Paper Compute blog on AI-native teams, agent telemetry, observability, runtime isolation, and turning successful sessions into reusable skills.https://papercompute.com/en-ushttps://papercompute.com/blog/dont-just-measure-code-measure-engineering-decisions/https://papercompute.com/blog/dont-just-measure-code-measure-engineering-decisions/The cost of an engineering decision is not the cost of making it. It's the work the decision causes afterward. We traced one architecture decision through 474 agent session records: $4,704.74 downstream of a $57 design doc, and four metrics (reach, duration, rework, amplification) any team with session records can track.Wed, 02 Sep 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/cassette-anatomy/https://papercompute.com/blog/cassette-anatomy/tapes captures your AI agent data, and a cassette is a small service that adds your own capabilities on top of it, served through the tapes API. This post walks through building one end to end: the API, its OpenAPI spec, and an MCP tool. Wed, 26 Aug 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/how-to-share-agent-knowledge-across-a-team/https://papercompute.com/blog/how-to-share-agent-knowledge-across-a-team/We analyzed two months of our team's sessions to find out what should become reusable knowledge. What we found is that repeatable work matters more than frequent work.Mon, 24 Aug 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/paper-clearings/https://papercompute.com/blog/paper-clearings/Parallel source trees are cheap; parallel runtimes are not. How a disposable, fully isolated platform environment per working tree lets multiple problems — worked by humans or agents — run side by side on one laptop.Thu, 20 Aug 2026 00:00:00 GMTEngineeringjasonhttps://papercompute.com/blog/tapes-open-core/https://papercompute.com/blog/tapes-open-core/How we're building the free and open source building-blocks for the future of AI dataMon, 17 Aug 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/dreams-on-paper/https://papercompute.com/blog/dreams-on-paper/Dreams sound mystical. In practice, they are a way to label agent traces, reflect across sessions, and turn repeated behavior into something the next run can use.Wed, 12 Aug 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/how-to-reduce-ai-agent-costs/https://papercompute.com/blog/how-to-reduce-ai-agent-costs/I audited our agent sessions expecting to find wasteful repeat calls. Repeats were only 4%. The real money was in agents re-reading what they already knew, plus a permission-check tax we could finally put a price on.Tue, 11 Aug 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/high-fi-traces/https://papercompute.com/blog/high-fi-traces/What we see our AI agents doing may not be exactly what's going on under the hood. Thu, 06 Aug 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/your-standup-is-a-search-query/https://papercompute.com/blog/your-standup-is-a-search-query/Your team keeps paying for the same solved problem, and because the tickets all close, nobody sees it happening. Repetition is evidence of a missing team capability, and standup is the detection layer that's been announcing it all along.Wed, 05 Aug 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/make-something-agents-want/https://papercompute.com/blog/make-something-agents-want/London's bridges are failing under load nobody modeled. Haussmann rebuilt Paris for traffic that hadn't arrived yet. Agent infrastructure faces the same choice: skills, shared team sessions, and reflections are how you build for the traffic that is coming.Tue, 28 Jul 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/the-coming-ai-layoffs-should-be-agents-not-engineers/https://papercompute.com/blog/the-coming-ai-layoffs-should-be-agents-not-engineers/Before companies use AI cost overruns to justify cutting people, they should terminate the agent workflows that consume money without producing reliable outcomes.Mon, 20 Jul 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/your-engineers-are-running-fable-5-on-merge-conflicts/https://papercompute.com/blog/your-engineers-are-running-fable-5-on-merge-conflicts/One session, $100.66, 91% Fable 5 — for confirming rollout steps. Multiply that across 200 engineers and you have a budget conversation nobody is ready for. The paper console makes the numbers visible.Thu, 16 Jul 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/stop-trusting-skills-you-havent-measured/https://papercompute.com/blog/stop-trusting-skills-you-havent-measured/Evidence gets a skill written. Evaluation keeps it worth keeping. A skill that isn't measured can silently make your agent worse, and, because skills execute, a wrong one does more damage than a wrong doc ever could. Here's why every skill needs an eval, and what a real one looks like.Wed, 01 Jul 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/stop-writing-skills-from-memory/https://papercompute.com/blog/stop-writing-skills-from-memory/Skills are about to become part of the AI engineering stack. If they don't carry provenance, we're just rebuilding prompt libraries with nicer names. A prompt starts with intention. A skill should start with evidence.Tue, 30 Jun 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/your-data-is-valuable/https://papercompute.com/blog/your-data-is-valuable/Frontier labs are paying a billion dollars a year for the same kind of session reasoning your team generates and deletes every night. The asset isn't the code the agent produced. It's the path the agent took.Mon, 15 Jun 2026 00:00:00 GMTThoughtsbekahhwhttps://papercompute.com/blog/ai-workflow-telemetry-primitives/https://papercompute.com/blog/ai-workflow-telemetry-primitives/AI agent workflows are no longer single requests. They are stateful, branching systems that mutate their own context over time. Request logs, traces, transcripts, and bills flatten that structure. Teams need telemetry primitives that preserve session lineage, cache topology, and prompt shape.Mon, 25 May 2026 00:00:00 GMTProductbekahhwhttps://papercompute.com/blog/paper-forest/https://papercompute.com/blog/paper-forest/Paper Compute's platform spans 25 repositories across 6 layers. Here's how we built a synthetic monorepo from git worktrees so coding agents stay coherent across the whole system — without Bazel, submodules, or a one-quarter migration.Thu, 07 May 2026 00:00:00 GMTEngineeringmatthttps://papercompute.com/blog/introducing-paper-compute/https://papercompute.com/blog/introducing-paper-compute/We built Paper Compute to close the Agent Gap — the inability to prove what an agent did, audit its actions, or replay its decisions. Here's what we built and why.Tue, 31 Mar 2026 00:00:00 GMTCompanybdougiehttps://papercompute.com/blog/tapes-deep-dive/https://papercompute.com/blog/tapes-deep-dive/How we built tapes to be a durable, cryptographically secure agent datasetFri, 27 Mar 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/tapes-sessions-to-skills/https://papercompute.com/blog/tapes-sessions-to-skills/I hit the same Confluent Cloud bugs across two projects. The second time, I extracted a skill from my tapes sessions instead of debugging from scratch. Now every agent I run knows the fix.Wed, 25 Mar 2026 00:00:00 GMTProductbdougiehttps://papercompute.com/blog/logs-are-the-self-healing-feedback-loop/https://papercompute.com/blog/logs-are-the-self-healing-feedback-loop/Agents don't need to remember everything. They need to observe what went wrong. Structured logs become telemetry, telemetry becomes signals, and signals become the context for the next run.Tue, 17 Mar 2026 00:00:00 GMTProductbdougiehttps://papercompute.com/blog/sandboxes-are-for-experiments/https://papercompute.com/blog/sandboxes-are-for-experiments/Once agents got good enough to clone repos, install dependencies, modify files, and execute code they just wrote, the natural question became: where does this actually run? Sandboxes were designed for short-lived, contained execution. Agents aren't snippets.Mon, 16 Mar 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/agent-played-pokemon/https://papercompute.com/blog/agent-played-pokemon/An AI agent played Pokémon Red for 1,000 turns without leaving the bedroom. The failure wasn't the model. It was the absence of telemetry.Mon, 09 Mar 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/agents-need-black-box-recorders/https://papercompute.com/blog/agents-need-black-box-recorders/Every commercial aircraft carries a flight recorder. Agent systems don't. The gap between an agent that works and an agent you can trust in production is telemetry.Wed, 04 Mar 2026 00:00:00 GMTProductbdougiehttps://papercompute.com/blog/stop-optimizing-the-model/https://papercompute.com/blog/stop-optimizing-the-model/Optimizing for model performance while ignoring execution infrastructure is like choosing the fastest race car without checking whether the track has guardrails. The real engineering challenge isn't squeezing more capability out of LLMs—it's figuring out how to actually run them safely.Tue, 03 Mar 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/introducing-stereos/https://papercompute.com/blog/introducing-stereos/stereOS is a NixOS based operating system designed for AI agents to be secure, self-managed, and efficientFri, 27 Feb 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/push-the-code-era-is-over/https://papercompute.com/blog/push-the-code-era-is-over/Agents are now consequential enough to warrant national-level oversight. Compliance doesn't require weaker AI—it requires observable AI. The observable era is the one that scales.Mon, 23 Feb 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/missing-primitive-agent-infrastructure/https://papercompute.com/blog/missing-primitive-agent-infrastructure/Visibility is the easiest piece. The hard part is analyzing and understanding what you're observing. The next phase of agent infrastructure won't be defined by visibility—it will be defined by durability.Thu, 19 Feb 2026 00:00:00 GMTCompanybdougiehttps://papercompute.com/blog/true-cost-of-claude-code/https://papercompute.com/blog/true-cost-of-claude-code/If you're paying $100/month but consuming multiples of that in value, you have to start wondering when that's going to catch up to you. The AI coding tool market is following a familiar playbook.Tue, 17 Feb 2026 00:00:00 GMTThoughtsbdougiehttps://papercompute.com/blog/continuous-ai-docs/https://papercompute.com/blog/continuous-ai-docs/Documentation that writes itself sounds like a gimmick. For tapes.dev, it's an operational reality. Here's how we wired continuous AI into our workflow to keep docs accurate as the codebase evolves.Mon, 16 Feb 2026 00:00:00 GMTProductbdougiehttps://papercompute.com/blog/cross-compiling-cgo/https://papercompute.com/blog/cross-compiling-cgo/How we solved the pain of cross-compiling Go projects with C dependencies by combining Zig as a drop-in compiler, Dagger for reproducible builds, and osxcross for macOS targets.Wed, 11 Feb 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/introducing-tapes/https://papercompute.com/blog/introducing-tapes/We built tapes to solve the observability gap in AI agent infrastructure. Every session recorded, every decision searchable, every conversation replayable.Tue, 10 Feb 2026 00:00:00 GMTEngineeringjohnhttps://papercompute.com/blog/the-missing-harness/https://papercompute.com/blog/the-missing-harness/Why AI Needs Distributed Systems InfrastructureFri, 19 Dec 2025 00:00:00 GMTCompanybdougie