A tour of your Overview dashboard
Your Overview page (labeled "My Overview" in the app) is what you land on after signing in. It turns your AI coding sessions for the selected period into one running story: what you paid, what the work was worth, where the spend went, and when it happened. This article walks through the page section by section, top to bottom.
At the top of the page you can narrow everything below by Date, Tool, and Repo using the filter chips next to the page title. Every section described below reflects whatever date range and filters are currently selected.
The Statement hero: paid vs. value
The first card on the page is your "personal value statement," numbered like a receipt (for example "№ 2026-07"). Its headline answers the core question in one sentence:
"You paid $X. The work was worth $Y."
Paid is either your subscription cost, if you've added a billing plan (shown as "Subscription · fully consumed"), or, if you haven't, what your usage would cost at published API rates (with an "add billing plan" link). Value is the human-equivalent cost — an estimate of what it would have cost to have an engineer produce the same work by hand.
Below the headline, a sub-line spells out the math in plain terms, for example: "That is ~1,043 engineer-hours at $150/hr, across 19 merged PRs in 4 repos."
Under that sits a row of smaller cards — the "value ladder" — read left to right:
You paid — your subscription cost, or API-rate cost if no plan is on file.
Same tokens, metered — the same usage priced at today's public API rates. This card only appears once you've added a billing plan, since without one your "paid" figure already is the API-rate cost.
Effective expert-hours — marked "proven," this is the human-equivalent cost: the engineer-hours the work represents, multiplied by an hourly rate.
Value multiple — highlighted in the row, this shows the payoff as a multiple, e.g. "every $1 paid → $12 of work."
The card's footer states that every figure traces back to a token-level receipt, with a "How we measure" link to the methodology page, and an Adjust calculations link that takes you to Preferences — see the note at the end of this article.
Spend to Value: your per-repo table
Below the statement is the Spend → Value table — a breakdown of the same period's delivered value, one row per repository, so you can see exactly where the spend went and what it returned.
The table header shows the active date range and repo count (for example "this month · 10 repos"), and each row has three parts:
Repo — the repo name, plus its merged-PR count and token volume.
Share of value — a bar showing that repo's percentage share of the period's total delivered value.
Value out — the dollar value of work delivered for that repo.
Rows are ranked by value delivered, highest first. Your top five repos each get their own row; everything past that is folded into a single "Other · N repos" row so the table stays short even with a lot of repos.
Hover (or tap, on touch) a row to open a small card with that repo's full period activity: merged PRs, commits, sessions, tokens, AI spend, and value out — plus, when there's enough spend to compare against, a leverage multiple (value ÷ spend) and a cost-per-merged-PR figure.
A footer line under the table calls out how much of the period's value your top repo carried, with a "View repos" link through to the full repo breakdown.
Checkpoints: the KPI strip
Next is a row of quick-glance tiles covering the same period:
Cost / merged PR — your average AI spend per merged pull request, with an API-rate comparison in its subtext once you have a billing plan.
Spend this period — total spend, with the number of sessions behind it.
Merged PRs — how many PRs merged this period, with the percent of spend that went to merged ("shipped") work.
Active days — how many days had at least one session, with total sessions logged.
Commits — how many commits this period are linked to an AI session.
Engineer-hours — the human-equivalent hours the period's work represents, with how much actual AI-active time it took to produce them.
The spend heat calendar
The "Heatmap" panel lays the period out as a calendar grid, Monday through Sunday, one cell per day. Darker cells mean more spend that day; days with no spend stay a plain neutral shade, and days still in the future (for the current period) render empty with a dashed border.
Cells with spend show the day number and the dollar amount spent. Hover or tap a day to see a card with its date, total spend, a mini breakdown of the top models used that day, token volume, session count, and merged PRs.
A caption under the grid reinforces the read: work tends to cluster into bursts rather than a steady daily drip.
Daily rhythm: your hourly strip
The "Daily rhythm" panel is a 24-segment strip, one segment per hour of the day, shaded by how many tokens you used in that hour. Hover or tap a segment for that hour's token count and its percentage of the day's total.
Below the strip, a large readout states what percent of your tokens land after 6pm, followed by a line noting that the taller (more saturated) the band, the heavier that hour was.
Model mix
The "Model mix" panel shows a single stacked bar representing 100% of the period's spend, split by model. Underneath, a legend lists each model with its provider logo, token volume, and percentage share of spend.
A note below the legend calls out what share of tokens came from prompt cache — cached tokens are billed at roughly one-tenth the normal rate — and, when relevant, what share of spend went to work that didn't end up merging ("exploration").
Daily consumption chart
The full-width "Daily consumption" panel is a stacked bar chart with one bar per day; each bar's segments are colored by model, matching the legend in the panel header. Hovering a bar breaks that day down by model and shows the day's total.
A caption under the chart is explicit about what these figures represent: what you ran each day, valued at today's public API rates — independent of your flat subscription, if you have one.
The footnotes
Three cards close out the page:
Exploration — the percent of tokens spent on work that didn't merge, with a short progress bar. This is normal, especially on greenfield work.
Cache savings — the percent of all tokens this period served from prompt cache at roughly one-tenth price; these tokens are excluded from the spend charts above.
Cost / merged PR — the same per-PR figure from the KPI strip, shown again alongside the equivalent API-rate cost for comparison, when available.
A final line at the very bottom of the page states what the whole page is benchmarked against: public API rates, and the hourly engineering rate used to price your engineer-hours.
Adjusting the calculations
Both the Statement hero and the line at the bottom of the page include an Adjust calculations link. It takes you to Preferences — the place to change the assumptions this page is built on, such as the hourly rate used to price your engineer-hours, so the value figures reflect your own numbers rather than ATTRIBUT's defaults.