Artificial intelligence
Models, assistants and agents in practice — what works in a product, what it costs and where AI is overkill.
Articles · 13
-
How AI search picks sources
How an AI answer is assembled in six steps, what decides whether a page is quoted, an answer-first page and robots.txt, how to measure presence in AI answers and why AI does not quote a site.
Read
-
Voice AI
Eight tasks for voice AI, a chain of models versus one voice model, tools for recognition and synthesis, seven rules of a good voice product and common mistakes.
Read
-
AI for an online shop
Ten tasks where AI pays off in a shop, search that understands buyers, six rules for product descriptions by AI and common mistakes.
Read
-
Data from documents
How a model turns invoices, contracts and forms into fields: the path from a letter to a record, eight kinds of documents with checks, a schema and a check by code, rules and mistakes.
Read
-
AI in support
Five stages of introducing an assistant into support, ten topics with a verdict, six metrics, seven rules of a good handover to a person and common mistakes.
Read
-
AI assistant security
Ten threats to an AI assistant in plain words, how prompt injection works, protection in code before and after the model, eight rules and common mistakes.
Read
-
Checking AI quality
What to measure for each kind of task, how to build a test set in six steps, an example set and rubric, what to watch after launch and common mistakes.
Read
-
Prompts for business
How to write prompts for a product: eight parts of a business prompt, a real prompt for sorting requests with a JSON schema and examples, eight rules and common mistakes.
Read
-
Why AI makes things up
Seven reasons why language models invent answers, seven kinds of fabrication with ways to catch them, eight ways to make it rare and an instruction with a check by code.
Read
-
AI agents
How an AI agent works, how it differs from a chatbot and automation, ten tasks with a verdict, seven rules of safety and common mistakes.
Read
-
Where AI is overkill
When a language model only gets in the way: the principle of rule or understanding, twelve tasks with an honest verdict, five questions before adding AI and how to combine a model with code.
Read
-
The cost of AI
What you pay for when a product uses a language model: tokens, what one request is made of, five scenarios of a monthly bill, how to calculate your own and eight ways to cut it.
Read
-
Claude, GPT or Gemini
How Claude, GPT and Gemini differ: twelve criteria, three tiers in every family, which model for which task, how to choose on your own examples and common mistakes.
Read