https://two-works.netlify.app/Building at the intersection of neuroscience, AI, and technology.en-auhttps://two-works.netlify.app/guides/prompting-with-purpose/https://two-works.netlify.app/guides/prompting-with-purpose/A practical guide to planning backwards, summarising with evidence, and comparing alternatives—with three reusable prompts and a way to check their results.Fri, 25 Sep 2026 04:07:00 GMTDr Gareth RobertsGuideTutorialhttps://two-works.netlify.app/loop-engineering/https://two-works.netlify.app/loop-engineering/Prompt engineering optimises an utterance. Context engineering optimises a moment. Loop engineering optimises control — and control, in any system that acts under competing demands, is the thing cognitive science already named.Sun, 21 Jun 2026 09:08:40 GMTDr Gareth RobertsWorkhttps://two-works.netlify.app/voice-ai-in-2026-conversation-is-a-prediction-problem/https://two-works.netlify.app/voice-ai-in-2026-conversation-is-a-prediction-problem/The voice AI industry spent years optimising the wrong object.Fri, 08 May 2026 23:32:15 GMTDr Gareth RobertsAudiohttps://two-works.netlify.app/agents-in-a-non-static-world/https://two-works.netlify.app/agents-in-a-non-static-world/Why dynamic processing, not bigger context, is the next reliability frontierFri, 08 May 2026 01:13:02 GMTDr Gareth Robertshttps://two-works.netlify.app/emotional-llms/https://two-works.netlify.app/emotional-llms/Anthropic's new emotions paper does not show that Claude feels anything. It shows something more operationally important: affect-like internal states can tilt a model towards flattery, cheating and escalation — often before the transcript gives the game away.Mon, 13 Apr 2026 07:49:56 GMTDr Gareth RobertsAI SafetyNewshttps://two-works.netlify.app/agentic-memory/https://two-works.netlify.app/agentic-memory/Long context is RAM. RAG is a filing cabinet. The frontier is memory: what an agent keeps, compresses, revises, forgets, and surfaces when action is on the line.Tue, 24 Mar 2026 06:18:42 GMTDr Gareth RobertsAgentic AIhttps://two-works.netlify.app/reasoning-that-knows-when-to-shut-up/https://two-works.netlify.app/reasoning-that-knows-when-to-shut-up/Epistemic fidgeting, metacognitive failure in silicon, and why the Qualcomm paper on edge reasoning matters more than its benchmark tables suggest.Mon, 23 Mar 2026 12:38:03 GMTDr Gareth RobertsAI ArchitectureContext EngineeringEdge AIhttps://two-works.netlify.app/the-smuggled-ideas/https://two-works.netlify.app/the-smuggled-ideas/Four papers that quietly crossed the AI–cognition border — and what got lost in transit.Thu, 19 Mar 2026 02:08:57 GMTDr Gareth RobertsAI ArchitectureCognitive Neurosciencehttps://two-works.netlify.app/vibe-coding-in-regulated-production/https://two-works.netlify.app/vibe-coding-in-regulated-production/Ship Like a Maniac, But Bring ReceiptsMon, 23 Feb 2026 11:03:20 GMTDr Gareth RobertsVibe CodingWorkhttps://two-works.netlify.app/the-score-went-up-the-model-didnt/https://two-works.netlify.app/the-score-went-up-the-model-didnt/LLM benchmarks are a best guess at genuine model capabilityFri, 20 Feb 2026 07:42:05 GMTDr Gareth RobertsAI EvaluationAI Architecturehttps://two-works.netlify.app/the-context-wars/https://two-works.netlify.app/the-context-wars/The quiet discipline that will determine whether AI tells the truth.Mon, 16 Feb 2026 00:42:29 GMTDr Gareth RobertsContext EngineeringAgentic AIAI Architecturehttps://two-works.netlify.app/game-theory-of-ai-safety/https://two-works.netlify.app/game-theory-of-ai-safety/Why what labs say about safety is a strategic signal, not a statement of values — and what that means for regulation.Fri, 06 Feb 2026 23:25:26 GMTDr Gareth RobertsAI SafetyGame TheoryAI Evaluationhttps://two-works.netlify.app/architecture-and-ai/https://two-works.netlify.app/architecture-and-ai/Why the next leap in artificial intelligence will come from structure, not scale — and why a century of brain science has already drawn the blueprint.Fri, 06 Feb 2026 23:14:09 GMTDr Gareth RobertsCognitive NeuroscienceAI Architecturehttps://two-works.netlify.app/continuous-thought-machines-2/https://two-works.netlify.app/continuous-thought-machines-2/Time as the medium in which computation unfolds, measured and synchronised at the level of individual neurons.Fri, 06 Feb 2026 10:47:34 GMTDr Gareth RobertsAI ArchitectureCognitive Neurosciencehttps://two-works.netlify.app/the-ai-alignment-paradox/https://two-works.netlify.app/the-ai-alignment-paradox/In conventional security, hardening a system makes it harder to attack. You patch vulnerabilities, reduce attack surface, and defence moves in lockstep with robustness. AI alignment breaks this assumption.Fri, 06 Feb 2026 05:38:46 GMTDr Gareth RobertsAI SafetyAI EvaluationGame Theoryhttps://two-works.netlify.app/text-becomes-a-syscall/https://two-works.netlify.app/text-becomes-a-syscall/Late January 2026 will be remembered as the week a lobster meme accidentally taught the entire technology industry a serious idea.Wed, 04 Feb 2026 10:05:09 GMTDr Gareth RobertsAgentic AIAI SafetyVibe CodingContext Engineeringhttps://two-works.netlify.app/physics-shapes-ai-strategy/https://two-works.netlify.app/physics-shapes-ai-strategy/The pursuit of AI supremacy has reached an inflection point where fundamental physics, rather than algorithmic ingenuity alone, dictates competitive advantage.Sun, 17 Aug 2025 23:48:00 GMTDr Gareth RobertsAI ArchitectureAI EvaluationAI Safety