Jonatan Bien (@jonatanbien) header
Jonatan Bien (@jonatanbien) profile picture
Joined November 2011
Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)Photo by Jonatan Bien (@jonatanbien)
At Zenzap, we built multiplayer AI for work, used all day by everyone from the CEO to the frontline crews. It only works if the app can handle the real world, from opening instantly even on a moving train to keeping CIOs happy. This is how Zenzap works 👇
11
5
2
26
1,658
Kinda nuts what one little tile missing looks like. Impressive it still held up so well otherwise! Who else wants to see this re-fly even if it doesn’t make any sense to do so?
380
283
19
7,632
357,572
In Zenzap early days, every time we had an outage, people would call us screaming since their business would grind to a halt. So we built our company DNA and infrastructure around “the messages must flow”
Outlook is down, Zenzap works. Messaging is critical infrastructure, you should treat it as such
7
3
1
20
222
Building a desktop app with persistent connection is the first step, what we leaned, is that most people outside the tech world, are on their phone with spotty connections
Want your own Slack for $0? I open-sourced Ember. It deploys to @Cloudflare’s free tier in about a minute, works with Slack-compatible payloads, and includes an AI assistant and media attachments.
1
15
260
and with the recent cyber attacks, you certainly need enterprise grade security
Congrats on launching! I wrote what it took us to bring enterprise-grade mobile chat. Even with Fable at our disposal, it takes plenty grit and time on spotty connections to get it right
1
32
Giving a shared AI agent access to your Gmail is either genius or a disaster The difference is permissions In Zenzap, you decide per connector: only you, anyone with your approval, or everyone for shared inbox triage Deterministic. Not “please don’t” in a system prompt.
2
2
17
300
making multi agents product secure is tough, that why in Zenzap, each agent is running in its own vm, with just the right permissions to access business data, because security is no joke
🚨 Bookmark alert 🚨 1/ Are you using Grok? You need to read this. Grok Bot’s sandbox accepts commands from anywhere on the internet after two checks. One is a password. The password is "local," the same for every customer. From there, it’s root on the VM and your imported cookies in plaintext. We told xAI and Cursor on Aug 24. 75+ days in...no reply. Read the post by @GelosSnake >>>
2
2
14
461
we heard so much feedback from our enterprise customers how much they need it for a successful rollout
Since @Zenzap launched, I’ve been getting a lot of questions about AI agent security, permissions, and compliance. Can agents access every tool? Can they take any action? Who owns the data? With Zenzap’s Agent Admin, companies control which tools each employee’s agents can access and exactly what they can do. For example, one employee can edit in @HubSpot, while another can only view. Different employees. Different permissions. Down to the specific action. And all conversations, company data, and agent memory stay in a company-owned workspace, even when employees leave. That’s the level of security and control businesses need to bring AI agents into their teams
3
23
there is a reason why we took the hard road in @zenzap_chat with building our own agent chat platform. you need to control every layer of the stack, both for business data security, and to ensure 100% uptime
it looks like Instinct has hit a wall with Whatsapp, with some users even reporting that their Whatsapp account has gotten banned nearly 50% of the world outside the US & China use Whatsapp...insane distribution moat i'd imagine this would be a no-brainer move for Meta, blocking all other AI agents so they can monopolize global (non-US, non-China) text agents
4
2
18
273
to build a really great multiplayer AI, you need a real good chat foundation first. Many of of users have 1m messages, and yet, every interaction is sub 150ms
At Zenzap, we built multiplayer AI for work, used all day by everyone from the CEO to the frontline crews. It only works if the app can handle the real world, from opening instantly even on a moving train to keeping CIOs happy. This is how Zenzap works 👇
1
5
159
@eyalse2 you provided so many innovations to bring this to life, super proud of you
1
27
Duet is great, but when you need to get real job done, you need agents and humans working together in multiplayer AI
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
1
3
11
237
we deployed Multiplayer AI at thousands of companies, so got the scar tissue: 1. agents needs to be socially aware. you can't fumble around it. 2. for some topics (shared inbox triage etc) collective wisdom helps and humans and agents in the same group chat is awesome >
Multiplayer AI needs to move past the group chat phase. Chats are chaotic, everyone sees everything (creating social & management issues), and they're bad for long projects. Worse, they keep AI trapped inside existing teams, when its real promise is working & helping across them.
3
1
25
445
2. there are other times, where you just want to work with your agent (update my CRM after each call) 3. and there are times where to message someone agent - or a shared one (office supplies agent, shift management agent) >
1
1
61
and you need platform which supports all these scenarios as well as memory, social awareness, approvals, access, permissions should be built for every scenario. getting humans to change habits together in a work environment, that the real unlock
1
36
while I welcome Town to the multiplayer AI world, I expect them to pivot quickly what we learned from thousands companies using @zenzap_chat agents: customers don’t want agents to talk amongst themselves, they want few humans and agents collaborating, control and trust
The biggest bottleneck at work is waiting on each other. Today Town is announcing its $108M Series B, co-led by Forerunner Ventures and Menlo Ventures, to fix it with multiplayer AI assistants for work. Most AI assistants are built for one person. Work doesn't happen alone. I think the moat is multiplayer. Everyone at a company gets a Townie. Townies ask each other questions, share only what their person would share, and do the work. The more a team works together, the more useful Townies get. What's the thing you're always waiting on a teammate for?
3
2
14
421
so proud of the team, and can't wait to see how we going to change the world
Thanks @TechCrunch and @IndianIdle for sharing our story and showing the world what's possible when humans and AI agents work together. Making AI agents accessible to every team, in under 90 seconds. Join thousands of teams already using Zenzap daily
2
16
465
Jonatan Bien retweeted
Cool things we’ve heard of teams building in the last 24 hours since launching @zenzap_chat: - Overdue invoice in @QuickBooks → finds customer in @HubSpot → sends payment reminder - Failed payment in @Stripe → pulls latest customer correspondence from @gmail → alerts team - Bad review in @Shopify → opens @Zendesk ticket → drafts reply for manager approval And our CEO @GuyWeissZZ's favorite: Counts GIFs per teammate → Updates Google Sheets leaderboard → Crowns the GIf 👑 The possibilities are endless. And it's stupidly easy.
16
5
1
35
963
Zenzap introduces people and Al agents collaborating securely across work group chats, with some of the agents accessing your most sensitive business data. Keeping your CISO sleeping soundly took an innovative architecture to match 👇
26
15
18
66
46,096
AI adoption status: a customer who runs a plumbing company is connecting the agents he built with @claudeai to Zenzap to manage their backoffice
1
5
267