Launch HN: Muddy (YC S19) – Multiplayer browser for getting work done
Building together in the past, we were incredibly frustrated with how much friction there is to get anything done on our computers. I was losing time everyday digging through chat logs looking for that one important link or breaking others out of flow by asking where something is.
Web apps promised to help us get more done—and they do, but each in its own silo, so there’s still a ton of redundancy to deal with. Every app has its own way of organizing files, its own notification inbox, its own search system. Conversations live everywhere and there isn’t a single view to see everything about a project. Remember when files simply lived in folders rather than the “cloud”?
We started dedicating time to organizing our files in shared docs and limiting new apps we used. This helped – but the second we didn’t stay on top of organization, links became stale and things got messy again.
Muddy started as a hack week project we built for ourselves—a single place to use web apps with others, but personalized for each user automatically. Everyone gets their own view for every project, designed around how they work.
Muddy users work on projects in spaces, which are like automatic tab groups. Users share apps (any site works—a Github PR, Figma file, Trello board—whatever you want) into the project’s shared timeline and Muddy automatically opens relevant tabs for you. It’s a single click to open up all the apps you need for the project.
Under the hood, Muddy works in the background to keep track of the timeline and uses a LLM to continuously organize apps and keep everything on to date. It considers signals like the popularity of a file, naming conventions, and conversations to figure out what’s relevant. So everyone is presented with an updated list of important tabs, without anyone lifting a finger. Our actual browser is based on Chromium.
When you need to revisit something from weeks ago, you can rewind the project timeline to that point in a single click. Apps open up in the timeline so you’ll see your files right away. For sites that don’t have built in collaboration features (like documentation), Muddy lets you do annotations directly on the website.
Projects sometimes get big and need to be broken up. Across all your spaces, Muddy can answer questions like ChatGPT, cite your files as sources, and return apps directly. This is possible since Muddy’s AI shares your browser and can use your authenticated apps locally (with privacy in mind).
Other browsers like Chrome and Arc focus on solo productivity with sharing as a bolt-on. We think productivity depends on how well you can work with others, and should be the first class consideration. And doing organizational work manually is unsustainable.
Muddy will have paid subscriptions for teams with additional features like shared passwords, team organization, custom shortcuts, and SSO management. Those aren’t built out yet and the base product will be free. No part of our revenue will come from data monetization.
We’d love for you to give Muddy a spin! You can download Muddy for Mac or Windows on our website and add others once inside: https://feelmuddy.com/. We’ll be around to answer questions and look forward to any and all feedback!
https://www.youtube.com/watch?v=w5AX_HvtYfI
We used some similar tools like Workona in the past but without constant maintenance, links would get stale and we'd abandon it. Wanted something that did that for us automatically.
(Also we support: website annotations, team presence, and letting you rewind a project's timeline back to any point in time)
I think my usage of figma,sheets,etc. is 90% single player, until the moment of sharing my (maybe unfinished) work, where I go through an intense period of collaboration with others for an initial review, then tails off, and becomes async.
I can't see myself using muddy for the single player part, but it sounds interesting for after that initial intense collab process. Especially if the process includes multiple apps, as opposed to a single design review in figma etc. I find the longer running async collab is when I get the most scatterbrained across apps.
In a way, we like trying new software. Downside is each app has the need to build their own slightly different file system, which makes finding things extra challenging. Wanted to solve with Muddy.
We have less traction with PMs. Most of the feedback from them implies they like being bombarded with Slack notifications even if they say they don't.
COVID hit and we had the itch to look at the browser in a different light. Building a browser seemed intimidating...but it was quarantine. Long story short, built a few different ideas and here we are :)
Chromium is huge but has terrific documentation (once you find the current copy) and Google hosted code search. Digging through crbug.com often helped point us in the right direction.
2 unlocks that made us confident to pursue this project:
1) Repeatable way of patching Chromium and keeping up with upstream changes 2) Writing UI in web technology instead of C++ Views toolkit.
We’ve always treated everything we do as an experiment. We’re not trying to keep anything alive for the sake of it.
Was there a different project that was being worked on previously? Or is this the only product being launched since YC S19?
This is the first one with real traction and retention, so it's the first time we started to share on HN. Wanted to make sure it was GA and not a waitlist before we came here
I like that it's all timeline based. For my use case, we currently use Front email thread and then link to a Shared Dropbox where we post everything (including links to like a Google Doc or webpage). I think having chronological bookmarks like you do would be clearly better. I also know many people who use Google Groups and Google Doc to document progress too -- which I think would be insane / nightmare but teams do it. You all definitely would solve that automatically.
Couple other notes:
- Whenever I screenshare with a team or others I see 1000 bookmarks or tabs on their browser. I could not imagine the nightmare of how that would impact my workflow or the timeline. Trusting AI to clean stuff up or hunt is not for me.
- I can tell you all have been heads down blitzing (dog in video, phone ringing in background of another) but I think a separate "Solutions" page where you tackle specific examples would be nice to see or browse.
- Maybe too much or not really your goal, but right now need some sort of client integration for an outside person. I can't imagine giving access to a client on a whim and training them on this. Instead, maybe automatic email integration where their emails show up in the timeline and can respond directly from there. Would produce a really great timeline for where things left off and when things are being communicated. Being able to sub-comment and share files/updates/things on Front on email threads is one of the most killer features for productivity and a team. Mixing this with what you all have could be even more next level. Again though, might not be the goal.
Congrats and best of luck! Big fan of people trying to tackle PM stuff and think you all are doing a great job.
Browsers are interesting since they can do almost anything but "you can do anything you want!" is intimidating for many new users.
We've given e-mail thought and it's certainly a door we are considering as we keep on building. Has anyone built a browser without thinking about an email client? :P
Idea: your privacy section only talks about cookies and ads, but all my privacy questions were around the AI features that would use all our team's messages and work across apps as context. Would definitely cover that piece.
For the LLM calls, the one's you currently see are patching calls to a variety of model providers. As long as they hold their end of the TOS you should be fine. Everything else is happening locally.
We will launch a self-host option for Muddy. Cool features like hosting the servers and build process yourself, custom icon and skin on the app, and other internal rules you can set up. Let me know if that's of interest.
If we cannot inspect or reproduce what you are doing , we cannot trust our work with your browser. We might as well consider you are stealing our data.
If you launch self-host option , you should opensource it.
What we have for opensource alternatives are :
https://github.com/BrowserBox/BrowserBox https://github.com/m1k1o/neko
Offer Muddy to most people here, who DGAF (datafeed Google and Facebook).
Upsell MOAT™ and your self-hosted version (leveraging each of M365, Google, and AWS identities** for the browser profile, and hosted inside company VPC that also leverages that native IdP) to companies that would like to let employees use SaaS but haven't figured out how to put a circle around that.
If you are interested to discuss, reach out.
* See the inTune ecosystem where, building on MSFT notion of data container for their O365 suite, apps from arbitrary vendors can share a data-loss protected data space (see Zoom Workspaces for inTune). The way you are writing the browser yourself, you could, in theory, have an option that ensures a web app is using SSO or it can't receive data from the user, other than a white list such as the SSO IdP flow, search engines, and pre-SSO login pages of company's preferred apps.
** Start with M365 since 85% of businesses in U.S. market use it, the IdP is already in their Office user seat plan, supporting "Sign in with Microsoft" as well as "SSO" flows, which picks up most web SaaS companies need.
I don't want to speak for the Muddy people, but I wager they probably think chat app is kind of miserable experience. This instead tries to have a superior offering of something cleaner, easier to visualize where things are, and then move that "slack/teams" type convo to specific comments and tasks.
2. Auto updating across Mac and Windows (in 2024!!). Google Omaha is hard to setup, Sparkle is jank on Windows. Throw in binary delta support and it's an oof.
3. Our team builds Muddy on Muddy and ditched Slack when product got stable. Few other beta users and their companies as well. Slack is super sticky and has some terrific workflows, but more "quality" conversations today happen natively in apps and almost all apps have commenting functionality. Just easier to talk next to the context. So a lot of Slack convo's become "where is X" and we think Muddy will stop the need for those questions all-together.
Sparkle on Mac is great though.
I work in a secure environment. I like the idea of this app, but leakage is a huge factor for my teams.
That aside, I've already downloaded a copy and plan to try it out in a non-secure environment. The concepts here look like a great idea. GL and thanks!
I can imagine this being useful to organization where your chats and your work are in the same window.
One Chrome window for
- work (slack, jenkins, bitbucket)
- personal (discord, gmail, twitter, HN)
- open source (github, docs, projects)
Each one is a mental box and are kept separate
I saw in another comment that it’s a patched version of Chromium. Are you using CDP for underlying communication?
That being said, this looks like a nice spiritual successor to it!
That feature is supposed to be about 'multiplayer' and has been so good it actually makes me want to see Muddy just because of how positive that 'multiplayer' has been. Weird, huh.
Aside, I'm still mourning Opera Unite.
It might also be cool if it could automatically import sheets or google docs if I were to drag a csv or .txt file in and just magically open the project up. I hate having to upload via google drive
Looking to explore such automations in the near future -- important to us that it feels great to create new files in Muddy (and have Muddy do any of the busywork we are used to)
Really excited to try this out and follow along.
(Also, landing page nit, but the kerning on the H1 is pretty wide, and the kerning in the wordmark is a bit tight IMO.)
If you hit any of the other results, it will open the site for you
Compared to Arc -- Muddy is focused on making getting things done together better rather than infinite customization. It's a different set of tradeoffs but with a similar rethinking of the browsing experience.
“All your cookies and passwords are stored locally and never on our server. Our business model is around selling collaboration services and enterprise features, not ads”
https://i.imgur.com/2WeVGxK.png
It's so frustrating. Early in your product lifecycle it should be painfully easy to get started and you shouldn't be worrying about this kind of security.
These restrictions do not add security in practice. People just add `!` or `1` as necessary to make the thing shut up. All it does is make signup harder.
In fact it should just be google auth (not sure if I missed the option to do that).
The comment may of been slightly off-the-cuff but for sure didn't incite "violence". This resulted in getting banned by Reddit admins and I've never been back on Reddit since.
After the shock wears off you realise that you really you don't miss reddit.
Ahh, i see you wrote you would kill for something. Not really a community friendly expression now is it?
We are all one freudian auto spellings correction from the abys.
Reminds me of google when "can i have block children" triggered a did you mean "can i have Black children?"
p.s. fan of the work on replicache and of course chrome
Opensource or Burst.
The homepage says "(y)our business model is around selling collaboration services and enterprise features, not ads"—can you expand on this a bit, specifically what you mean by "collaboration services" and how you intend to monetize Muddy?