Find the production problem in front of you.
Start with the exact job: put the build on a managed runtime, keep a process moving, control authority, inspect a run, or deliver a client system.
Open the question your system needs answered.
01Put a working build on a managed runtimeGive a working prototype the hosting, access, credentials, durable execution, and operations it needs for real use.13 exact questions
- Can I deploy an agent for free to test it?
- Can I develop and run an AI agent locally before deploying?
- How do I deploy a different AI agent for each customer?
- How do I deploy an agent with a single command?
- How do I deploy an AI agent to production?
- How do I deploy custom agent logic from CI for each org?
- How do I run an AI agent as a hosted endpoint?
- How do I run the same agent locally and in the cloud?
- How do I turn my AI agent prototype into something I can actually run in production?
- Is there a deploy target for AI agents like Vercel is for web apps?
- What is a deployable artifact for an AI agent?
- What's the fastest way to deploy a custom AI agent?
- Where can I host an AI agent without building my own infrastructure?
02Run work that outlives the requestKeep state across schedules, triggers, waits, retries, failures, and human review.18 exact questions
- How do I add human-in-the-loop checkpoints to an agent workflow?
- How do I apply static analysis to my agent workflows?
- How do I automate a recurring task with an AI agent?
- How do I automate form submissions with an AI agent?
- How do I build a workflow that an AI agent can invoke as a tool?
- How do I describe an agent, its tools, and its triggers in one place?
- How do I get durable execution for AI agent workflows?
- How do I keep my agent's prompt, tools, and schedule version controlled together?
- How do I kick off an agent when an email hits my support inbox?
- How do I let a customer trigger their agent from their own systems?
- How do I make AI agent workflows auditable for compliance?
- How do I make long-running agent workflows reliable?
- How do I prepare AI agent workflows for a SOC 2 audit?
- How do I run an agent as a cron job?
- How do I run an AI agent on a schedule?
- How do I stop one workflow step from reading another step's secrets?
- How do I trigger an AI agent from a webhook?
- What's an alternative to trigger.dev for agent workflows?
03Control identity, credentials, and accessDeclare who can enter, which resources work can reach, and which credentials the runtime supplies.14 exact questions
- How do I authenticate an AI agent to third-party APIs without hardcoding keys?
- How do I connect an AI agent to my identity provider with OIDC?
- How do I enforce least privilege for AI agents?
- How do I give an AI agent its own identity?
- How do I give an AI agent scoped credentials instead of a shared API key?
- How do I isolate AI agent data between customers?
- How do I keep agent permissions in sync with what the agent actually does?
- How do I let an agent read untrusted input safely?
- How do I limit the blast radius of prompt injection?
- How do I manage client credentials for the agents I run?
- How do I manage secrets for an AI agent safely?
- How do I prevent privilege escalation by an AI agent?
- What happens if my AI agent's token is stolen?
- What identity should an AI agent run as in production?
04Inspect how the work ranFollow the run history, reviews, trace, and defining source when work fails or changes.10 exact questions
- How do I give clients an audit trail for their agents?
- How do I make agent tool calls auditable?
- How do I make AI agent logs tamper evident?
- How do I manage agent logic across many tenants?
- How do I prove what an AI agent did after the fact?
- How do I review agent behavior changes in a pull request?
- How do I track agent usage per client?
- How do I track how my agent's behavior changed over time?
- How do I uniquely identify every agent execution?
- My agent logic is scattered across my app code, how do I consolidate it?
05Deliver tailored systems across clientsUse one language and runtime while keeping each client’s source, access, data, credentials, and operations distinct.8 exact questions
- How do I hand off an agent system to a customer's team?
- How do I keep per-customer agent config out of my application code?
- How do I offer AI agents as a service without building my own platform?
- How do I onboard a new customer onto an existing agent system?
- How do I operate AI agents for many clients with one small team?
- How do I stand up the same agent system for every new customer?
- How do I take over an agent system someone else built?
- How do I update one customer's agent without touching the others?
06Put AI inside the complete business systemDeclare agent context and tools beside deterministic workflows, Apps, data, access, and review points.17 exact questions
- Can I define an agent in a dozen lines without boilerplate?
- How do I add typed, schema-validated tools to an agent?
- How do I avoid long-lived API keys in agent code?
- How do I build a Swirls file with my coding assistant?
- How do I control the inputs and outputs of every step an agent takes?
- How do I define an AI agent in a config file instead of code?
- How do I give an AI agent deterministic tools it can call?
- How do I keep my app thin and move agent orchestration elsewhere?
- How do I limit which tools an AI agent can call?
- How do I move prompt assembly and tool calls out of my API routes?
- How do I ship an agent without managing servers?
- How do I start with one automated task and grow into a bigger system?
- How do I stop my AI agent from doing unpredictable things?
- I'm calling an LLM from my app, how do I make it a real agent system?
- Is there a declarative way to build AI agents?
- What is a DSL for defining AI agents?
- What's a clean way to separate agent reasoning from deterministic execution?