Avi Hacker, J.D.

Repo: github.com/ahacker-1/cre-…

If you work in CRE acquisitions and want to see where agentic workflows are going, this is the repo to watch. The question is not "can AI write a memo?" The question is: can the whole deal team become software?

Yesterday I pushed a big revision to the open-source CRE Acquisition Orchestrator. It is no longer just a dashboard around agents. It is becoming an agentic deal-team workspace for multifamily acquisitions.

My AI assistant went down. So another AI agent on my computer diagnosed it, repaired the gateway, verified Telegram was back online, and told me Greg was back. That sounds absurd, but I think it’s a glimpse of where personal AI agents are headed. I wrote about the full setup here: open.substack.com/pub/avihacker/…

My AI chief of staff went down today. So I asked my other AI agent to fix him. It diagnosed the gateway, found the stale runtime, repaired the service, restarted OpenClaw, verified Telegram was live, and told me Greg was back online. We have officially entered the “AI agents doing IT support for other AI agents” era. You can’t make this stuff up.

@ClaudeDevs @alexalbert__ There goes my sleep

Claude Code weekly limits are increasing 50%, now through July 13. Live now for all Pro, Max, Team, and seat-based Enterprise users.

@garrytan @Alfred_Lin You have to stay nimble.

Everyone building with AI is pattern matching on what worked 6 months ago and it's already wrong. The models change, the capabilities change, and the frameworks you built last quarter break. The only way through is what you describe: hold your mental models loosely, zoom in and out constantly, and stay humble about what you think you know.

There is no merit in making something more complex than it has to be, but simplicity for simplicity’s sake will also fail us. The challenge is not to reject simplicity but to use it humbly. We should not confuse simplicity and, by extension, our reductionist thinking with reality. Here are some methods to complement simplicity: 1) Face the Limits | Simplifying frameworks has a domain of validity. Knowing where a simplification breaks down is as important as knowing when and where it applies. 2) Zoom In and Out | Each problem looks different from different angles and altitudes. By zooming in and out, we’ll likely apply different frameworks and mental models to simplify our problem. Diversity of vantage points and mental models can lead to different understandings and be illuminating. 3) Convergence | When we apply various simplifying frameworks, do our answers converge? If not, we need a deeper understanding of our problem. 4) Probabilistic Thinking | As a simplifying method, we view the world as static and stationary, yet we know the world is not static and non-stationary. It is often helpful to apply a probabilistic approach. 5) Humility | Simplifying conclusions and beliefs should be loosely held. Be careful of narratives that feel too elegant. While we struggle for clarity, the truth is often messy.

Great take - this is the way to go

@ClaudeDevs @bcherny How is this different from/loop until x

How do you keep Claude working until the job is done? Claude Code helps with this in a few ways, including one we shipped recently: /goal.

Anthropic's legal plugins are a warning shot for AI wrappers. If your product is just: prompt + public LLM + nicer UI that's not a moat. The future is skills, connectors, workflows, verification, and context. The model is table stakes. The operating layer is the product.

Also curious where Hermes is better than Claude Code or OpenClaw. What would you set up first?

For people already using Hermes: What are the best tips, tricks, and workflows I should know? Especially curious about: - memory setup - skill creation - scheduled automations - Telegram/Slack workflows

I just got my Hermes agent set up. Now I have Hermes, OpenClaw, and Claude Code all running alongside each other, helping me operate the business. This is the part of AI that still feels under-discussed: Not one chatbot. A small team of agents, each with a different job.