Pablo Mercuri
@mercurithinks
Followers
46
Following
5K
Media
11
Statuses
129
@thinklazarus CEO
San Francisco, CA
Joined March 2019
The IRONMAN suit for LLMs already exists. it's Linux. Thousands of hackers built it over 30 years. Runs 90% of the cloud. Every tool an agent needs is already there. Why are we reinventing this? MCPs could've been CLIs. Skills are just scripts and files. Give agents Linux.
1
2
63
245
2K
34K
Filesystem beats retrieval. agents are simple, stop over engineering them. using MCPs to retrieve data is just legacy SaaS forcing itself into a new world where it doesn't belong.
0
0
7
this is real, but the UI was never the moat. we paid for SaaS because data was chaos and someone else made it not our problem. but if apps become disposable weekend projects, where does the work actually live? how do agents and throwaway tools share context without it all falling
Zoho’s business would be first to be competed away by people building their own custom software built by people using @Replit @emergentlabs and @Taskade Why pay $30/seat/month for over bundled SaaS when soon even nontech ops ppl can vibe-code a custom solution in a weekend?
0
1
2
Multi-agent systems need a shared file system where work lives
1
0
1
Ephemeral apps generate fresh every time you need them, so you get a beautiful new app with no memory of what you did yesterday.
0
0
1
This is the way ! MCPs should be for abilities (tools), not data retrieval. that’s the whole thesis behind @thinklazarus and why our enterprise agents are ahead of everyone else
If you haven't seen it yet, our API shapes have evolved a lot over the past months, with tool search, file-based memory and programmatic tool use. Claude is increasingly using bash, files & code execution to get things done.
0
0
2
That's how companies should hire people! Forget leet code - show us you know how to solve real problems.
We made 5 challenges and if you score 47 points we'll offer you $500K/year + equity to join us at 🦥@UnslothAI! No experience or PhD needed. $400K - $500K/yr: Founding Engineer (47 points) $250K - $300K/yr: ML Engineer (32 points) Challenges: 1. Convert nf4 / BnB 4bit to
10
56
1K
Does your company have an agentic workforce? It should. Get one at @thinklazarus
0
0
1
The simplest upgrade: give your agents a mailbox. @thinklazarus
1
0
3
Today we're launching Lazarus, a shared filesystem where your team and agents work side by side. Drop files, connect tools, and watch as agents pass work between each other like any great team would. We just made AI collaboration as simple as sharing a folder. Come see.
0
1
11
DM for access codes
Where do your AI agents save their work? They don't. Until now. Dropbox for AI Agents. Free start → https://t.co/TevPCTok2p
0
0
0