Build cloud agent coworkers that take on real, multi-step work across the tools you already use. Describe the outcome you want, and Skydive creates a working agent in minutes. No code, no prompt engineering, no workflow wiring required. Skydive agents live in your stack, execute repeatable work, and get sharper over time.
Hey Product Hunt! I'm Marcus, Co-founder of Skydive 👋
The problem
Everyone has access to the same AI models now, but you still have to prompt AI tools to get good results. I've seen one too many terminals built for managing Claude Code instances.
Today's tools usually fall into two camps:
💬 Chatbots answer questions and generate content, but you still have to take action yourself. They forget what you taught them the last time you worked together.
⚡ Workflow builders automate repetitive processes, but you have to design every step, maintain the logic, and update it whenever something changes.
Skydive is designed to actually help you grow your business by autonomously taking action throughout your company.
Meet Skydive
Skydive lets you hire AI agents that take on real responsibilities across your company.
Describe the job, and your agent gets to work. Your agent has it's own codebase and computer in the cloud, but you never have to think about it. It just works. If you want it to use your skills, you just tell it to. If you want it to learn to write like you, it can do that too. It also self-improves over time by learning from previous conversations.
🖥️ Every agent has its own computer
Agents use websites, apps, and files just like a person would. They click, type, log in, create documents, and complete work from start to finish.
💬 Agents work where you work
You can talk to agents in Slack, email, iMessage, on the web, or in your terminal. They go wherever you go and feel right at home working alongside you. They even connect to your desktop and can use the same programs you use.
🤝 Built for teams
Agents are experts in their own area. They collaborate with each other, share context, and hand work off automatically so bigger projects actually get finished. Every agent can be shared with other members of your team so multiple people can coordinate work.
🌙 Automation that never sleeps
Turn recurring work into routines. Your agents monitor, take action, and keep work moving whether your laptop is open, closed, or you're halfway around the world.
🧠 Agents that improve over time
Correct an agent once, and it remembers. Your preferences, feedback, and company knowledge carry forward into future work automatically. When you're done working, your agents are dreaming and ingesting the lessons from the previous day's work.
Who is it for?
Founders, operators, and fast-moving teams that need to do more without adding headcount. If your work spans multiple tools, teammates, and recurring processes, Skydive was built for you.
We'd love your feedback ❤️
We're just getting started, and we'd love to hear what you think.
If there's one responsibility you'd hand off to an AI agent, tell us in the comments. We'll be around all day answering questions and shipping improvements.
To celebrate our Product Hunt launch, we're giving everyone an additional $5 in credits with the code PRODHUNT5 - if you need more, shoot my cofounder an email zaria[at]anything.com :)
If you're launching on product hunt soon, you can use this agent we've built to organize your launch. We used it ourselves, so we'll see how this goes, lol
@themarcuslowe Congrats on the launch. Help me understand, is it chat only or chat-first? I mean to ask, is there also a web app from where we can control the agents?
@themarcuslowe@rohanrecommends Hey Rohan! Yep, there’s a full web app where you can create, manage, and talk to your agents. You can also talk to them from Slack, email, iMessage, or the desktop app.
Report
Maker
@themarcuslowe@rohanrecommends@zariazinn The best part (imo) about agent comms is the multi-platform element: I can ping my Skydive agent in Slack about a task or project, then text them about the same thing from my phone when I'm running out the door - and then pick it up in web-app later. The persistent memory architecture for each agent that maintains across all the places I already am is outrageously useful
Report
@themarcuslowe Skipping the chat terminal to focus on persistent background execution hits a real operational need.
@rohanrecommends
Dhruv here another co-founder. We made it so you can talk to your agents everywhereee
Slack
email
text
web
desktop app
CLI
TUI
API
So just make an agent. Train it by chatting with it where you work. Then set up some routines (background work) by telling it what to do always.
When you want to give it feedback grab the closest way to talk. It's the same agent everywhere
Report
Congrats. Are we able to control permissions to agents? Example, access to production tools and company data?
@himani_sah1 Yes! You can control exactly what each agent has access to (tools, data, repos, etc). Each agent is isolated and access can be revoked at any time.
@himani_sah1 Yes! You can also set privacy level for each agent - private to yourself, internal to your company, or open to public in sales use cases.
Report
Maker
@himani_sah1 In terms of access to anything that might have exposure/sensitivity risk, I've found that my Skydive agents have a (reassuring) foundation of diligence + proactivity. if you're trying to connect to a tool outside of our oAuth integration stack, they'll guide you through the process of storing API keys/tokens in their secrets panel, will verbalize when login details (for example) shouldn't be shared over chat, and will suggest when it's safer for them to open their browser for you to plug in credentials yourself. with trust being paramount in agentic tooling, Skydive's soc2 compliance is also key here! @marcuslowe_@dhruvtruth
@himani_sah1
Yes on multiple levels:
1. Multi agent. Each agent only gets the integrations and access it needs vs everything. This helps scoping and access rejection. Agents can collaborate like humans do with different access and rules
2. Visibility: each agent can be set to internal, private, or external.
Internal: By default, all agents are internal. That means only people in your workspace can talk to the agent. If we don't know who the person is, we'll do an identity check before we let them talk to it.
Private: that means the agent can only talk to you. And only you can see it.
External: this is the most dangerous setting. But it's your choice to flip it when you feel your agent is well trained. It means that anyone can talk to your agent from anywhere where you expose them. There's many reasons you might want this (EAs, Customer Support, etc) but:
Biggest hack is usually this: you give your agent an integration and someone prompt hacks it for access. So be careful.
3. We're SOC2, CASA, Google Limited Data Use compliant, going for many other compliance certifications.
Our goal is for this to be used across the entire company, with built-in team controls, admin controls. We also have enterprise plans, which add even more capability, safety, and workspace configuration.
Let me know if we don't have something you think we need. Thanks for the feedback!
Report
“Cloud agent coworkers " is a great way to frame this. If this agents actually get better through repeated work, that could be a huge productivity boost.
@luke_bell you nailed it here - my longest running customer facing Skydive agents have immense context now stored in memory so each of their "lanes" of expertise are well honed and improving every day.
@luke_bell my best performing and go-to agent isn't even work related. he takes care of my gym workout programming so i dont have to think about it everyday. kinda like how zuckerberg wears the same shirt everyday to minimize decision making lol
@luke_bell Exactly! Each Skydive agent has unique memory and cloud computer, and will get advanced in the scope of work you give them as you prompt them more. And the best part of “cloud” is that you can talk to your agents from anywhere - iMessage, Slack, Email, or web!
Report
Maker
@luke_bell@marcuslowe_@dhruvtruth@zariazinn the self-improvement feature is wild. especially when you start to witness your agent(s) taste level + standards reach parity with yours, and ultimately supersede it. productivity is baseline. productivity x high-calibre output is Skydive's biggest moat imo - the product doesn't just amplify your productivity, it accelerates your execution ability while elevating it.
While my customer facing support agents will always be my favorite part of my workday, its the niche agents like ChargeKnight (Stripe Disputes Manager) and Alfie (Affiliate Fraud Prevention and Payments Manager) that win for me on those areas of the business I care about but don't work in daily. They build expertise off best practices (white papers on Stripe Disputes) from external sources and adopt them into our company - Chargeknight even pays for himself, he returns more cash to us than he costs to operate by a factor of 2-4x
Alfie identified spoofing in our affiliate program with pending payouts of over 18k. The vigilance of Skydive agents helping me around the clock is just unbeatable.
@amanda_o_connor1 It's an agent builder that lets you create any agent you want! Each agent has a cloud computer and persistent memory, and can run background automations to self-improve. Common use cases include personal assistants, engineers, social media managers, and more!
@amanda_o_connor1 Great question Amanda! Basically you can use Skydive to create AI agents that can connect to the tools you use like gmail, notion, github, and more. By connecting to those tools, your agents can complete real tasks for you. Things like serving as a personal assistant (managing your email & calendar), becoming an engineer (pushing prs), or designing mocks in figma. They have their own computer they can use to browse the web or host/create files. Really you name it, the agents can probably help. You can chat with them on iMessage, email, or Slack and they will have the same memory/personality across all surfaces. It's basically like hiring a coworker!
@iamanantgupta You get to decide. You can tell each agent what it owns, when it should check with you, and what requires approval. So it depends on the task and how you’ve set up your agent.
@iamanantgupta You can think of these agents as new hires! They are fully autonomous and take work to completion, but only to the degree you want them to. If there are key pieces that you would like review or transparency with, just ask! The clearer you are about what you want and need, the better the agent will be for you. Like a new hire, they get smarter and better overtime as they learn more about your standards and conventions!
@zeeshan_aslam2 I hear you! Now I don't worry about the first prompt anymore and instead ask my Skydive agents how to achieve goals themselves - and they just go make progress proactively!
Create
Hey Product Hunt! I'm Marcus, Co-founder of Skydive 👋
The problem
Everyone has access to the same AI models now, but you still have to prompt AI tools to get good results. I've seen one too many terminals built for managing Claude Code instances.
Today's tools usually fall into two camps:
💬 Chatbots answer questions and generate content, but you still have to take action yourself. They forget what you taught them the last time you worked together.
⚡ Workflow builders automate repetitive processes, but you have to design every step, maintain the logic, and update it whenever something changes.
Skydive is designed to actually help you grow your business by autonomously taking action throughout your company.
Meet Skydive
Skydive lets you hire AI agents that take on real responsibilities across your company.
Describe the job, and your agent gets to work. Your agent has it's own codebase and computer in the cloud, but you never have to think about it. It just works. If you want it to use your skills, you just tell it to. If you want it to learn to write like you, it can do that too. It also self-improves over time by learning from previous conversations.
🖥️ Every agent has its own computer
Agents use websites, apps, and files just like a person would. They click, type, log in, create documents, and complete work from start to finish.
💬 Agents work where you work
You can talk to agents in Slack, email, iMessage, on the web, or in your terminal. They go wherever you go and feel right at home working alongside you. They even connect to your desktop and can use the same programs you use.
🤝 Built for teams
Agents are experts in their own area. They collaborate with each other, share context, and hand work off automatically so bigger projects actually get finished. Every agent can be shared with other members of your team so multiple people can coordinate work.
🌙 Automation that never sleeps
Turn recurring work into routines. Your agents monitor, take action, and keep work moving whether your laptop is open, closed, or you're halfway around the world.
🧠 Agents that improve over time
Correct an agent once, and it remembers. Your preferences, feedback, and company knowledge carry forward into future work automatically. When you're done working, your agents are dreaming and ingesting the lessons from the previous day's work.
Who is it for?
Founders, operators, and fast-moving teams that need to do more without adding headcount. If your work spans multiple tools, teammates, and recurring processes, Skydive was built for you.
We'd love your feedback ❤️
We're just getting started, and we'd love to hear what you think.
If there's one responsibility you'd hand off to an AI agent, tell us in the comments. We'll be around all day answering questions and shipping improvements.
To celebrate our Product Hunt launch, we're giving everyone an additional $5 in credits with the code PRODHUNT5 - if you need more, shoot my cofounder an email zaria[at]anything.com :)
If you're launching on product hunt soon, you can use this agent we've built to organize your launch. We used it ourselves, so we'll see how this goes, lol
https://www.skydive.com/templates/product-hunt
Thanks for checking out Skydive! 🚀
@themarcuslowe Congrats on the launch. Help me understand, is it chat only or chat-first? I mean to ask, is there also a web app from where we can control the agents?
Skydive
@themarcuslowe @rohanrecommends Hey Rohan! Yep, there’s a full web app where you can create, manage, and talk to your agents. You can also talk to them from Slack, email, iMessage, or the desktop app.
@themarcuslowe @rohanrecommends @zariazinn The best part (imo) about agent comms is the multi-platform element: I can ping my Skydive agent in Slack about a task or project, then text them about the same thing from my phone when I'm running out the door - and then pick it up in web-app later. The persistent memory architecture for each agent that maintains across all the places I already am is outrageously useful
@themarcuslowe Skipping the chat terminal to focus on persistent background execution hits a real operational need.
Congrats on the launch! 🚀
Skydive
@themarcuslowe @jurgen_pergega Thanks so much Jurgen!
@themarcuslowe @jurgen_pergega we agree! Thank you Jurgen!
Anything
Congrats. Are we able to control permissions to agents? Example, access to production tools and company data?
Skydive
@himani_sah1 Yes! You can control exactly what each agent has access to (tools, data, repos, etc). Each agent is isolated and access can be revoked at any time.
Skydive
@himani_sah1 Yes! You can also set privacy level for each agent - private to yourself, internal to your company, or open to public in sales use cases.
@himani_sah1 In terms of access to anything that might have exposure/sensitivity risk, I've found that my Skydive agents have a (reassuring) foundation of diligence + proactivity. if you're trying to connect to a tool outside of our oAuth integration stack, they'll guide you through the process of storing API keys/tokens in their secrets panel, will verbalize when login details (for example) shouldn't be shared over chat, and will suggest when it's safer for them to open their browser for you to plug in credentials yourself. with trust being paramount in agentic tooling, Skydive's soc2 compliance is also key here!
@marcuslowe_ @dhruvtruth
Anything
“Cloud agent coworkers " is a great way to frame this. If this agents actually get better through repeated work, that could be a huge productivity boost.
Skydive
@luke_bell Yes, that’s one of the things we’re most excited about. They build context the more you work together.
Anything
@luke_bell you nailed it here - my longest running customer facing Skydive agents have immense context now stored in memory so each of their "lanes" of expertise are well honed and improving every day.
Skydive
@luke_bell my best performing and go-to agent isn't even work related. he takes care of my gym workout programming so i dont have to think about it everyday. kinda like how zuckerberg wears the same shirt everyday to minimize decision making lol
Skydive
@luke_bell Exactly! Each Skydive agent has unique memory and cloud computer, and will get advanced in the scope of work you give them as you prompt them more. And the best part of “cloud” is that you can talk to your agents from anywhere - iMessage, Slack, Email, or web!
@luke_bell @marcuslowe_ @dhruvtruth @zariazinn the self-improvement feature is wild. especially when you start to witness your agent(s) taste level + standards reach parity with yours, and ultimately supersede it. productivity is baseline. productivity x high-calibre output is Skydive's biggest moat imo - the product doesn't just amplify your productivity, it accelerates your execution ability while elevating it.
Anything
While my customer facing support agents will always be my favorite part of my workday, its the niche agents like ChargeKnight (Stripe Disputes Manager) and Alfie (Affiliate Fraud Prevention and Payments Manager) that win for me on those areas of the business I care about but don't work in daily. They build expertise off best practices (white papers on Stripe Disputes) from external sources and adopt them into our company - Chargeknight even pays for himself, he returns more cash to us than he costs to operate by a factor of 2-4x
Alfie identified spoofing in our affiliate program with pending payouts of over 18k. The vigilance of Skydive agents helping me around the clock is just unbeatable.
Skydive
@dylan34 Dylan you are such a pioneer user of agents, it's incredible!
Skydive
@dylan34 forever impressed by how you can create specialized agents on Skydive and your customer support crew Dylan!
Skydive
@dylan34 thats why u the goat
Congrats on your launch but I need to understand what it's all about and how it can help me
Skydive
@amanda_o_connor1 It's an agent builder that lets you create any agent you want! Each agent has a cloud computer and persistent memory, and can run background automations to self-improve. Common use cases include personal assistants, engineers, social media managers, and more!
Anything
Skydive
@amanda_o_connor1 Great question Amanda! Basically you can use Skydive to create AI agents that can connect to the tools you use like gmail, notion, github, and more. By connecting to those tools, your agents can complete real tasks for you. Things like serving as a personal assistant (managing your email & calendar), becoming an engineer (pushing prs), or designing mocks in figma. They have their own computer they can use to browse the web or host/create files. Really you name it, the agents can probably help. You can chat with them on iMessage, email, or Slack and they will have the same memory/personality across all surfaces. It's basically like hiring a coworker!
GrowMeOrganic
How does the tool decide when to rely on an agent and when to ask human input? Or that totally depends on the task?
Skydive
@iamanantgupta You get to decide. You can tell each agent what it owns, when it should check with you, and what requires approval. So it depends on the task and how you’ve set up your agent.
Skydive
@iamanantgupta You can think of these agents as new hires! They are fully autonomous and take work to completion, but only to the degree you want them to. If there are key pieces that you would like review or transparency with, just ask! The clearer you are about what you want and need, the better the agent will be for you. Like a new hire, they get smarter and better overtime as they learn more about your standards and conventions!
Anything
I like that the focus is on outcomes rather than prompt writing. Not everyone wants to spend time figuring out the perfect instructions.
Skydive
@zeeshan_aslam2 Exactly, you should be able to give it the outcome and let it figure out how to get there.
Skydive
@zeeshan_aslam2 I hear you! Now I don't worry about the first prompt anymore and instead ask my Skydive agents how to achieve goals themselves - and they just go make progress proactively!
Anything