Is agent-native team messaging worth building in 2026?

Q: What is Ando and how is it different from a Slack bot?

Ando is a team messaging platform, launched September 24, 2026 by founder Sara Du, where AI agents get their own identity and inbox instead of being installed as an app. Agents can join channels without being tagged, message people first, and sit in transcribed calls. Her framing: agents were 'treated as apps you install even as they were becoming participants in the team,' and the goal is removing the 'meat proxy,' the human who manually relays an agent's work to the rest of the company.

Q: Why doesn't Slack just add this itself?

Slack has shipped its own AI agent features, but the structural complaint isn't about a missing feature, it's that Slack was built as a closed system with a human on every end. Fivetran CEO George Fraser argued in March 2026 that Slack's own API restrictions are the real blocker: it holds the context AI agents need most and won't open it up, which is the opposite of what an agent-native workspace requires.

Q: What would actually make agent-native messaging worth building in 2026?

Not a better chat UI. The wedge is interoperability and audit: a bridge that lets agents operate natively in Ando- or Buzz-style rooms while the compliance record, the human system of record, and existing Slack integrations stay intact during a multi-year migration most companies will never fully complete. Regulated industries (Ando's own early customers are software, real estate, and finance) need that bridge more than they need a fourth chat app.

Q: Who should care about this: founders or engineers?

Both, differently. A founder building the platform itself is now competing with a free, Apache-2.0, 34,000-star incumbent backed by a public company, a brutal starting position. An in-house engineer choosing where their team's agents will live and talk is the more immediate decision: pick the room before the room picks itself by accretion of a dozen point-solution bots.

Ando launched today with $20M to give AI agents their own Slack-style inbox. Block already gave the same idea away for free in March, and it sits at 34,272 stars. The number that actually decides this isn't the raise. It's the six months a funded startup spent racing a free repo to the same finish line.

What actually shipped today

maybe worth building tracked two companies solving the identical problem six months apart, one that raised $20 million to do it and one that gave it away for free at 34,272 stars. Ando came out of stealth on September 24, 2026 with $20 million in pre-seed and seed funding from Accel, Index Ventures, and Emergence. Founder Sara Du built it as a direct Slack and Microsoft Teams competitor, one where an AI agent gets its own identity and inbox instead of living behind a bot integration. Agents can join a channel without being tagged, start a message unprompted, and sit in on a transcribed call. Du's framing of the problem: "Agents were treated as apps you install even as they were becoming participants in the team." The goal is killing what she calls the "meat proxy," the person stuck manually relaying an agent's output to the rest of the company. Ando is already working with customers in software, real estate, and finance across 15 countries, though it's tuned for small teams of two to 40 humans right now.

The company that already built this, for free

Here's the number that changes the verdict: Block, the public company Jack Dorsey runs, open-sourced Buzz in March 2026, a self-hostable workspace where humans and AI agents share the same rooms on a signed event log, and it has since collected 34,272 stars and 4,537 forks under Apache 2.0. Buzz treats "whether the author is a person or a process" as identical inside the event log, the exact pitch Ando is making with venture money. A repo with that kind of traction and a Fortune 500 company's engineering budget behind it is not a side project. Anyone building the same platform layer now has to out-execute a free competitor with a head start and no burn rate to answer to.

Why the timing isn't a coincidence

Buzz was created on March 6, 2026, the same day Fivetran CEO George Fraser published "Anthropic, please make a new Slack," a post that hit 272 points on Hacker News. Fraser's core complaint, in his own words: "Today, if I want Claude's help with something that came up in a Slack thread, I have to relay the context between Slack and Claude by copy-pasting. This is absurd. I am not a sub-agent!" His argument wasn't that Slack lacks an AI feature. It's that Slack is a closed system built for one human per seat, and its own API restrictions keep it that way even though it holds the context agents need most. That complaint went semi-viral the same week a major company shipped the fix as open source. Ando is the funded, productized version of the same idea arriving six months later.

When this is worth building

Skip the general platform. Build in one of these shapes instead:

The migration bridge. Most teams will not rip out Slack this year or next. A layer that lets agents act natively in an Ando- or Buzz-style room while the compliance trail and existing Slack integrations stay live during the transition is worth more than a fourth chat app.
The regulated-industry version. Ando's own early customers, finance and real estate, need audit logging, retention policy, and access control tuned to their compliance regime, not the generic small-team default either platform ships with today.
The interop standard. Buzz's event log and Ando's inbox model are both proprietary shapes right now. Whoever writes the adapter that lets an agent identity work across both owns a real wedge neither company is racing to close.

When it isn't

Skip it when the pitch is just "Slack, but agents are members too." That specific claim is already free, open source under the same license most of the winning open-weight and open-source AI bets use, and backed by a public company's engineering org. A closed, funded competitor making the identical claim is fighting the worst kind of battle: against a rival with more traction, more backing, and a zero-dollar price that yours can't match on day one. The same logic applies if you're weighing whether to build the underlying agent versus the room it talks in: check who already gave the room away before you fund one.

Related: Is an agent identity layer worth building in 2026? solves who's allowed in the room. This is about the room itself, which assumes identity is already settled and is the layer above it.

Frequently asked questions

Is agent-native team messaging worth building in 2026?

Not as a general Slack replacement. Ando raised $20 million to launch one on September 24, 2026, but Block's open-source Buzz has given away the same core idea since March 2026 and sits at 34,272 stars under Apache 2.0. The open wedge is the layer that lets a team keep Slack as the human system of record while giving agents a governed, auditable second channel, not a third inbox competing for the same seat.

What is Ando and how is it different from a Slack bot?

Ando, launched September 24, 2026 by founder Sara Du, is a messaging platform where AI agents get their own identity and inbox instead of being installed as an app. Agents can join channels without being tagged, message people first, and sit in transcribed calls. The goal is removing the "meat proxy," the human who manually relays an agent's work to the rest of the company.

What is Block's Buzz and why does it matter here?

Buzz is an open-source, self-hostable messaging platform built by Block where humans and AI agents share the same rooms on a signed event log. It's Apache 2.0, free, and had 34,272 stars and 4,537 forks as of September 24, 2026, more community proof than most funded seed-stage products ever collect.

Why doesn't Slack just add this itself?

Slack has shipped its own AI agent features, but Fivetran CEO George Fraser argued in March 2026 that Slack's own API restrictions are the real blocker: it holds the context AI agents need most and keeps it closed, which is the opposite of what an agent-native workspace requires.

Is this the same problem as agent identity and access control?

Related but distinct. An agent identity layer answers who is allowed to act and how a site verifies it's really an agent. Agent-native messaging answers where the conversation happens once an agent is already trusted, the room it joins, the inbox it owns.

What would actually make agent-native messaging worth building in 2026?

Interoperability and audit: a bridge that lets agents operate natively in Ando- or Buzz-style rooms while the compliance record and existing Slack integrations stay intact during a migration most companies will never fully complete.

Who should care about this: founders or engineers?

Both. A founder building the platform is competing with a free, Apache-2.0, 34,000-star incumbent backed by a public company. An in-house engineer choosing where a team's agents will live is the more immediate call, pick the room before it gets picked by accretion of a dozen point-solution bots.