enough context to actually be useful, and doing all of that without blowing through your token budget.
0:54
The more she dug into it, though, the more she decided the plumbing wasn't really the problem. She told TechCrunch, quote, "The deeper I went, the more I felt Slack and Teams were built for a world that was starting to pass us by. Agents were treated as apps you install even as they were becoming participants in the team."
1:12
Her answer to that is Ando, which came out of stealth on Thursday with a team messaging app built for humans and AI agents to use together.
1:22
It also came out with $20 million in funding.
1:25
Accel led the pre-seed round, and Index Ventures and Emergence Capital led the seed, with Contrary Capital in there as well.
1:33
Company did not put a valuation on any of it, and the two rounds were announced as one combined number, so we don't know exactly how that $20 million splits between them. Du calls the problem she's going after "meat proxies." Say you've got an agent that did some piece of work for you, maybe it wrote some code, or it pulled together some research, whatever.
1:53
And then you, the human, have to take that output and carry it over to your team, paste it into a channel, explain what it means, and then relay everybody's questions back to the agent.
2:03
She told TechCrunch, quote, "The human effectively becomes the messenger between the agent and the rest of the company."
2:09
And I think a lot of people watching this probably have felt that.
2:12
if you've been using AI regularly, You've got Claude or ChatGPT open on one side of your screen Slack or something else on the other, and you are the copy-paste layer in between the two.
2:23
What do wants instead is for agents to, quote, "participate in the shared conversations where work already gets coordinated."
2:30
And she kept going, quote, "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."
2:40
On the surface, Ando looks a lot like the thing you already use.
2:44
It's got channels, it's got DMs, it's got group chats.
2:47
It also has live calls, which Ando calls Jams, and those get transcribed so an agent can follow along with them too, or even take part in real time. The difference is in how the agents are treated.
2:59
In Ando an agent gets its own identity, its own inbox, and its own set of permissions, kind of the way a new hire would.
3:07
And it keeps context and memory from the conversations it's part of, at least according to the company.
3:12
Agents also get a lot of freedom in there.
3:15
They can browse the channels and decide for themselves which ones to join.
3:19
They can jump into a conversation without anybody tagging them.
3:22
And if an agent decides somebody on the team needs to know about something, it can just message that person directly, rather than sitting around waiting for someone to sign off on it.
3:30
Ando says it's agent agnostic as well, meaning you're not locked into one AI company's agents. You can bring the ones you already use, like Codex, Claude, Grokbot, and so on.
3:40
The argument is that the more agents a team piles up from different places, the harder it gets to keep them all on the same page, so one shared place for them gets more valuable the messier your setup gets.
3:51
Du is also pretty upfront about the fact that she's not trying to be a plug-in for Slack. The company is pitching Ando as a full replacement, for Slack or Teams or whatever else you're using right now.
4:02
In the launch announcement she said, quote, "This isn't about whether Slack or Teams can add agents.
4:07
Of course they can. Starting from scratch, unencumbered by legacy scaffolding, lets us design around what agents are actually capable of rather than what existing software allows them to be.
4:17
I think agents are capable of surprising us if we actually give them the permissions and the context to."
4:23
The early days did not go great, though, and Du is pretty honest about that. She told TechCrunch, quote, "A lot of people we showed it to early on were understandably unimpressed.
4:31
It was, in many respects, just a jankier messaging platform.
4:34
People would get stuck on that before they even got to what was different about the agents." Which makes sense, right?
4:40
If you're asking people to leave Slack, the very first thing they're going to judge you on is whether you're as good as Slack at being Slack.
4:47
The agent stuff is supposed to be the reason to switch, but the chat app has to clear that bar first.
4:53
What told them they were onto something, she says, was that the customers started sticking around and using it for longer and longer stretches.
4:59
And she gave TechCrunch an example of an agent doing something she didn't expect. It noticed that two conversations happening in two different channels were really about the same problem.
5:08
Nobody asked it to do anything about that.
5:10
It just pulled the people from both conversations into a group chat, laid out the context for everybody, and even suggested a decision.
5:17
Du said, quote, "That was really delightful for me.
5:19
I felt like 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." I'll note that's one example from the founder, not some measured result across all their customers.
5:32
But you can see why it would be the demo moment, because that's the kind of thing a person with 50 on red channels basically can't do.
5:41
Ando says it's got teams using it in software, real estate, and financial services. The press release says that's across a dozen countries, and Du told TechCrunch 15, so somewhere in that range.
5:51
She also acknowledged that a lot of those teams are small right now.
5:55
They haven't shared customer counts, paid seats, or revenue, or at least not yet. The one usage stat that they did give is that the Ando team itself has worked entirely inside Ando since January.
6:07
Du herself is a second-time founder.
6:09
Her first company was Alloy Automation, which does integrations and workflow automation and started out in ecommerce, and she started it as a Thiel Fellow. Index says she started coding in high school in Atlanta and left Harvard at age 19 to go do Alloy.
6:23
So she's kind of spent pretty much her whole career getting different pieces of software to talk to each other, and Ando is the kind of that same problem, except now some of the pieces are agents.
6:32
The investors are all making more or less the same argument.
6:36
Ivan Zhou, a partner at Accel, said, quote, "Agents aren't simply another feature inside existing workplace software.
6:42
They're becoming participants in companies themselves."
6:45
And he said Accel thinks that creates, quote, "a new category of coordination software, and that Ando has the opportunity to define it."
6:52
Sofia Dolfe, a partner at Index, went even bigger, quote, "The most important communication platforms have always grown into something bigger: the operating layer for how organizations work."
7:03
She pointed to what Ando calls its structured context model.
7:07
The way she describes it, Ando has a memory layer that tags messages with things like intent behind them, what stage something is at, the topic, and the surrounding context. I take that as the investor's description of the design, though, because we haven't seen any outside evidence yet that it actually makes agents work better.
7:23
And Santi Subotovsky at Emergence said, quote, "Teams need a platform that understands the context of their collaboration and helps them identify which tasks are better suited to people and which to AI."
7:36
Du told TechCrunch the new money goes to hiring more people and paying for a lot more tokens, because agents reading every channel all day isn't cheap. And her bigger vision, she told TechCrunch, is this, quote, "I think agents will let a very small teams operate at a scale that previously required hundreds of people.
7:52
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." End quote.
8:02
In my opinion, the obvious question here is the one Du already answered herself, which is can't Slack and Microsoft just do this?
8:12
And the answer is they already are, at least to a point.
8:14
Slack turned Slackbot into an AI agent back in January, and Microsoft has been putting Copilot into Teams and hooking it up to the rest of Office for a while now. Plus Jack Dorsey announced Buzz back in July, which is also a group chat for people and their agents, though that one is pitched more at developers. So Ando isn't alone in seeing this, and it's probably the smallest player in the game right now.
8:35
Where I think Du has a real point is that bolting agents onto a product built for humans is different from designing around them for day one.
8:44
In Slack, a bot is basically an app.
8:46
It mostly waits to be called, and it sees what it gets invited to.
8:50
Letting an agent wander into channels on its own and DM your coworkers is a big product decision and a big trust decision, and it's the kind of thing an incumbent with a ton of enterprise customers is probably going to be slow to turn on by default. A startup can just go ahead and do it.
9:06
But switching costs in messaging are really tough.
9:10
Your chat app isn't just a chat app, it's years of decisions and history.
9:14
And a lot of companies it's also the shared channels you've got with customers and partners. Moving all of that because the new thing has smarter agents is a big ask, especially when Du herself admits it started out feeling janky. So I think the realistic path is probably the one that Slack itself took, which is you win the new companies first.
9:34
Small, AI-native teams that are starting from zero don't have anything to migrate, and those are exactly the kinds of teams Ando says it has right now. If those companies grow up on Ando, and Ando grows with them.
9:46
you know, great. The other thing I'd keep an eye on is the very thing that makes it cool. An agent that can message you whenever it thinks you need to know something could save you a lot of time, or it could turn into one more coworker who won't stop bugging you.
10:01
Getting that balance right is probably the real product here, more than the messaging part. But I do think the meat proxy problem is something most people with an AI subscription have felt at this point, so I'll be curious to see how many teams are actually willing to move over to fix it.
10:16
That's it for this episode. Like and subscribe.
10:18
And if you'd like to get this in your ears every morning, check out the podcast, Startup Bios, wherever you get your podcasts.