MagiCrew

Great question. The short answer: the user doesn't have to decide — the main agent does.Here's how it works in practice:Super Magic has a generalist main agent that receives your request first. It handles most tasks on its own — research, writing, data analysis, code, file operations — because switching agents has a cost (context, latency, coordination overhead). So the default bias is: do it yourself unless there's a good reason not to.The main agent delegates when:A task needs a specialized capability it doesn't have — e.g., slide design, canvas generation, or interacting with a specific platform API (Lark, DingTalk, etc.). These are handled by dedicated sub-agents or skills with domain-specific tools.Independent subtasks can run in parallel — e.g., "research competitor A, B, and C" can fan out to multiple agents working simultaneously instead of sequentially, then merge results back.The user explicitly asks for it — you can @ mention a specific agent to force-route a task.The key design choice is that all agents share the same workspace filesystem. So when the main agent delegates to a sub-agent, it doesn't need to serialize the full context into a prompt — the sub-agent can just read the files. And when the sub-agent finishes, its output is already in the shared workspace for the next step. No "handoff summaries," no lost context.Think of it less like "which tasks need collaboration" and more like one person deciding when to ask a specialist vs. doing it themselves — except the specialist already has access to everything on your desk.Thanks for the kind words! 🙌

When multiple agents work together, how do you decide which tasks should stay with one agent and which ones need collaboration?

Congrats @enzy_magicrew for the launch.

I am an idiot :) . So explain to me why this tool is better than using Perplexity Computer (Other than cost)

congrats on the launch. the "shared workspace so context carries between agents" bit is the interesting part to me, most multi agent tools I've tried just re-explain everything at each handoff. since it's open source, can I bring my own LLM keys and self host, or is the open part just the agent definitions while the platform itself stays hosted? asking because running several specialized agents per workflow instead of one chat could add up fast on the LLM bill

Congrats on this launch!

Do you have plans for connecting these agents directly with tools like Slack, Notion, or Google Drive?

Congrats on the launch@enzy_magicrew!The Costco of AI analogy really lands sorting over choice hesitation is exactly what's missing in this space right now.

Curious! How this actually handles replacing between agent types without it feeling awkward, that's normally where these tools fall apart for me.

Congrats! Been burned by so many half completed AI outputs so this connects hard. Subscribing to see if it really delivers on the finish the thing promise.

honestly the thing I want most is research + deck creation in one flow, I waste so much time reformatting stuff between tools

Congrats @MagiCrew! Does each agent learn from the others' work, or are they still working in their own lane right now?

Congrats on the launch, @enzy_magicrew The costco comparison is such a different way to frame it 🙌Quick question, does memory carry across agents, or is each one siloed for now?

Hi everyone, I'm Enzy, co-founder of MagiCrew.I spent years as a journalist at the United Nations, then built and sold a company. Two very different experiences, same lesson: the bottleneck is almost never ideas. It's the gap between what people can think and what they can actually get done.My co-founders and I met at business school. We'd each built things, broken things, and paid the full price for that gap. When AI became genuinely capable, we saw something we hadn't seen before — a real chance to close it. Not for enterprises with six-figure IT budgets. For everyone.The moment that crystallized MagiCrew wasn't dramatic. It was a pattern.People weren't struggling to find AI tools. They were struggling to finish anything with them. Every tool handed you something half-done. A draft that needed reformatting. An analysis that needed to become a deck. The output was always one step away from being useful.The problem wasn't the tools. It was the missing layer above them — the one that understands what you're actually trying to accomplish, coordinates the right capabilities, and delivers something you can use.We sometimes describe MagiCrew as the Costco of AI — not because it's cheap, because it's curated.Costco doesn't carry everything. They find the best version of what you actually need, and that's all they stock. You don't comparison-shop. You trust the shelf.That's what we do with agents. We went deep on each use case — research, data, presentations, meetings, social media, creative work — and built a genuine specialist for each. Not wrappers. Not prompt templates. Agents that know your projects, remember your context, and build on everything you've already done. The longer you work inside MagiCrew, the less you have to explain.Every agent we ship has to answer one question: does this solve a genuine problem, or are we just building what's easy to demo?The bigger mission is simple: the most powerful AI capabilities in the world shouldn't require six subscriptions and a technical degree to access. We want to take the best technology available, make it genuinely useful, and put it in the hands of anyone who needs it — at a price that actually makes sense.That's AI for everyone. That's why we built MagiCrew.We'd love to hear from you — what's the one workflow you wish AI could actually handle end-to-end?