The newest teammate has an inbox
Picture a busy team chat. A designer asks why a feature changed. An engineer has the answer, but it sits three channels away. Someone must find the thread, explain the decision, and bring the right people together.
Ando wants an AI agent to help with that part.
On September 24, 2026, the startup emerged from stealth with a workplace messaging platform built for conversations among people and AI agents. It also announced $20 million in funding from Accel, Index Ventures, and Emergence Capital. Ando describes the announcement as a seed round; TechCrunch reports that the $20 million spans pre-seed and seed funding.
The pitch is easy to grasp. Most team chat apps let you summon an AI assistant. Ando wants agents to have a standing place in the team: an identity, an inbox, appropriate permissions, and access to the conversations they are allowed to follow.
That could make agents more useful. It could also make a crowded workday even louder. Ando’s challenge is to prove that its extra participants help teams move work forward, rather than create another stream of messages to survive.
Why start over with team chat?
Founder and CEO Sara Du arrived at the idea while helping companies build tools for AI agents in 2025. Teams wanted agents inside Slack, she told TechCrunch. But moving messages between systems, supplying the right context, and controlling computing costs made the setup awkward.
There was a social problem beneath the technical one. An agent could complete a useful task, yet a person still had to relay its findings to everybody else. The employee became a go-between for the software.
Du’s answer was a messaging product designed around agents as participants. In her account of Ando’s launch, agents can join channels, speak with colleagues in group messages, and pick up work without waiting to be tagged every time. Humans still get familiar features such as direct messages and calls.
Starting fresh gives Ando room to shape how context moves through a workplace. It also gives the startup a formidable problem. People already know how to use their current chat tools, and their teams have years of habits built around them. A clever agent has to earn its place in those habits—or make a compelling case for changing them.
What “agent-native” means here
“Agent-native” sounds like something a startup would put on a slide. In Ando’s case, it describes a specific product choice: an agent appears as a member of a workspace instead of only as a button inside somebody’s private assistant.
Ando says agents can take part in channels, threads, direct messages, and live conversations it calls Jams. They can build context from the conversations they are permitted to join and contribute when they have something useful to add. The company also says teams can bring agents they already use, including ones built with different AI systems. They do not have to commit to a single model provider.
Consider a product team with an agent that reviews bug reports. A person posts an issue. The agent sees it, gathers relevant history, and offers to handle a straightforward fix. Another agent might review the result. A human can then judge the change before it ships. Ando describes a version of this workflow within its own team.
That example shows the ambition. It does not prove every outside team will get the same result. Good teamwork depends on an agent knowing its job—and knowing when to leave a conversation alone.
A chat room with a memory
Every workplace has an information problem. One channel contains the decision. Another contains the reason. A third contains the person who remembers both. Finding all three can feel like an office treasure hunt, minus the treasure.
Ando argues that agents can connect those scattered pieces. On its product site, the company says conversations should create context that agents can use over time. Du offers a concrete example: when somebody raises a new feature request, one Ando agent surfaces related discussions, while another supplies missing project-management context. (The modern team messaging platform)
TechCrunch reports another possibility Du has observed: an agent notices two groups discussing the same problem, starts a shared conversation, briefs the participants, and suggests a decision. It is an appealing picture. Nobody has to discover the duplicate debate at 4:58 p.m. on a Friday.
But memory needs boundaries. A system must know which conversations an agent may use, which details belong in a new group, and when people would prefer to make the connection themselves. Ando’s central product question may be whether it can turn stored context into timely help without making every conversation feel observed by the entire office.
The $20 million bet
Ando’s funding gives it room to build. The company announced backing from Accel, Index Ventures, and Emergence Capital. Accel says it led the pre-seed round, while Index and Emergence led the seed round. The investors see an opportunity to create a shared place where people and multiple AI agents coordinate work.
Those investors have a point of view—and a financial stake. Their backing shows confidence in Ando’s approach, but it is not proof that customers will leave established platforms.
For now, the startup is testing its idea with smaller organizations. Ando says its platform is used by teams in software, real estate, and financial services across more than a dozen countries. TechCrunch reports that many of those teams remain small. Ando’s own employees have worked exclusively in the product since January, according to the company.
Du told TechCrunch the funding will help Ando hire and pay for more AI usage. Those costs matter. A chat app must be quick and dependable; an agent-driven chat app also has to pay for the work its agents do. Twenty million dollars can support a serious experiment. It cannot, by itself, make an AI teammate worth listening to.
Small teams get the first look
Anyone eager to move an entire company onto Ando may need patience. The company’s website says it is working through a waitlist and currently works best for teams of up to 30 human members, plus their agents. Du’s launch post describes an initial focus on small teams and plans to expand access to larger ones.
That is an important distinction between launching a product and making it broadly available. Ando has a live platform with early users. It is still onboarding more teams gradually. There is no basis yet to describe it as a ready-made replacement for every large organization’s messaging system.
The initial audience makes sense. A small startup can change its workflow faster than a company with thousands of employees, sprawling access rules, and chat history nobody wants to move. A compact team can also see more clearly whether an agent saves time or simply adds commentary.
Ando says it will keep a connection to Slack for teams making the transition. Its stated goal, though, is to become the place those teams work every day. That is a bigger ambition than selling an AI add-on. It asks customers to move the room where decisions happen.
Privacy has to survive the group chat
Giving agents a place in team conversations raises an immediate question: what can they read?
Ando says agents cannot read a user’s private direct messages unless that user forwards the context. The company also describes agents as having identities and permissions. Those boundaries matter because useful context and unrestricted access are very different things.
Its security page says Ando completed a SOC 2 Type I review in July 2026 and that its Type II observation period remains in progress. The page also describes encryption in transit and at rest, access controls, and other security practices. These are the company’s published statements about its program; they do not remove the need for a customer to examine how particular agents receive permissions.
That examination gets practical quickly. Can an agent repeat a confidential detail in a broader channel? Who chooses which rooms it joins? What happens when its job changes? Teams would need clear answers before letting an agent act on sensitive conversations.
Ando has made the boundaries part of its pitch. Whether they feel intuitive in daily use will matter just as much as the safeguards described on a page.
The competition is already awake
Ando’s story would be simpler if established chat platforms had ignored AI. They have not.
Slack has expanded Slackbot into an AI agent that can find information and help with tasks inside its platform. Microsoft has also integrated Copilot across Teams and its wider productivity software. TechCrunch notes that newer entrants are exploring ways for humans and agents to work in the same messaging space.
Ando is therefore making a narrower argument than “chat needs AI.” Its argument is that a platform built around agents from the start can handle shared context and participation better than a platform adding those capabilities to an established design.
Maybe it can. But a customer comparing products will care about the whole workday: message reliability, search, calls, integrations, administration, security, and how little effort migration takes. An agent that connects two conversations brilliantly will earn applause. An app that drops an ordinary notification will hear about it immediately.
Du acknowledged the early product was rough in her TechCrunch interview. Some people saw a less polished messaging app before they saw what its agents could do. That candor captures Ando’s challenge: the new idea has to be good, and the everyday chat experience has to hold up too.
Who decides when an agent speaks?
Proactive agents can save a team time. They can also become the colleague who replies to every message with a helpful three-paragraph summary. Nobody has requested that colleague.
Ando says it wants the system to respect human attention. Its site lists that as a product focus: as agents become more active, the platform should get better at deciding what deserves a person’s attention. An agent may be able to join a conversation without being tagged, according to TechCrunch. Knowing when to do so is the hard part.
A useful interruption might point out that another team has already solved the problem under discussion. A poor one might restate the thread everyone just read. Those examples illustrate the tradeoff; they are not measured results from an independent evaluation of Ando.
Organizations will also have to decide who is responsible when an agent suggests a course of action. An agent can collect context and propose a next step. People still need to judge the proposal, especially when a decision affects customers, money, or colleagues.
The dream is less busywork. The test is whether the agent can deliver that without claiming more attention than it saves.
Bring your own agents
One of Ando’s more interesting choices is its openness to agents built elsewhere. The company says teams can bring cloud or local agents into the workspace. Du’s launch post names several agent tools and says Ando also offers its own hosted option for teams that do not already have a preferred setup.
That approach fits a workplace where one agent writes code, another researches questions, and a third helps with product planning. A common conversation space could let them share relevant work with humans instead of sending everyone between separate interfaces.
It also gives Ando a harder engineering job. Different agents may have different abilities, costs, and ways of handling context. A platform that invites all of them in must help teams understand which agent can do what—and which permissions each one needs.
The company has not promised that every agent will perform equally well inside Ando. The value of an open platform depends on how smoothly those connections work in practice. Still, the idea is clear: the team chat should not force a company to pick one AI system for every task.
Pricing without a message meter
There is a small but telling detail in Du’s launch post: Ando currently prices the product per human seat. The company does not want teams counting every agent message, action, or conversation as they decide whether to use it. Du says that model could evolve as teams develop more established ways of working.
The post does not publish a dollar price, so prospective customers would need to ask Ando what a seat costs and which services that price includes. The company has offered the pricing structure, not a public price list.
The choice reveals the experience Ando hopes to encourage. If each agent contribution feels like another item on a meter, people may hesitate to let agents participate. A human-seat model aims to make collaboration feel more natural.
Of course, the computing behind an active agent still costs money. Ando must balance predictable customer pricing with the cost of the work agents perform. That is a business question the launch does not answer yet. It is also one that will become sharper if teams invite more agents into more conversations.
Can Ando become the room where work happens?
Ando has launched with funding, early users, and a focused idea: give AI agents a proper place in team communication. Its agents can participate in conversations, carry context, and help connect work across a company. The product is opening gradually, with small teams at the front of the waitlist.
The opportunity is real. So is the burden of proof. Ando must show that its agents make good contributions, respect boundaries, and leave people with fewer loose ends. It must also make the ordinary parts of messaging feel effortless. Nobody moves an entire team for a clever demo if the daily chat experience is a chore.
For now, the fairest description is that Ando is testing a different way to organize work. It has not demonstrated that Slack or Teams are obsolete. It has put a pointed question to them: if agents are going to help with real work, should they be present when the team makes its decisions?
That question will be answered in the least glamorous place imaginable: an ordinary Tuesday, halfway through a busy group chat.
Sources
- Ando’s launch announcement, September 24, 2026
- TechCrunch’s reporting and interview with Sara Du, September 24, 2026
- Sara Du’s introduction to Ando
- Ando’s product and availability information
- Ando’s security information
- Accel’s investment announcement, September 24, 2026
- Emergence Capital’s investment announcement, September 24, 2026
The Kingy Brief
Get the next Kingy Brief.
Source-checked AI changes, original tests and one practical thing to try.
Free · Choose your subjects · Double opt-in · Unsubscribe anytime