Prompt Engineering

**AI Topics Discussed on 03 Feb, 2026**

—

by

Fraser

**AI Topics Discussed on 03 Feb, 2026**

Creative & Visual Media

Heather Cooper (@hbcoop_) showcased workflows for content production using generative video models. She transformed Midjourney images into Veo 3.1 videos with structured JSON prompts specifying scene descriptions, camera movements, atmospheric effects, lighting, and mood for ultra-cinematic fantasy scenes like a cosmic cathedral hall with magical lights.

Additional examples included “Nano Banana Pro” and Hailuo 2.3 generations.

fofr (@fofrAI) shared impressive generative video simulations of fluid dynamics, such as a magic glass that pours endlessly without emptying, highlighting promising behaviors for physics-based content creation.

A.I.Warper demonstrated advanced Sora prompting by replicating a detailed street interview video prompt, achieving high realism with a cameo appearance, and plans to share a tutorial on inserting custom AI characters into Sora videos.

Software Development

Guillermo Rauch (@rauchg) noted ElevenLabs’ new Skills integration, enabling AI coding assistants like Claude Code, Cursor, and OpenCode to better handle audio APIs in agent workflows via a simple npx command.

Discussions highlighted comparisons between coding agents like Codex and Claude Code, with users debating their effectiveness in refactoring and debugging tasks.

One approach shared involved using Claude Opus 4.5 for initial project planning, generating master plans and knowledge bases to accelerate building with Claude Code.

No-code tools for mobile app development were promoted, enabling professional apps with paywalls without writing code.

Ethan Mollick noted IDEs and CLIs as the next battleground, emphasizing the need for strong models and harnesses beyond chatbots.

Automation & Orchestration

Guillermo Rauch (@rauchg) emphasized the shift in engineering toward building scalable agentic systems: “The new engineering is building the agents that ‘take your job’, but now do it at 100x the scale.” He highlighted tools like CLI agents in parallel sessions (e.g., tmux), Skills and MCPs for directing behavior, and sandboxes for infinite parallelism in product development loops, including PRs and incident response.

Alex Volkov (@altryne) reacted to agent hype around Moltbot/OpenClaw demos (e.g., calendar invites, emails), comparing them to established workflows like n8n, and shared a meme underscoring familiarity for automation builders.

He also praised OpenClaw’s autonomous task completion, like creating emails and generating business ideas.

OpenRouter shared a leaderboard of top models used by OpenClaw users, revealing preferences in agentic workflows.

Agent self-setup via plain English instructions was discussed as a marketing opportunity and potential security risk, exemplified by an ElevenLabs agent.

Comprehensive workflows with Claude were outlined, integrating extended thinking, co-founder role-playing, and task-specific projects for research and MCPs.

Strategy & Ecosystem

Cristóbal Valenzuela (@c_valenzuelab) stated, “certain technologies are inevitable, and the question isn’t whether they arrive but who builds them and with what values,” amid discussions on ethical practices.

He raised alarms about potentially toxic tactics in the ecosystem, questioning why model providers aren’t addressing them.

Alex Volkov (@altryne) reflected on the explosion of models in 2025, predicting 2026 will be even crazier.

Jonathan Fischoff highlighted the value of X’s new polls with images feature for gathering preference data to train reinforcement learning models.

A market gap for AI advisors was identified—experts who can audit business operations and implement AI reductions like 40% process cuts—bridging normies and deep nerds.

SkillStack’s launch as a vetted marketplace for Claude Code skills was praised as a strong 2026 investment for buyers and sellers.

Publishing challenges in AI research were raised, where slow peer review lags behind rapid preprint advancements.

Emerging trends included world modeling’s significance in robotics, per NVIDIA’s Jim Fan.

A bold Nature commentary declared AGI achieved by Turing standards.

Concerns grew over an omnicausal anti-AI coalition pushing for halts rather than targeted policies.

SpaceX/xAI developments were noted by venturetwins.