# Notes from AgentCompile.

What we're building, what we measured, and how to add it to your agent.

- [Introducing AgentCompile](/blog/introducing-agentcompile.md): October 7, 2026. Your agent does the same jobs over and over, and thinks each one through from scratch. AgentCompile learns those jobs from its own history and runs them compiled: same answers, far fewer model calls.
- [How to reduce LLM calls in an AI agent](/blog/how-to-reduce-llm-calls-in-ai-agents.md): October 4, 2026. An agent calls its model on every turn of its loop. Here is what each common fix actually cuts, which ones make calls go away, and where to start.
- [τ-bench, explained: the benchmark behind our numbers](/blog/tau-bench-explained.md): October 1, 2026. Every number on this site comes from τ-bench retail. What the benchmark is, how it grades an agent, why we use it, and what it can't tell you.
- [Prompt caching, model routing, a semantic cache or compiling: what each one cuts](/blog/prompt-cache-model-routing-or-compiling.md): September 28, 2026. Four ways teams try to make agents cheaper and faster, side by side: what each one does, what it leaves alone, and which ones work together.
- [Where AgentCompile fits, and where it doesn't on purpose](/blog/where-agentcompile-fits.md): September 25, 2026. Support desks, SaaS workflows, phone lines, data teams, back offices. The industry changes; the shape of the work doesn't. A map of where AgentCompile works, where it's heading, and what we leave alone.
- [Customer support and operations agents](/blog/customer-support-agents.md): September 22, 2026. Refunds, exchanges, cancellations, account updates. Customers ask in a thousand ways; the jobs underneath are a short list.
- [Workflow agents across SaaS tools](/blog/saas-workflow-agents.md): September 19, 2026. Email, calendar, CRM and ticketing. Schedule, update, file, reply: the same shape every time.
- [Voice agents](/blog/voice-agents.md): September 16, 2026. Booking, rescheduling, checking an order, changing an account. The same jobs, out loud.
- [Data and SQL agents](/blog/data-and-sql-agents.md): September 13, 2026. Weekly revenue by region, every Monday, with new dates. The question repeats, so the query can too.
- [Document and extraction agents](/blog/document-extraction-agents.md): September 10, 2026. Invoices, receipts and contracts into fields. Reading stays with the model; what comes after is the repeated part.
- [Internal ops and IT agents](/blog/internal-ops-and-it-agents.md): September 7, 2026. Reset access, provision an account, rotate a key. Lookups, a write, and usually an approval.
- [Computer-use and desktop agents](/blog/computer-use-agents.md): September 4, 2026. The same jobs, clicked through native apps. Where AgentCompile is heading.
- [Multi-agent handoffs](/blog/multi-agent-handoffs.md): September 1, 2026. One agent chooses which specialist takes over. The choice repeats, and so does each specialist's work.
- [Your agent needs muscle memory](/blog/ai-muscle-memory.md): August 29, 2026. A pianist doesn't think about scales. Your agent still thinks through every refund. On skills that become automatic, and why agents don't have them yet.
- [What your agent does all day](/blog/what-your-agent-does-all-day.md): August 26, 2026. Agentic work looks open-ended from the outside. Read the logs and most of it is the same handful of jobs, done again and again.
- [Trust is earned on your own history](/blog/trust-is-earned-on-history.md): August 23, 2026. Before a job runs without your agent, it has to get your past conversations right. Why we start from history, and what stays protected.
- [58% fewer agent calls, with the same answers](/blog/58-percent-fewer-agent-calls.md): August 20, 2026. What we measured on τ-bench retail, how we measured it, and how to read the numbers.
- [Not a cache. Not fine-tuning.](/blog/not-a-cache-not-fine-tuning.md): August 17, 2026. Two things AgentCompile gets mistaken for, and what it is instead.
- [Adding AgentCompile in one line](/blog/adding-agentcompile-in-one-line.md): August 14, 2026. Install the SDK, wrap the client your agent already uses, and give each conversation its id.