Research, playbooks, and field notes on IT, strategy, marketing, automation, design, and the systems that make them compound.
Buy off-the-shelf unless the software is either shallow enough to build cheaply or deep enough to be the thing you compete on. The middle, meaning anything moderately complex that is not your advantage, is where custom builds go wrong. Company size matters far less than most people assume.
SEO takes 30 to 60 days to tell you whether it is working, 3 to 12 months to produce rankings, and longer than that to produce revenue. Those are three separate clocks and most answers to this question report only the middle one, which is why the answer never feels useful.
You do not need three programmes. SEO, AEO and GEO describe one discipline at three scopes: ranking a page, being quoted as an answer, and being represented across AI-generated responses. The work overlaps heavily. If you have been quoted three separate retainers, that is a pricing decision rather than a technical one.
AI agent development cost in 2026 runs from about $8,000 for a proof of concept to $500,000 and beyond for an enterprise multi-agent system, with most mid-market production builds landing between $40,000 and $120,000. If you have three quotes that disagree by six times, they are not competing. They are quoting different products.
Self-hosting n8n costs about $12 to $30 a month for a server, plus four to six hours to set up and roughly one to two hours a month to maintain. The software itself is free. Price those hours at any realistic internal rate and self-hosting is more expensive than n8n Cloud until your volume gets large.
llms.txt is a plain-text file placed at the root of a website that lists, in Markdown, the pages an AI system should read and what each one covers. It was proposed in September 2024 by Jeremy Howard of Answer.AI as a way to give language models a clean, low-noise map of a site's documentation.
An AI agent is software given a goal rather than a script. It decides which steps to take, uses tools such as your database or email system to take them, checks whether the result worked, and retries if not. Businesses use agents for work with variable inputs and reviewable output: triage, research, drafting, classification.
Every B2B keyword research guide will tell you that search volume is the wrong metric. They are right, and almost none of them gives you the metric to use instead. B2B keyword research means finding the search terms your buyers use, tagging each one by what the searcher wants, and pricing each one in expected pipeline rather than expected traffic. The formula is volume times click rate times lead rate times close rate times deal value. Everything else is sourcing and sorting.
For most business processes, the right answer is plain workflow automation, not an AI agent. A chatbot answers questions. An automation runs steps you defined. An agent picks its own steps. If your process already has a known sequence, an agent adds cost and unpredictability and gives you nothing back.