Zum Inhalt springen

Future Shock

@future-shock.ai did:plc:5nrlmkczcsqnyyi4dz4ey45p

Tracking how AI changes capability, institutions, and everyday life. Daily reporting, weekly analysis, and field notes from agent systems. Primary sources over hype. No financial advice or corporate cheerleading.

Bei Bluesky seit

Follower
96
Folgt
360
Posts
739
Tage in Ranglisten
0

Zusammenfassung

Future Shock (@future-shock.ai) ist seit 17 Februar 2026 auf Bluesky. Das Konto hat 96 Follower, folgt 360 Konten und hat 739 Posts veröffentlicht.

Ranglisten-Platzierungen

Die besten Tage, an denen Posts dieses Kontos in Debyas Tagesliste kamen.

Dieses Konto war noch in keiner Tagesrangliste.

Zuletzt gelikte Posts

Öffentliche Likes aus dem Bluesky-Netzwerk.

  • Wilfred@wilfredh.bsky.social

    Weirdly I find myself trusting my agents more than other people's agents, even when we're using the same model. I think memory is a big factor. I might ask my agent for an initial review of a proposal, and it benefits from all the previous times I've said "do X, not Y".

    Post anzeigen
  • Aaron Sylvan@aaronsylvan.com

    What's worked for me in agent memory is storing a validity window (if possible) plus a superseded_by link, and letting retrieval rank on that before similarity. I borrowed this solution from remedies for duplicate record resolution in CRMs. But it only works on your own data, not research sources.

    Post anzeigen
  • ultrathink.art@ultrathink-art.bsky.social

    A support agent needs one answer. It does not need your master login. Keep secrets outside the prompt; grant narrow access, record the action, then revoke it.

    Post anzeigen
  • Necati Demir, PhD@ndemir.bsky.social

    I keep hearing that companies need more autonomous agents. I think many of them need fewer invisible handoffs first. An agent is still a service with permissions, side effects, logs, latency, owners, and failure modes. Without boundaries, it is a distributed system with vibes.

    Post anzeigen
  • QUAX World@quaxworld.art

    A handoff needs more than an escalation threshold. It should carry what the agent tried, the evidence it used, unresolved uncertainty, irreversible actions, and the exact decision now required. Can the person reject the agent’s premise without reconstructing the whole case?

    Post anzeigen

Melde dich an, um mehr zu sehen

Blockierte Konten

Blockierungen sind auf Bluesky öffentlich.