The Ledger

The Economics Of Building AI-native.

Field notes on AI cost, probabilistic runway, and staying alive while your product scales.

Latest posts

• The AI spend control plane, defined — Cloud spend got FinOps. AI spend is more volatile, more granular, and changes by the minute. Here are the five jobs a tool has to do to actually govern it. Cost Economics · October 8, 2026
• Observability tells you what happened. Control decides what happens next. — LLM observability tools are great at explaining last night's bill. None of them can stop tonight's. That gap is the whole category. Guardrails & Control · October 8, 2026
• Why AI guardrails belong at the edge, not in your app code — Budget checks written inside your application are easy to bypass, hard to keep consistent, and always one deploy behind. The right place to stop spend is before the request leaves. Engineering · October 8, 2026
• The four laws of a cost SDK — An SDK that sits on every AI call is in the critical path of your product. If it can break your app, it shouldn't exist. Here are the rules ours is built around. Engineering · October 8, 2026
• Content-blind by design — your cost tool should never see your prompts — To tell you what your AI costs, a tool needs token counts. It doesn't need your customers' conversations. Here's why we drew that line, and where it holds. Building AtlasBurn · October 8, 2026
• Telling a busy agent from a broken one — A healthy agent working hard and an agent stuck in a loop both make lots of calls. Kill the wrong one and you break your product. Here's how to tell them apart from the traffic alone. Guardrails & Control · October 8, 2026
• What actually gets captured — the honest boundary of fetch instrumentation — AtlasBurn auto-captures AI calls by patching global fetch. That's elegant and it covers most stacks — but it has a real edge, and pretending it doesn't just creates support tickets. Engineering · September 19, 2026
• Setup should be one command, not a copy-paste ritual — The gap between "installed" and "actually sending data" is where most integrations die. We closed it with a single command that shows you the diff before it touches anything. Engineering · September 17, 2026
• Your AI cost tool shouldn't make you open a dashboard — We turned AtlasBurn into an MCP server. Now you ask your assistant "what's my AI runway?" and get an answer from your real data — without leaving the terminal. Building AtlasBurn · September 15, 2026
• The zero-baseline principle — why AtlasBurn assumes nothing — Most dashboards fill the gaps in your data with defaults and hand you a confident number. We made the opposite choice, on purpose. A field note on building AtlasBurn. Building AtlasBurn · September 10, 2026
• Runway isn't a number — it's a distribution — Your runway is usually reported as a single figure. For an AI-native company, that number is a lie of precision. Here's how to model it honestly. Forecasting & Risk · September 9, 2026
• Retry storms — how one outage triples your AI bill — When a provider degrades, your carefully-built retry logic doesn't protect you. It amplifies the cost. The outage doesn't lower your bill — it multiplies it. Guardrails & Control · September 8, 2026
• Why 2× customers doesn't mean 2× AI cost — The most dangerous assumption in an AI-native financial model is that cost scales linearly with users. It rarely does — and the gap is where runway quietly disappears. Cost Economics · September 7, 2026
• How much can a runaway agent burn in an hour? — A single misbehaving agent loop doesn't fail loudly — it succeeds, thousands of times, at full price. We ran the numbers. Guardrails & Control · September 5, 2026
• Model overkill — paying premium prices for commodity work — The most common AI cost leak isn't a bug. It's a premium model quietly answering questions a model a tenth of the price would handle just as well. Engineering · September 3, 2026
• Capturing token usage across OpenAI, Anthropic, and Gemini — Every provider reports usage differently, and streaming hides it in a different place again. Here's why cost instrumentation is annoying — and where to do it once. Engineering · August 31, 2026
• VaR for founders — the worst month you should plan for — Banks have priced tail risk for decades. Founders running variable AI spend need the same tool: not "what will I probably burn," but "how bad is the bad case. Forecasting & Risk · August 26, 2026
• AI inference is COGS. Start treating it like one. — Most AI-native teams bury inference spend in "R&D" or a cloud line item. But it's a cost of goods sold — and where you book it changes whether you actually know your margins. Cost Economics · August 21, 2026

Browse by topic

• Cost Economics — Unit economics, margins, and the real cost of scaling AI.
• Forecasting & Risk — Probabilistic runway, Monte Carlo, and modeling what comes next.
• Guardrails & Control — Runaway agents, budgets, and stopping cost before the invoice.
• Engineering — Instrumentation, SDKs, and provider cost optimization.
• Building AtlasBurn — Behind the engines — how we build AtlasBurn in the open.