How Ember works

AI features

Ember's AI features — chat and insights — read your plan through the same calculation engine that drives your dashboard. They never invent a number: every answer is either pulled straight from your data or computed by the engine and then explained in plain language.

What it can see

Both features have read-only access to your own scenario — assets, debts, pensions, contributions, planned events, residency timeline and the projection outputs. Nothing is sent anywhere else, and the AI cannot change your data: it can only read it and describe what it finds.

AI chat

Chat with Ember is a conversational way to ask about your plan — “what happens if I retire two years earlier?”, “why did my FIRE date move this month?”, “how much of my pot is in tax-advantaged wrappers?”. It calls the same projection and Monte Carlo routines the dashboard uses, so the numbers it quotes always match what you'd see if you ran the calculation yourself.

Answers report inflation-adjusted (real) figures by default, matching the rest of the app — say so explicitly if you want a nominal comparison.

AI insights

AI insights are one-shot, purpose-built reports rather than a back-and-forth conversation: a plain-language diff of what changed in your plan recently, a ranked explanation of why your FIRE date moved, a rolling monthly review, and a data-quality check that flags stale FX rates or gaps in your data.

Descriptive by default, advisory only when you ask

Every AI surface in Ember runs in descriptive mode unless you explicitly switch it on: it can explain your numbers, summarise changes, and describe trade-offs, but it is instructed never to recommend a specific security, allocation, or trade.

A small number of features (for example, the investment insight on an individual holding) offer an advisory mode you opt into per call, which relaxes that guardrail so the model can suggest a course of action for that specific question. It is never on by default and never persists to your next question.

Ember is an information and modelling tool, not financial advice, in either mode — see the full disclaimer.

Metered on Ember's own AI key

Every AI call runs on Ember's own hosted AI key — there is no option to supply your own API key. Usage is metered: paid plans include a monthly allowance of AI calls that resets each calendar month, and Blaze's allowance uses a stronger model than Ember's.

Your current allowance, usage so far this month, and any AI credit balance are shown on Settings → Account.

AI credits

Once a month's allowance runs out — or if you're on the free Spark plan — you can buy AI credits as a one-off top-up rather than upgrading your plan. Credits come in a small set of pack sizes and are priced in your own currency at checkout; Blaze accounts buy credits for the stronger model, everyone else buys credits for the standard one. Exact pack sizes and prices are shown on Settings → Buy AI credits before you pay anything.

Cost caps

Behind the monthly allowance sits a hard spending cap per account, so a single unusually long conversation can't run away with cost. If you hit the cap, Ember falls back to drawing from any AI credits you hold rather than failing outright; with no credits available you'll be prompted to buy a pack or wait for next month's allowance to reset.


Not sure which plan you need?

See Plans & billing for what each tier includes.