# Build on Anthropic > A complete, dated, linked reference for building on Anthropic: every customer case study, engineering pattern, and platform capability, organized for founders shipping on the Claude Developer Platform. ## Docs - [Case Studies](https://buildonanthropic.com/case-studies): 256 published Anthropic customer stories, organized by company stage, industry, use-case pattern, and how hard the evidence actually is. - [The Startup Cut](https://buildonanthropic.com/case-studies/startups): Anthropic tags 29 customer stories as startups, and 21 of the 256 are Y Combinator companies at every stage from current batch to exit. - [Strength of Evidence](https://buildonanthropic.com/case-studies/strength-of-evidence): Not all published metrics are equally defensible. A tiering of the corpus by method, plus the specific pages that contradict themselves. - [The Index](https://buildonanthropic.com/case-studies/the-index): Every published Anthropic customer story in one searchable table: 256 rows, each linked to the official first-party page. - [Changelog](https://buildonanthropic.com/changelog): A log of every change to this wiki (new, updated, removed), newest first, grouped by month. - [Agent Once, Session Every Run](https://buildonanthropic.com/concepts/agent-once-session-every-run): The Managed Agents usage rule: create a versioned agent config once, then create a session per run. Creating agents in the request path is the documented anti-pattern. - [Consumption Pricing](https://buildonanthropic.com/concepts/consumption-pricing): The Claude Developer Platform bills by tokens consumed, not seats sold. A builder's bill scales with how much their product actually runs, which aligns the platform with usage that works. - [The Four Ways to Build](https://buildonanthropic.com/concepts/four-ways-to-build): Manual loop, Tool Runner, Agent SDK, Managed Agents: the four paths onto the platform, separated by who supplies the harness and who supplies the deployment. - [Concepts](https://buildonanthropic.com/concepts): The Build on Anthropic ontology: nineteen concepts covering the structural spine of the platform, the Managed Agents primitives, the cost levers, and the patterns proven in the customer corpus. - [Long-Running Work](https://buildonanthropic.com/concepts/long-running-work): Agent runs measured in hours, not seconds: the workload class that makes the harness and deployment decisions matter, with published runs of four, six, and seven hours. - [Memory Stores](https://buildonanthropic.com/concepts/memory-stores): Workspace-scoped persistent memory for Managed Agents: a filesystem mounted into every session, versioned, with an audit trail, so what an agent learns survives the run. - [Outcomes](https://buildonanthropic.com/concepts/outcomes): The Managed Agents primitive for defining done: you supply a rubric, and the platform runs an iterate, grade, revise loop with a separate grader until the work passes. - [Prompt Caching](https://buildonanthropic.com/concepts/prompt-caching): Cache reads cost a fraction of base input price, but the cache is a strict prefix match: one changed byte invalidates everything after it. Stable content goes first. - [Scheduled Deployments](https://buildonanthropic.com/concepts/scheduled-deployments): Cron for agents on Managed Agents: a schedule fires a fresh session on the clock, every firing leaves a run record, and the whole thing can be paused or archived. - [Self-Correction Mid-Task](https://buildonanthropic.com/concepts/self-correction): An agent that checks its own output against reality before handing it over, and fixes what it finds: validate in a sandbox, look at the rendered result, retry on failure. - [Silent Failure](https://buildonanthropic.com/concepts/silent-failure): The failure mode that kills trust in unattended agents: the agent thinks it finished, and it did not. Loud failure is recoverable; silent failure is the trust problem. - [The Audit Trail](https://buildonanthropic.com/concepts/the-audit-trail): Citation to source as a product feature: every answer carries a pointer to the exact place it came from, so the human verifies instead of hunts. - [The Batch API](https://buildonanthropic.com/concepts/the-batch-api): 50 percent off input and output tokens for anything that can wait up to 24 hours. The discount stacks with prompt caching. - [The Company Brain](https://buildonanthropic.com/concepts/the-company-brain): A persistent store of how a specific company works that agents read on every run and write back to as they learn. The pattern an entire YC batch was asked to build. - [The Deployment](https://buildonanthropic.com/concepts/the-deployment): Everything an agent runs on: sandboxing, session persistence, tenant isolation, secrets, and capacity. The infrastructure half of the agent question. - [The Effort Dial and Model Splitting](https://buildonanthropic.com/concepts/the-effort-dial): Spend frontier-model tokens on judgment and small-model tokens on commodity steps, and turn the effort parameter down where deep thinking is wasted. - [The Harness](https://buildonanthropic.com/concepts/the-harness): The software loop around the model: tool dispatch, context management, retries, compaction. The half of an agent that is code, not model. - [The Human Gate](https://buildonanthropic.com/concepts/the-human-gate): An approval step the platform enforces before an agent's action commits. A prompt can ask for approval; a permission policy guarantees it. - [The No-API Integration](https://buildonanthropic.com/concepts/the-no-api-integration): Using computer use to operate the screens of systems that will never get an API. The largest architecture requirement in the newest YC batch, and a proven pattern in the customer corpus. - [Vaults](https://buildonanthropic.com/concepts/vaults): The Managed Agents credential store: secrets never enter the sandbox, they are substituted into outbound requests at egress, so a prompt injection has nothing to steal. - [Glossary](https://buildonanthropic.com/reference/glossary): One-line-per-term summary across the whole wiki. - [Graphic Style](https://buildonanthropic.com/reference/graphic-style): Visual conventions for illustrations and diagrams on this wiki. The default is comic-strip article summaries in the Supersuit Up family style. - [Reference Index](https://buildonanthropic.com/reference): The dated master list of what Anthropic has published for builders, organized by surface, so you can find the primary source instead of a summary of it. - [Tools](https://buildonanthropic.com/reference/tools): One page per tool. What it is, what it costs, when to use it. - [Voice Rules](https://buildonanthropic.com/reference/voice-rules): Writing constraints every page is held to. Edit per wiki. - [Start Here](https://buildonanthropic.com/start-here): The map to everything Anthropic has published about building on the Claude Developer Platform, organized so a founder can find the one thing that matters to them. - [The Tools](https://buildonanthropic.com/tools): Two paste-in files: one gives any founder a personalized architecture read, the other lets an operator mint a custom edition of that experience for their own audience. --- Generated at build time. 32 pages indexed.