Explore

Will Percey, Solutions Architect knowledge base

new_releases Recently Updated

buildApplied Neuro-SymbolicWiring models to exact engines
securityAI SecurityOWASP LLM Top 10, 2026 edition
policyAuthorisation PolicyDeciding what an agent may do
deployed_codeSandboxing & IsolationContaining what agents execute
terminalCode ExecutionThe generate, run, observe loop
exploreSteeringCorrecting an agent mid-flight
fact_checkDeterministic ConversationScripted speech in a generative agent
join_innerNeuro-Symbolic AILearned perception, symbolic guarantees

language My Websites

View All arrow_forward

article Blog

View All arrow_forward

cloud Cloud Providers

View All arrow_forward

smart_toy LLM & Agentic AI

View All (30 more)arrow_forward

neurology Deep Learning & MLOps

View All (15 more)arrow_forward

database Data

View All arrow_forward

architecture Architecture & Patterns

View All (1 more)arrow_forward

functions Computer Science Fundamentals

View All (5 more)arrow_forward

apps Technologies

View All arrow_forward

stacks Tool Stacks

View All arrow_forward

radar Tech Radar

View All arrow_forward
sports_esportsGoogle GenieGoogle DeepMindCreating video games in real-time in front of your eyes. Too expensive but bright future.
terminalOllamaOllamaThe new CLI goes beyond ollama run. ollama launch turns it into a platform, local models inside Claude Code, agent wrappers, and an enormous model catalogue
business_centerNemoClawNVIDIAThe first enterprise-grade claw variant from a major vendor. Compliance, confidential computing, and NeMo integration in one platform — the claw to watch for regulated industry deployments.
securityNanoClawqwibitai (community)Proof that a claw agent can be both capable and fully auditable. Container isolation by default, Claude SDK foundation, and a codebase small enough that one person can genuinely understand the whole thing.
smart_toyKimi K2Moonshot AIImpressive multi-agent orchestration, a pattern to follow once it matures.
forumMoltbookOpenClawBy the creators of OpenClaw. A social network where only AI agents can post, humans just observe.
widgetsGoogle OpalGoogle LabsVibe coding for non-developers. Describe an app in plain English and Opal builds it. Conversational and visual editing modes, with deeper Google integrations on the roadmap.
hiveHiveAdenGoal-driven agent development where you declare outcomes instead of hardcoding workflows. Self-healing graphs, built-in judge and escalation, and a TUI dashboard for observability.
groupsMetaGPTDeepWisdomRole-playing agents that mimic a full dev team. Niche but interesting for exploring multi-agent coordination patterns and development automation.
verifiedAIUC-1AIUCThe missing piece for enterprise AI adoption: a verifiable, independently audited standard that replaces bespoke vendor security questionnaires.
graphic_eqElevenLabsElevenLabsVoice is the missing interface for agentic systems. Their Conversational AI SDK turns any agent into a voice agent with tool calling and sub-second latency.
record_voice_overEVAServiceNow ResearchThe accuracy-vs-experience tradeoff is real and previously unquantified. Directly relevant to voice agent work — EVA's dual-axis scoring is the right mental model for evaluating spoken AI.
menu_bookNotebookLMGoogle LabsAI-powered research and note-taking.