https://sunrisegenai.com/blog/ Production essays on agents, retrieval, and data platforms from Sunrise Gen AI. en-us https://sunrisegenai.com/blog/prefill-vs-decode/ https://sunrisegenai.com/blog/prefill-vs-decode/ Fri, 09 Oct 2026 00:00:00 GMT Prefill is a compute burst that sets time to first token. Decode is a memory-bandwidth walk that sets inter-token latency. Shipping GenAI means budgeting both, plus the cache and the batcher behind them. https://sunrisegenai.com/blog/pre-ship-security-gate/ https://sunrisegenai.com/blog/pre-ship-security-gate/ Tue, 06 Oct 2026 00:00:00 GMT When agents write most of the code, pre-ship security cannot be a vibe check the night before launch. Secrets, authorization, data rules, abuse limits, supply chain, and logs become numbered gates the harness runs on every release. https://sunrisegenai.com/blog/design-skill-branding/ https://sunrisegenai.com/blog/design-skill-branding/ Fri, 02 Oct 2026 00:00:00 GMT Coding agents invent a new look every run unless taste lives as a loadable skill. Tokens, type, spacing, forbidden patterns, and a verify gate so the studio stays the same. https://sunrisegenai.com/blog/agents-as-markdown/ https://sunrisegenai.com/blog/agents-as-markdown/ Tue, 29 Sep 2026 00:00:00 GMT Most production agents are Markdown contracts, not framework code. Role, instructions, constraints, tool refs. Describing the agent is the product surface. Description without a harness is still a costume. https://sunrisegenai.com/blog/harness-cache-hits/ https://sunrisegenai.com/blog/harness-cache-hits/ Tue, 22 Sep 2026 00:00:00 GMT Prompt stability, tool-trace shape, and cache reuse are how you score a harness. An engineered layout earns hits. A vibe layout burns the floor. https://sunrisegenai.com/blog/reusable-skills/ https://sunrisegenai.com/blog/reusable-skills/ Fri, 18 Sep 2026 00:00:00 GMT A reusable skill is a named contract the harness can load. Role, constraints, tools, and a recipe that survives the next project. Slop skills are prompts in a costume. Production skills are how the loop stops relearning the same job. https://sunrisegenai.com/blog/numeric-quality-gates/ https://sunrisegenai.com/blog/numeric-quality-gates/ Tue, 15 Sep 2026 00:00:00 GMT Verify for coding agents is a stack of measurable contracts in the harness. Pick the gates that fit the job. Fail closed. Retry or escalate. Never ship on confidence theater. https://sunrisegenai.com/blog/slim-context-packs/ https://sunrisegenai.com/blog/slim-context-packs/ Sat, 12 Sep 2026 00:00:00 GMT Long policy in the prompt is not control. Slim packs plus hard stops — allowlists, budgets, stop sequences, refuse-when-unsourced — are control. Evaluation is product surface. https://sunrisegenai.com/blog/inspectable-agent-memory/ https://sunrisegenai.com/blog/inspectable-agent-memory/ Tue, 08 Sep 2026 00:00:00 GMT Opaque vectors and session blobs make memory a black box. Hierarchical files, portable over MCP, are how a person opens the last write before the next action fires. https://sunrisegenai.com/blog/agent-is-a-harness/ https://sunrisegenai.com/blog/agent-is-a-harness/ Tue, 01 Sep 2026 00:00:00 GMT A model generates tokens. An agent finishes work. The difference is the loop, the tools, the permissions, and a test gate in front of done — the harness is the product.