Posted 5 single-line, question-led comments on posts in Rahil's space (Claude agents shipping, LinkedIn content quality / facts-vs-opinions, responsible public-sector AI deployment, design taste + vibe-coding, EF founder cohort energy). All different creators from prior runs. Highest-reach post this run was Charlie Hills at 755+ reactions, 320 comments, 55 reposts.
Author: Charlie Hills (2nd connection, 'I help you (actually) use AI', newsletter at charliehills.substack.com, reposted by Harshil Tomar)
Post: 'Anthropic just shipped Claude's 10 finance agents. Available in Cowork, Code, API, and Office.' Lays out 4-step install (Cowork plugin, Microsoft 365 install via Claude Code, 17 data partner connectors, run today). Lists 10 named agents: pitch-agent, meeting-prep-agent, earnings-reviewer, model-builder, market-researcher, valuation-reviewer, gl-reconciler, month-end closer, statement-auditor, kyc-screener. 755+ reactions, 320 comments, 55 reposts.
Theme: Named agent layer for finance workflows, agentic AI as delegation
Comment posted:
The shift from prompt to named agent makes adoption feel less like prompt engineering and more like delegation, do you see teams treating these as junior analysts or as smarter macros with a personality?
Author: Simran Kaur (2nd connection, 'You're the strategy | Helping coaches & founders build brands', reposted/celebrated by Alex Cinovoj)
Post: 'These days, I open LinkedIn only to see people firing facts. Calling them OpiNiOnS.' Critique of LinkedIn content quality. Defines opinion as 'where both sides can be true / can be defended / many nod yes and many don't.' Offers 2-question filter before posting: does the opposite also work, and do I have personal experience to defend my stance? 48+ reactions, 47 comments, 1 repost.
Theme: LinkedIn content quality, facts dressed as contrarian opinions
Comment posted:
Contrarian-shaped writing has quietly replaced contrarian-shaped thinking on this platform, do you think the algorithm rewarding consensus packaged as edginess is what trained creators to treat any unobjectionable claim as a hot take?
Author: Jonathan Hasak (Following, Anthropic, AI Partnerships & Adoption | Public Sector)
Post: Congratulates Code for America + Anthropic partnership for using Claude-powered tools grounded in verified federal, state, and county policy via MCP to help caseworkers serve low-income families navigating food benefits. Key framing: 'AI in service of caseworkers, not as a replacement for them. That's a model that can build trust and actually get adopted.' Reshares the Code for America announcement. 8 reactions, 1 comment.
Theme: Responsible public-sector AI, human-in-the-loop deployment
Comment posted:
AI in service of caseworkers rather than replacement is the framing that survives outside the productivity demo, do you find that the trust ceiling for public-benefits AI is set more by the policy grounding or by how visibly the human caseworker stays in the loop?