AI Weekly Malaysia

IBM Research introduces ALTK-Evolve, an agentic memory system that learns reusable guidelines from an LLM agent's own trajectories without weight updates or human labels. It shares ACE's (Agentic Context Engineering) core philosophy of never compressing learned lessons into summaries, but differs in delivery: ACE maintains one comprehensive evolving playbook while ALTK-Evolve consolidates into individually retrievable guidelines, which the authors argue reduces token consumption at inference time.

Why: If you're building LLM agents that repeatedly call APIs and fail on multi-step tasks, this directly compares two approaches to agentic memory that avoid fine-tuning. The key decision: whether to feed one large playbook (ACE) or individually retrievable guidelines (ALTK-Evolve) at inference time — and the latter claims lower token costs. Builders should evaluate whether their agent's failure patterns (mis-pagination, wrong entity resolution, returning unasked values) warrant trajectory-based learning, and which retrieval structure fits their token budget.

OpenAI published a customer story where startup Model ML uses GPT-5.6 Sol to generate editable PowerPoint and Excel files for finance workflows, reporting 21% fewer tokens per deck than Fable 5, 36% fewer tokens per Excel workbook than Opus 5, and a tearsheet build time drop from ~1 hour to 5 minutes.

Why: This is a vendor case study, not independent benchmarking. The only actionable signal is that GPT-5.6 Sol is being positioned for structured-document generation (PowerPoint/Excel) with token-efficiency claims against named competitors—founders building finance or reporting automation could test those specific token-cost claims in their own pipelines before committing.