—Built in public
Everything we ship, the moment we ship it. No roadmap teasers — just what landed in production this week.
DataBridge now checks for domains that accept unknown mailboxes and surfaces them as "Needs review · Catch-all domain" instead of treating them as invalid. Catch-all addresses are excluded from the fully-clean export by default, visible in dataset KPIs, and shown in Resolve even when the count is zero so you can tell the check actually ran.
End-to-end audit of the eight-validator pipeline ahead of public launch. Four real bugs caught and fixed: (1) null-MX domains (RFC 7505 — example.com and any corporate apex that explicitly disables mail) were being marked deliverable; now correctly flagged as unreachable. (2) Typo suggestions fired against short legitimate domains — x.com → me.com, b.com → me.com, etc — because the Damerau-Levenshtein budget wasn't length-aware. Short domains now require distance 1, longer ones still allow 2. (3) The .zip and .mov TLDs (Google 2023 release, immediate phishing favourites) were missing from the risky-TLD set; added. (4) Homograph attacks (Cyrillic 'аpple.com', Greek 'αpple.com') were slipping through as deliverable. Non-ASCII domain detection added — real IDNs use Punycode and remain accepted.
The Audit tab used to show only the upload event because the live flows weren't writing rows. Now: every applied fix (Remove Invalid / Duplicates / Role / etc), every revert, every download, and a 'Processing complete' event written by the worker the moment it flips a dataset to ready — all surface in chronological order with the affected row count. Revert events also include the field being restored (e.g. 'Revert · Duplicates · Restored 3 rows back to pending') so a compliance audit can reconstruct exactly what changed.
Free users now see an emerald 'Upgrade →' pill in the header on every dashboard route, plus the 'Upgrade to apply all fixes' button on dataset details routes straight to /checkout?plan=pro. Both used to detour through /dashboard/billing or a paywall modal — now it's one tap to the actual upgrade form. Paid users keep the plan badge that links to billing (where managing seats and invoices is what they actually need).
Gmail iOS and Android auto-invert light-themed transactional emails — the previous welcome / confirm / dataset-ready templates ended up with gray-on-gray CTAs and mangled accent colors. Each template now ships an explicit dark variant (via prefers-color-scheme + Outlook's [data-ogsc] mirror), and the CTAs are bulletproofed with HTML bgcolor + <font color> legacy tags so the label stays readable even when the client ignores CSS entirely. The user's email shown in the confirm body also got pinned to brand emerald instead of the auto-blue Gmail / Apple Mail apply to anything that looks like an address.
The full feature × tier grid and the vendor comparison table both required ~340px of horizontal scroll on iPhone SE / 320-class phones. Phones now get a stacked card-per-tier layout (Free / Pro / Agency, each with their feature checklist grouped by section) and a price-punchline card for the vendor compare ($29 DataBridge vs $80–$390 alternatives). The wide grids stay on tablets and up where they have room to breathe.
Two endpoints, one Bearer token. POST /api/v1/clean returns every row with a verdict + reason (audit-friendly); POST /api/v1/clean/filter actually drops the rows you don't want — pass remove_duplicate, remove_role, remove_disposable, etc., or use ?keep_only=deliverable as a shortcut. Output preserves your original columns so it's import-ready for Mailchimp, HubSpot, Klaviyo. Full reference at /docs/api.
GET /api/v1/datasets lists every run with cursor pagination + a source filter (api / dashboard). GET /api/v1/datasets/{id} pulls a single run's summary. Use the dataset_id returned from /clean to look up runs later from your CRM or audit log — no re-upload needed.
Every successful call returns X-DataBridge-Quota-Used / -Remaining + X-RateLimit-Remaining-Minute / -Hour headers (and a `quota` block in JSON mode), so consumers can pace themselves without a follow-up call to /api/usage. The 402 QUOTA_EXCEEDED carries the same block — failure responses are useful too.
By default oversize CSVs return 413 with a hint. With ?truncate=1 we process the first N rows your plan allows and surface a `truncation` block (original_rows / processed_rows / dropped_rows) so the caller knows what was clipped. The persisted dataset carries the truncated flag too.
Every /api/v1/clean and /api/v1/clean/filter call now persists as a dataset under /dashboard/datasets — same as a manual upload, with a small `</> API` badge so you can tell apart programmatic runs from CSV uploads. Filter the list by source (All / Dashboard / API) on the same page.
Pro stays at $29/mo locked in for life if you sign up before the cohort closes. Live countdown on the pricing page so you can see exactly how long the offer holds.
Switch the toggle on /pricing to see Pro at $23/mo (billed yearly) and Agency at $63/mo. Same features, two months free.
Pricing page now ships with a full feature × tier breakdown — Limits, Validation, Tools, Performance, Support — so "which plan is right for me?" is a 5-second question instead of a 5-minute compare-cards exercise.
One entry point for support: Aurora's chat. Bug reports, feature requests, and direct-to-human email all live behind a single "Human" button inside the chat — no more split between two modals. Free plan keeps full access.
End-to-end visual rebuild: cream surfaces, emerald accents, editorial typography, mint hairlines on every card. The landing, dashboard, dataset detail, billing, settings, legal, and login all share one design system now.
New 3-step flow on every dataset (Insights → Typos → Download) that walks you from raw upload to clean export without leaving the page. Score ring shows tonal feedback (mint / amber / rust) instead of always-green at 30%.
Compare and Column Mapper now live as tabs on the dataset detail page and operate on your resolved data automatically. Apply fixes once, every downstream tool inherits the cleaned subset.
Inline 1-email validator now collapses the verdict + 6 checks into a single row so the primary CTA never gets pushed below the fold on 13" laptops. Live activity pill ticks up addresses cleaned in the last 24h.
Toggling Processed off no longer leaves a phantom line at the top of the chart. Y-axis now rounds up so a 65k peak is fully visible instead of clipping at the 60k tick.
Activity card delta arrows used to escape the card border when totals hit 6 digits. Stack rebuilt: label + delta on top, full localized number below — and 10M+ values compact gracefully to "10M".
Validate → fix preview → apply, end-to-end. Detects unclosed quotes, trailing commas, mixed encodings, and malformed rows; offers a one-click repair before processing. Free users see the diagnostic, Pro+ apply the fix in place.
Every unique domain in your CSV is now verified with a live DNS MX lookup before processing. Catches typo'd and parked domains that regex-only checkers miss — surfaced as a dedicated "Unreachable domains" insight.
New /login with ?next= preservation across the Google OAuth round-trip. Already-registered detection handles both Supabase flavors with a one-click "Sign in instead" path.
Rows processed before signup now merge into the user's usage row on login (previously got stranded as orphan rows the sidebar couldn't read).
Detected common address typos (gmial.com → gmail.com, outlok.com → outlook.com) are surfaced in-report with one-click apply or remove — per-row decisions persist across reloads.
Pro feature: one-click column renaming for Mailchimp, Klaviyo, Beehiiv, Kit, HubSpot, ConvertKit and ActiveCampaign. Works on any CSV, no pre-cleanup needed.
Upload two CSVs and instantly see which contacts are in both, only in the first, or only in the second. Export each segment with exactly the columns you want.
Want to shape what ships next? Email Damian — feature requests get read within a day.