<![CDATA[Stories by Agent Native on Medium]]> https://medium.com/@agentnativedev?source=rss-3dea44a5d468------2 https://cdn-images-1.medium.com/fit/c/150/150/1*dt5tcaKMBhB6JboQ9lIEAA.jpeg Stories by Agent Native on Medium https://medium.com/@agentnativedev?source=rss-3dea44a5d468------2 Medium Sat, 14 Nov 2026 01:11:06 GMT <![CDATA[Local Inference Engine That Makes Agents Faster on Mac]]>

Local models have improved enough to handle useful development work.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/local-inference-engine-that-makes-agents-faster-on-mac-d8b1ff3e924f?source=rss-3dea44a5d468------2 https://medium.com/p/d8b1ff3e924f Thu, 12 Nov 2026 09:21:07 GMT 2026-11-12T09:21:07.504Z <![CDATA[Your Agent Knows SQL But Doesn’t Understand Your Data]]>

LLMs keep getting better at writing SQL, but enterprise databases are still full of ambiguous business language.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/your-agent-knows-sql-but-doesnt-understand-your-data-d580492f531e?source=rss-3dea44a5d468------2 https://medium.com/p/d580492f531e Wed, 11 Nov 2026 11:03:37 GMT 2026-11-11T11:03:37.768Z <![CDATA[Open-Source Jev on Apple Silicon and CUDA]]>

The hardest part of production is making thousands of simple decisions reliably.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/open-source-jev-on-apple-silicon-and-cuda-5edcb1aa7392?source=rss-3dea44a5d468------2 https://medium.com/p/5edcb1aa7392 Mon, 09 Nov 2026 09:23:21 GMT 2026-11-09T09:23:21.066Z <![CDATA[Unlock 2x Tensor Throughput on NVIDIA RTX GPUs]]>

RTX 3090 delivers about 75 TFLOPS on a BF16 matrix multiplication benchmark.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/unlock-2x-tensor-throughput-on-nvidia-rtx-gpus-5d178ab121d0?source=rss-3dea44a5d468------2 https://medium.com/p/5d178ab121d0 Sun, 08 Nov 2026 07:53:11 GMT 2026-11-08T07:53:11.779Z <![CDATA[Understand Any Codebase Instantly (with up to 99.7% fewer tokens)]]>

In this article, we will turn a repository into a queryable graph, which will provide you 30 to 99.7% fewer tokens on the development…

Continue reading on Medium »

]]> https://agentnativedev.medium.com/understand-any-codebase-instantly-with-up-to-99-7-fewer-tokens-c7f944f049e0?source=rss-3dea44a5d468------2 https://medium.com/p/c7f944f049e0 Sat, 07 Nov 2026 07:19:21 GMT 2026-11-07T07:19:21.953Z <![CDATA[6 Open-Source Projects for Serious AI Infrastructure]]>

If you are building agents that need to run somewhere other than a notebook, these six open-source projects are worth knowing.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/6-open-source-projects-for-serious-ai-infrastructure-50a1f8b769ff?source=rss-3dea44a5d468------2 https://medium.com/p/50a1f8b769ff Fri, 06 Nov 2026 13:02:45 GMT 2026-11-06T13:02:45.818Z <![CDATA[Context Engineering Is Dead. Long Live Cache Engineering.]]>

What happens when we give the agent direct control over its active KV cache?

Continue reading on Medium »

]]> https://agentnativedev.medium.com/context-engineering-is-dead-long-live-cache-engineering-ec238e672346?source=rss-3dea44a5d468------2 https://medium.com/p/ec238e672346 Thu, 05 Nov 2026 08:56:24 GMT 2026-11-05T08:56:24.478Z <![CDATA[Claude Bug Hunter]]>

To guard against cyber attacks, you have to package security methodologies, patterns from disclosed reports, tool integrations, and…

Continue reading on Medium »

]]> https://agentnativedev.medium.com/claude-bug-hunter-f3982a907b34?source=rss-3dea44a5d468------2 https://medium.com/p/f3982a907b34 Wed, 04 Nov 2026 07:21:13 GMT 2026-11-04T07:21:13.734Z <![CDATA[Pure C inference engine to run MoE models (2.8T) on consumer hardware]]>

744-billion-parameter model on a laptop with 25 GB of RAM sounds absurd. The quantized checkpoint alone takes up hundreds of gigabytes.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/pure-c-inference-engine-to-run-moe-models-2-8t-on-consumer-hardware-761407e1143e?source=rss-3dea44a5d468------2 https://medium.com/p/761407e1143e Tue, 03 Nov 2026 08:07:19 GMT 2026-11-03T08:07:19.629Z <![CDATA[Unified Interface for Agent Harnesses]]>

Everybody can connect an LLM to their backend in an afternoon but running a coding agent as a reliable product feature takes a lot longer.

Continue reading on Medium »

]]> https://agentnativedev.medium.com/unified-interface-for-agent-harnesses-a641a62a0696?source=rss-3dea44a5d468------2 https://medium.com/p/a641a62a0696 Mon, 02 Nov 2026 17:04:39 GMT 2026-11-02T17:04:39.645Z