WRITING
Blog
Notes from building IsoKron — design decisions, architecture trade-offs, and what we learned from the customer-review checkpoint.
The hash chain we ship — why your audit log can't be tampered with
IsoKron team
Per-tenant chain heads, RFC 8785 JCS canonicalization in the application layer, hourly Merkle anchors to immutable storage. The architecture that lets us promise an audit log that's evidence, not just record.
- security
- audit
- foundation
Bring Your Own Fleet — why your existing GPUs are the right execution layer
IsoKron team
Your Qwen-Coder-class compute is closer to your code, your data, and your money than any hosted fleet. Why IsoKron is built around BYO-fleet from day one — and where we're going with hardware-native execution agents.
- byo-fleet
- infrastructure
- ai-codegen
Why AI codegen needs structured outputs, not chat history
IsoKron team
Chat-history-based agents lose schema. Structured-output enforcement at every stage means the compiler can't silently corrupt your project. Zod, JSON Schema, and the discipline that comes with them.
- architecture
- structured-outputs
- ai-codegen
BYOK economics — why we pay only for security
IsoKron team
Bring your own keys, pay your upstream provider directly, and let us pay for the operator-side security critic on every compile. Here's the economic argument for why this is the right shape.
- pricing
- byok
- economics
The four-layer tenant isolation we ship
IsoKron team
Postgres RLS alone isn't enough when service-role connections bypass it. Here are the four independent layers we enforce — and the 109 tests that keep them honest.
- security
- multi-tenant
- postgres
Why structured graphs beat markdown for AI codegen
IsoKron team
Markdown architectures lose the relationships between decisions, components, and tickets. Structured graphs preserve them — and that's the difference between AI codegen that scales and AI codegen that drifts.
- architecture
- ai-codegen
- knowledge-graph
chain_height: pendingverified pending