AI agent glossary
The AI agent terms business teams run into, defined in plain words, each with a note on what it means when you work with Aether.
- AI agentAn AI agent is software that does a task for you: it reads a request, decides what to do, uses your systems to do it, and checks the result before it answers.
- AI agent calibrationAI agent calibration is setting an agent up to work the way your business does: your policies, tone of voice, process steps and the cases it must hand to a person.
- AI agent memoryAI agent memory is what an agent keeps from earlier work, such as a customer's previous messages, and how long it keeps it: one task, a day, or until you delete it.
- AI agent portalAn AI agent portal is where a business sees and manages the AI agents working for it: what they did, which projects they belong to, and who on the team can see what.
- AI document processingAI document processing is using AI to read documents such as claims, invoices and applications, pull out the details you need, check them, and file them as clean records.
- AI guardrailsAI guardrails are the limits that keep an AI agent safe to use: what it may do, what it must never do, how long it may work on a task, and what it can and cannot reach.
- Audit trail for AI agentsAn audit trail for AI agents is the step-by-step record of what an agent did on each task: what came in, what it looked up, what it decided and what it sent back.
- Human in the loopHuman in the loop means an AI agent hands a decision to a person when it should not make it alone, such as a large refund, a complaint or an unusual case.
- Managed AI agentsManaged AI agents are AI agents a provider builds, runs and maintains for you, so your business gets the work done without building or operating the technology.
- Prompt injectionPrompt injection is an attempt to trick an AI agent with instructions hidden in a message or document, such as telling it to ignore its rules or reveal private data.
Ready to put your first agent into production?
Bring us one process. We’ll map the skills, rules and integrations, and help you choose cloud, on-premises or a managed service.