Best AI for AdvisorsWhat is a context window?
Short answer
How much text an AI model can take in and consider at once, including your prompt, files and the conversation so far.
Definition
The context window is the AI's working memory for a single conversation. Everything you send (the prompt, uploaded documents, earlier messages) and everything it writes back has to fit. It's measured in tokens, roughly pieces of words.
Bigger windows let a tool read a long trust document or a full year of meeting notes at once. When a conversation goes past the limit, older parts drop out or the tool refuses the input.
Why it matters for advisors
It explains odd behavior. If an assistant "forgets" something from the start of a long session, or summarizes only part of a 90-page document, you may have hit the window.
It also affects tool choice. Reviewing a full estate plan or comparing several tax returns needs a tool that can hold all of it, or one that uses retrieval to pull in the right pages.
It also explains some odd behavior. In a very long chat, the assistant may lose track of instructions you gave at the start. For important work, start a fresh conversation and restate the key instructions rather than relying on something you said an hour ago.
Example
An advisor asks an assistant to compare three years of tax returns in one chat. The summary misses the first year. Splitting the task into one return per request, then comparing the summaries, fixes it.
Watch out for
A big context window isn't the same as accuracy. Models can still miss details buried in long documents, so check key figures.
Related terms
- TokenThe small chunk of text (often part of a word) that AI models read, write and are priced by.
- Large language modelThe type of AI model behind chat assistants: trained on huge amounts of text to predict and generate language.
- Retrieval-augmented generationAn AI setup that first looks up relevant documents, then writes an answer based on them, so output is grounded in your sources.
Learn more
Last updated .