Speaker notes for this slide
Sources & logo credits
- Attention Is All You Need — Vaswani et al. (2017)
- Language Models are Few-Shot Learners — Brown et al. (2020)
- Training language models to follow instructions with human feedback — Ouyang et al. (2022)
- Grok Bot
- Claude Duet
- Meta Muse
- OpenClaw
- ChatGPT, Grok, Qwen, DeepSeek & Doubao marks: LobeHub icon library
- Claude mark
- Gemini mark
- Qwen, DeepSeek & Doubao colour marks: LobeHub icons-static-png
- ReAct: Synergizing Reasoning and Acting in Language Models — Yao et al. (2022)
- Toolformer — Schick et al. (2023)
- OWASP Top 10 for LLM Applications: LLM01 Prompt Injection
- Meta provider mark
- OpenClaw mark
- ChatGPT
- Claude
- Gemini
- Qwen — Alibaba’s model family
- DeepSeek
- Doubao (豆包) — ByteDance
- Which always-on agent fits?
- What is a world model?
Original product descriptions checked 31 October 2026; always-on comparison checked 4 November 2026. Examples labelled as exercises are proposed demonstrations, not measured product results. Marks identify products or providers, not endorsement. Claude Duet uses the Claude mark; Meta Muse uses the Meta provider mark.