An A/B test asks which version of a page is better on average. Personalisation asks a different question: which version is better for the person actually reading it.
That difference matters more than it sounds. A test that ends with version B winning by four percent has found the page that suits your largest group of visitors slightly better. Everyone outside that group gets the page that lost, or the page that won despite not being written for them. The result is one page tuned to the average of people who are not average.
There is a practical cost too. A test needs enough traffic to reach significance, then it ends, and you set up another one. The work is in the running of them. Personalisation does not converge on a single winner, because there is no single winner to find: a developer and a marketing lead genuinely need different paragraphs, and neither of them is wrong.
The two are not opposed. Plenty of sites test layout and structure, where there really is one better answer, and vary the content inside that layout by who is reading. Testing the frame and personalising the contents is a reasonable way to hold both.
This is one atom from the Spectare content library. Spectare assembles the right atoms for each visitor in real time, based on who they are and how they arrived.