Building Developer Tools That Last: A Conversation with Matan Kushner

End-to-end latency in real-time pipelines comes from coordination across systems, not from any single component. Four common workloads, tested two ways, show where multi-hop architectures compound delays and where collapsing storage, messaging, and compute into one runtime changes the math.

Comparison

Kafka-Centered Stacks vs. a Single Harper Cluster: Where Real-Time Latency Actually Comes From

End-to-end latency in real-time pipelines comes from coordination across systems, not from any single component. Four common workloads, tested two ways, show where multi-hop architectures compound delays and where collapsing storage, messaging, and compute into one runtime changes the math.

Comparison

Kafka-Centered Stacks vs. a Single Harper Cluster: Where Real-Time Latency Actually Comes From

Most teams treat a cache as a black box: URL-keyed blobs with a TTL, useful for speed and nothing else. In Harper, cached data lands in a real table inside the same query engine. That means filtering, joining, real-time subscriptions, and vector search all work against it.

Your API cache is secretly a database

Cache

Tutorial

Most teams treat a cache as a black box: URL-keyed blobs with a TTL, useful for speed and nothing else. In Harper, cached data lands in a real table inside the same query engine. That means filtering, joining, real-time subscriptions, and vector search all work against it.

Kris Zyp

SVP of Engineering

Most teams treat a cache as a black box: URL-keyed blobs with a TTL, useful for speed and nothing else. In Harper, cached data lands in a real table inside the same query engine. That means filtering, joining, real-time subscriptions, and vector search all work against it.

Your API cache is secretly a database

Most teams treat a cache as a black box: URL-keyed blobs with a TTL, useful for speed and nothing else. In Harper, cached data lands in a real table inside the same query engine. That means filtering, joining, real-time subscriptions, and vector search all work against it.

Your API cache is secretly a database

Most teams treat a cache as a black box: URL-keyed blobs with a TTL, useful for speed and nothing else. In Harper, cached data lands in a real table inside the same query engine. That means filtering, joining, real-time subscriptions, and vector search all work against it.

Your API cache is secretly a database

Learn how to keep AI-assisted side projects on track using a repo-native coordination layer that preserves context across session gaps.

How to Keep Massive AI-Assisted Side Projects on Track

A.I.

Blog

Learn how to keep AI-assisted side projects on track using a repo-native coordination layer that preserves context across session gaps.

Aleks Haugom

Senior Manager of GTM

Learn how to keep AI-assisted side projects on track using a repo-native coordination layer that preserves context across session gaps.

How to Keep Massive AI-Assisted Side Projects on Track

Learn how to keep AI-assisted side projects on track using a repo-native coordination layer that preserves context across session gaps.

How to Keep Massive AI-Assisted Side Projects on Track

Learn how to keep AI-assisted side projects on track using a repo-native coordination layer that preserves context across session gaps.

How to Keep Massive AI-Assisted Side Projects on Track

Deserializing entire records to read one field is a bottleneck at scale. Structon stores objects in a binary format where any field is reachable by byte offset, with lazy getters that never allocate until you access a property. It's the encoding Harper has used internally for years, now a standalone package.

Introducing Structon: Random-Access Binary Encoding for JavaScript

JavaScript

Tutorial

Deserializing entire records to read one field is a bottleneck at scale. Structon stores objects in a binary format where any field is reachable by byte offset, with lazy getters that never allocate until you access a property. It's the encoding Harper has used internally for years, now a standalone package.

Kris Zyp

SVP of Engineering

Deserializing entire records to read one field is a bottleneck at scale. Structon stores objects in a binary format where any field is reachable by byte offset, with lazy getters that never allocate until you access a property. It's the encoding Harper has used internally for years, now a standalone package.

Introducing Structon: Random-Access Binary Encoding for JavaScript

Deserializing entire records to read one field is a bottleneck at scale. Structon stores objects in a binary format where any field is reachable by byte offset, with lazy getters that never allocate until you access a property. It's the encoding Harper has used internally for years, now a standalone package.

Introducing Structon: Random-Access Binary Encoding for JavaScript

Deserializing entire records to read one field is a bottleneck at scale. Structon stores objects in a binary format where any field is reachable by byte offset, with lazy getters that never allocate until you access a property. It's the encoding Harper has used internally for years, now a standalone package.

Introducing Structon: Random-Access Binary Encoding for JavaScript

Production agents spend 56–74% of their time waiting on tool calls and infrastructure, not the model. Microservices were the right answer for the web era; agentic workloads need the stack pulled back into a single unified runtime.

Microservices Split the Stack, but Agents Want It Whole

Blog

Production agents spend 56–74% of their time waiting on tool calls and infrastructure, not the model. Microservices were the right answer for the web era; agentic workloads need the stack pulled back into a single unified runtime.

Aleks Haugom

Senior Manager of GTM

Production agents spend 56–74% of their time waiting on tool calls and infrastructure, not the model. Microservices were the right answer for the web era; agentic workloads need the stack pulled back into a single unified runtime.

Microservices Split the Stack, but Agents Want It Whole

May 2026

Production agents spend 56–74% of their time waiting on tool calls and infrastructure, not the model. Microservices were the right answer for the web era; agentic workloads need the stack pulled back into a single unified runtime.

Microservices Split the Stack, but Agents Want It Whole

Production agents spend 56–74% of their time waiting on tool calls and infrastructure, not the model. Microservices were the right answer for the web era; agentic workloads need the stack pulled back into a single unified runtime.

Microservices Split the Stack, but Agents Want It Whole