AI agent

The difference from a chatbot is fundamental and lives in the loop. A chatbot receives a message and returns text. An agent receives a goal, plans a step, calls a tool, gets a result, evaluates it and decides what to do next, repeating until the goal is met or a limiter fires. It is the combination of tools and a repeating loop that makes an agent an agent. A typical tool set in a commercial system: knowledge base search, reading and writing to the CRM, calls to an internal API, sending a notification, creating a task.

Where agents genuinely pay back: tasks made of several steps, where the steps resist being specified in advance, and where the cost of an error is moderate and reversible. Triaging inbound mail into CRM leads, first-line qualification of enquiries, gathering data from several systems into one report, drafting documents from templates, monitoring and flagging anomalies. Where the steps are known in advance and stable, an ordinary workflow or integration is cheaper, more reliable and needs no supervision, and replacing it with an agent adds non-determinism exactly where none was required.

Designing an agent is eighty percent limiters, and that is engineering reality rather than pessimism. You need an allowlist of the actions and data the agent can reach; a step and cost budget per task, or the loop can spin indefinitely; mandatory human confirmation for irreversible actions such as messaging a client, making a payment or deleting records; a full log of every tool call for incident review; and separate validation of whatever enters the input, because user-supplied data can carry instructions. An agent without those bounds in production is not automation, it is an incident generator with unclear accountability.

Related terms

• LLMAn LLM is a large language model: a neural network trained on vast amounts of text to predict the next token, which is why it can write, summarise and answer, yet it stores no facts the way a database does and can be confidently wrong.
• RAGRAG is a pattern where a language model first retrieves relevant fragments from your own knowledge base and answers from them, which delivers current data and a source citation without any fine-tuning of the model itself.
• Prompt engineeringPrompt engineering is the design of instructions for a language model: role, context, task, output format and constraints are written so the result is reproducible, and quality is judged against a set of test cases rather than by impression.
• APIAn API is a fixed set of rules by which one program requests data or actions from another: method addresses, request and response formats, authentication and rate limits, documented so that both sides can evolve independently.
• SLAAn SLA is a service level agreement: it fixes system availability as a percentage, response and resolution times by incident severity, the hours support operates, and the compensation owed when the provider fails to meet those numbers.

Related services

• AI and ML developmentWe build AI that solves a defined task and pays for itself, not demos for the sake of demos. Classifiers, recommendations, text and document processing, LLM assistants, and model integration into an existing product.
• IT consulting and product auditThe most expensive mistakes in software happen before the first line of code: a misread problem, a stack chosen out of a contractor’s habit, and a specification that does not exist. Consulting exists to settle all of that before the development meter starts running: what to build, out of what, at what cost and in which order. The result is a document, not an opinion on a call.
• Web app and Telegram Mini App developmentFor when a site is no longer enough and you need a product: an account area, a dashboard, an internal tool or a Mini App inside Telegram. We design the architecture, write the backend and frontend, and take it to release.

Read more

• An AI Assistant on Your Own Knowledge Base: How RAG Works and What It CostsHow a RAG assistant differs from a button-tree bot and from an LLM with no data, where hallucinations come from and what actually reduces them, what SaaS costs against a custom build, and which month the project breaks even.
• Where AI and ML Actually Pay Off in a Business, and Where They Are Expensive HypeMachine learning pays off where there is a repeated decision, a lot of similar data and tolerance for error. Missing any one of the three means do not. Here are the use cases that work, their realistic accuracy, the running costs and how to run a four to six week pilot.