Tally — Features

Operational

Track this month.

Five surfaces handle the daily and monthly view. The screenshots below are from the demo persona's May 2026 — a Melbourne household with two mortgages, $11,718 spent of a $15,915 budget, three days left in the month.

Dashboard

One page, the whole month. Net worth, this-month vs last-month spend, cashflow, discretionary composition, recent activity, and the next seven days of upcoming bills.

This / Last month toggle across every card
Discretionary composition — pie of just the spend that responds to choices, ignoring the fixed costs that don't move
Pace + projected end-of-month on the budget card
Register summary — open items across Comments / Budget / Actions
Two layouts: Cockpit (default) and Stream (single-column)

Transactions + categories + AI

85+ pre-seeded categories tuned for Australian households. Every leaf carries four independent axes, so the same row reads four different ways.

Claude Haiku categorisation on ingest — every new transaction gets categorised inside three seconds
Confidence threshold < 0.7 lands in the Review queue, never silently mis-tagged
Rules engine with priority + hit-count tracking
Find similar — match a tx pattern to bulk-recategorise
Optional property tag — for households with an investment property; ignorable for everyone else

Four axes per leaf: parent · funding source · kind · essential.

Budget grid

Budgets in Tally are caps with variance, not contracts with grades. Each row shows budget, spent, remaining, variance, pace, and any operational flags (fast burn, essential, chronic over).

Monthly

Per-category budget vs spent
Inline-editable
Trailing 3-month avg shown as reference

Subscriptions

Recurring charges, annual run-rate
Snooze / dismiss states

Annual expenses

Year-ahead view of quarterly + annual bills
Grouped by predicted next month

Recurring detector + Bills

A daily cron walks the transaction history and identifies any merchant that's appeared 3+ times at a stable cadence. No user setup. The Bills page then shows what's coming in the next 30 / 60 / 90 days.

Cadence detection — weekly / monthly / quarterly / annual / irregular
Drift signal — surfaces when a subscription quietly went up
User curation — snooze, ignore, edit cadence; user-set fields override the detector
Confidence shown — values below 0.7 carry a "verify before relying on it" disclaimer

Cash flow + Sankey

Every dollar, from where it came to where it went. The Sankey is the headline visual — read it as a flow, not a pie. Click any node to drill into the transactions behind it.

Sankey diagram — full income → category routing
Moorr funding-source view — same numbers, advisor side
Subscription audit embedded — annualised cost of every recurring charge
Period toggle — this month, last month, FY-to-date, custom range

Macro

Plan the decade.

Five surfaces handle the long horizon. A cash-position forecast (Moorr-style), a loan engine with stacked scenarios, monthly net-worth snapshots, Moorr funding-bucket mapping, and per-property reporting. The numbers used across the screenshots below are the demo persona's: a Brunswick PPOR ($620k mortgage at 6.10%) and a Bendigo IP ($410k IO loan at 6.35%).

Forecast

Paste your Moorr cashflow projection once. The chart overlays it against the actual cash position, month by month. Horizons range from one quarter out to 2029.

Moorr cashflow overlay — projection line + actual line on the same axis
Multiple horizons — quarter / 12 mo / 24 mo / 60 mo / to 2029
Per-account breakdown on hover — which account drives the trajectory
Snapshot capture — pin today's bank balances to the curve manually

Forecast widget on the Dashboard. The full Forecast page is a richer version of the same.

Loans + scenarios

P&I, IO with transition, offset, lump sums, named scenarios — every shape an Australian mortgage takes. The IP loan in the screenshot is $410,000 at 6.35% interest-only for 5 years then reverting to P&I. Adding a +0.75% rate scenario shifts lifetime interest by ~$65,000.

Full AU repayment shapes — P&I or IO with transition, weekly / fortnightly / monthly, true-fortnightly vs half-monthly
Lump-sum payments — model one-off extra principal at specific dates (tax refunds, bonuses)
Offset modelling — interest accrues on (balance − offset); daily cron snapshots the offset balance from a linked savings account
Named scenarios — "Rate +0.25%", "Extra $200/fortnight", "Switch to P&I from year 5". Stack as many as the question needs.
Drift signal — link a loan to its bank account; the worker surfaces ahead-of-plan or behind-of-plan against the schedule
Three chart modes — balance over time, interest vs principal per period, cumulative interest

Net worth flow

Monthly account snapshots across every linked + manual account. The persona below sits at $446,600 net worth — $1.5M in assets (Brunswick PPOR, Bendigo IP, CommSec, cash, crypto) against $1.06M in liabilities (two mortgages, a car loan, a credit card).

30 / 90 / 365-day horizons
Asset / liability breakdown grouped by subtype (transaction, savings, investment, mortgage, loan, crypto, manual)
Manual valuations — anchor a value on a date; the running sum walks from there
Per-account 30-day delta — quick glance at where the line moved most

Moorr alignment

Tally complements Moorr — Stuart Wemyss's household-finance methodology. The plan lives in Moorr; the actuals live in Tally. Every category carries its Moorr funding bucket as a first-class column.

Every leaf category mapped to one of five Moorr funding buckets
"Group by Moorr" toggle on Budget Monthly — same numbers, advisor view
Forecast tab overlays the Moorr cashflow projection against the actual cash position
Insights surface Fixed / Variable / Discretionary breakdown — what the buckets should be sized for

Every category maps to one of five Moorr buckets.

Per-property reporting optional

Households with an investment property can tag transactions, bills, and loans to that property. The Transactions filter pulls a property-scoped P&L without leaving the app — net rental income, deductible expenses, depreciable items. Households without an IP can ignore the surface entirely.

Per-property income + expense breakdown on Transactions, Budget Monthly, BudgetTrends
System-tagged categories that mirror Housing for separate ATO reporting (Rental income, Landlord insurance, Strata, Loan interest)
Recurring transactions inherit the property tag — set once, future occurrences carry it forward

Filter set to a property on the Transactions page; every other page picks up the filter via the URL.

Live data

Australian Open Banking.

Tally receives transactions via Fiskil — an Australian CDR-accredited Data Recipient. CDR (Consumer Data Right) is the Treasury-regulated framework that replaced screen-scraping for bank data in 2020. Bank credentials never touch Tally or Fiskil — you authenticate at the bank itself, the bank confirms what's allowed, and Fiskil receives only the data you've authorised.

No passwords on Tally — bank authentication happens at the bank's own portal. Required by CDR.
Per-account consent — you choose which accounts to share. Revoke any time from your bank's CDR dashboard.
Webhook delivery — transactions arrive within minutes of clearing. No daily polling, no manual refresh.
HMAC-signed payloads — every webhook is signed by Fiskil; Tally verifies the signature before processing. Replay attacks rejected by timestamp window.
12-month CDR consent windows — by regulation. Tally prompts you to re-authorise before expiry.
Raw payload archive — every webhook payload kept in encrypted R2 for 90 days for forensics and reproducibility.

Banks + brokerage supported

All four major Australian banks plus most mid-tier ADIs. Brokerage coverage is narrower — Treasury's CDR rollout to investment products is still in progress — but the most- asked-for platforms are already supported.

NAB CBA Westpac ANZ ING Macquarie HSBC Bendigo Suncorp Bankwest St George UBank Up Bank Australia ME Bank CommSec SelfWealth Stake + most CDR-enrolled ADIs

Sharing

Built for couples. Three access levels via the admin's Settings → Users.

None

Their own empty data silo. Default for new users.

Full

See and edit the data — transactions, categories, loans, register, everything. For partners.

View only

See the data, but the worker blocks every non-GET request. Useful during focused work.

Each user signs in with their own Google account; the audit log shows who
One-click sign out ends the session within seconds, ahead of JWT expiry
Per-user activity log — every login attempt recorded with IP, device, outcome

Security & privacy

No passwords on Tally. Sign-in is delegated to Google OIDC — whatever 2-step verification Google asks for is what protects the account.
1-hour session JWT. Even a stolen device's cookie self-destructs within the hour.
Allowlist-only. No self-signup; the admin invites users explicitly.
Per-user audit log — every sign-in attempt recorded with IP, device, outcome. Force sign-out any session from one click.

Under the hood

Built on Cloudflare's edge stack so it's fast everywhere and cheap to run.

Tech stack details

Cloudflare Workers (Hono framework) — sub-100ms API responses, globally distributed
D1 — SQLite at the edge, ACID transactions
R2 — webhook archive (raw payloads kept 90 days) + receipt attachments
KV — aggregate cache + JWKS cache
Anthropic Claude — Haiku for categorisation, Sonnet for insights
React 18 + Vite + Tailwind + Radix on the frontend
Drizzle ORM — typed queries, portable if we ever outgrow D1

What Tally does.