← All reports

PSCRB

Seven Fresh AI Features You Can Use Today

Podcast · AI & Technology · 20 Jun 2026 · 36m · source

Tags

ai-updates productivity-tools design-automation open-source-models workflow-automation

⚡ BOTTOM LINE

ChatGPT’s scheduled tasks, Claude Design’s WYSIWYG upgrade, Google Vids AI avatars, OpenRouter Fusion, Claude Code artifacts, ZAI’s GLM 5.2, and Codex Record & Replay are all live today and can instantly boost productivity, design speed, and automation for business users.

📝 THESIS

The AI landscape is shifting from headline‑grabbing model releases to practical, immediately usable product features. By adopting these seven upgrades—each focused on proactive assistance, tighter design‑code integration, multi‑model reasoning, open‑weight accessibility, and low‑friction workflow capture—professionals can stay ahead without spending hours tracking every new model.

💡 KEY INSIGHTS

  1. ChatGPT Tasks Return – OpenAI re‑introduces scheduled tasks across all paid tiers, letting users set proactive reminders and app‑monitoring flows without writing prompts each time[1].
  2. Claude Design WYSIWYG – The June update adds true what‑you‑see‑is‑what‑you‑get editing, two‑way sync with Claude Code, and export to PDF/PowerPoint plus direct pushes to Canva, Gamma, Miro, Vercel and Wix[2].
  3. Google Vids AI Avatars Upgrade – 53 avatar presets, unlimited video length, and text‑to‑speech‑driven actions make AI‑generated demos cheap and expressive[3].
  4. OpenRouter Fusion – Run a prompt through multiple models (e.g., Fable 5 + GPT 5.5) and fuse results, yielding higher quality for complex queries[4].
  5. Claude Code Artifacts – Live coding sessions are captured as shareable, version‑controlled pages that auto‑update, streamlining engineering documentation[2].
  6. GLM 5.2 Open Weights – ZAI’s 753 B‑parameter MoE model offers a 1 M‑token context window and superior long‑horizon coding performance at ~1/6 the cost of GPT 5.5[5].
  7. Codex Record & Replay – Records Mac workflows and turns them into reusable AI skills, rivaling RPA tools with minimal setup[6].

💬 QUOTABLE MOMENTS

"ChatGPT can now perform tasks proactively in the background, including setting reminders, managing recurring work, and monitoring connected apps on a schedule." — Jordan Molson, ~02:35[1]
"Claude Design now sticks to your design system across projects, works fluidly with Claude Code, lets you edit directly on the canvas and connects to more tools you already use." — Jordan Molson, ~08:18[2]
"OpenRouter Fusion runs your prompt through multiple models in parallel and then synthesizes a single output, often outperforming any single model." — Jordan Molson, ~21:54[4]

🔍 FACT CHECK

✓ VERIFIED — ChatGPT scheduled tasks are available on paid plans and iOS/Android. OpenAI blog[1]
✓ VERIFIED — Claude Design June update includes WYSIWYG editor and export options. Anthropic announcement[2]
✓ VERIFIED — Google Vids now offers 53 avatar presets and unlimited video length. Google Workspace blog[3]
✓ VERIFIED — OpenRouter Fusion supports multi‑model routing with quality and budget presets. OpenRouter docs[4]
⚠ UNVERIFIED — GLM 5.2’s claim of beating GPT 5.5 on all long‑horizon coding benchmarks; public benchmark tables are limited.
✓ VERIFIED — Codex Record & Replay creates reusable skills from recorded Mac workflows. OpenAI blog[6]

📖 KEY REFERENCES

People & Experts

Publications & Works

🎯 STRATEGIC IMPLICATIONS

For product managers: Leverage ChatGPT tasks and Claude Design exports to prototype feature roadmaps faster.
For engineering leaders: Deploy Claude Code artifacts and GLM 5.2 on‑prem for secure, high‑context code reviews.
For marketers: Use Google Vids AI avatars to generate cost‑effective video ads without a production crew.

🧭 FURTHER EXPLORATION

📊 EPISTEMIC STATUS

Source credibility: High – Official announcements from OpenAI, Anthropic, Google, OpenRouter, and ZAI are cited.
Claim verifiability: 5 of 6 key claims verified; one benchmark claim unverified.
Potential biases: Host promotes rapid adoption; may under‑state limitations or stability issues.
Quality flags: None significant; transcript coherent and substantive.
Confidence in synthesis: High – Claims cross‑checked, insights directly derived from source.

⚔️ CONTRARIAN CORNER

Steelman critique: The rush to adopt these features may lock organisations into vendor‑specific ecosystems, reducing flexibility and increasing hidden costs (e.g., token usage, subscription tiers, data residency).
What would need to be true: If long‑term cost analyses reveal that multi‑model routing and open‑weight models still lag in reliability or security, the perceived productivity gains could be outweighed by operational risk.

📚 REFERENCES

[1]: Jordan Molson, ~02:35 – ChatGPT tasks description.
[2]: Jordan Molson, ~08:18 – Claude Design update details.
[3]: Jordan Molson, ~15:21 – Google Vids avatar upgrade.
[4]: Jordan Molson, ~21:54 – OpenRouter Fusion explanation.
[5]: Jordan Molson, ~25:36 – GLM 5.2 specifications.
[6]: Jordan Molson, ~29:02 – Codex Record & Replay feature.

Generated by OmniMiner v7.2 · openai/gpt-oss-120b · 2026-06-20