Not a metaphor. This is the actual mechanism.

Why stop looking
after one place?

Every lead gets checked against the company site, LinkedIn, and recent activity — then the contact gets verified, retried, and only pattern-guessed as a last resort. Nothing ships because we stopped looking early.

Find verified email
Fallback chain
First tryCompany site scrape
SkippedLinkedIn lookup
SkippedPattern fallback
Verification
CheckedDeliverable ✓
€299/mo — one flat rate 3-source fallback, extendable with your own provider keys Every email verified before it ships ~€0.006 marginal cost per lead

Why this exists

Clay is good at this. It's just built for a company that already has a RevOps hire.

Clay's waterfall is a genuinely good idea — try one data provider, fall back to the next, keep the first hit. But you're renting a spreadsheet-grid canvas, a data marketplace, and a credits meter to get there, and the meter runs separately for actions and for the data itself. Small teams outgrow the entry tier fast, and the learning curve is real — RevOps admins call it a 40-hour rabbit hole before it's "somewhat functional."

MagFlow AI does the part that actually matters — find the person, verify the contact, explain why they're worth reaching — without the configuration surface. You get a form, not a grid.

See the full comparison →
01 / scrape

Three sources, not fifty vendors

Company site, LinkedIn, and a recent-activity search — routed through a bypass-proxy layer so a Cloudflare wall doesn't end your list at row twelve.

  • No data marketplace to shop across
  • Fixed pipeline — cost per lead stays predictable
  • Built for markets Clay's US-heavy data thins out on — UK, Ireland, EU SMBs
02 / corroborate

Confirmed by a second source, or labeled as unconfirmed

A waterfall stops the moment any one source answers. We check whether a second, independent source says the same thing before a field is shown as high-confidence — a name that only appears once stays labeled that way, visibly, instead of being presented the same as one two sources agree on.

Corroboration check
Sources checked
site"Priya Chandran" found
linkedinconfirms — 2 sources ✓
activityno mention
03 / verify

A bad email doesn't ship silently

Every extracted address is checked. A miss triggers a retry, then a pattern fallback — labeled as exactly that, never presented as a fact you'd have to discover was wrong mid-campaign. Re-verify the whole list again, free, right before you export — email validity decays, so a check from three weeks ago isn't good enough.

  • verified / likely / pattern-guess / not-found — never blended together
  • Free re-verification pass before export
  • Click any field to see the exact source sentence it came from
04 / personalize

A landing page that already knows their name

Every lead gets a minimal page built from what we actually found — their company, their logo, the signal that makes this worth opening. No video, no PDF, no generated media — those cost real money per lead and we'd rather that cost go into the accuracy work above.

  • Logo pulled live via Logo.dev
  • Copy assembled from corroborated fields, not freshly written per lead
  • See it live in the product tour →
05 / send

Wherever you already send from

Export to CSV, or connect your own Smartlead / Instantly account and push leads straight into a live campaign. Your account, your sending reputation — we never touch your inbox.

  • Universal CSV / webhook export
  • Optional direct push (Smartlead, Instantly)
  • Nothing here requires you to change how you send

The actual cap

You never pay for an empty row.

MagFlow AI includes 1,500 leads a month. A dead end — no email, no signal, genuinely nothing found — doesn't count against that number. You're only using cap against leads that actually produced something, which is the direct fix for the exact complaint Clay's own users have about paying for lookups that come back empty.

The trade you're actually making

A flat number instead of a meter you have to watch.

no credits

One price, every month

€299/mo covers the pipeline. No separate action credits and data credits ticking down at different rates.

no grid

A form, not a canvas

Set your target once. No column-mapping, no conditional logic to debug before your first list comes back.

no lock-in

Leaves how it arrived

CSV out, always. If you plug into Smartlead or Instantly, that's your account — not ours to hold hostage.

Field notes

We're writing down what we find as we build this in public.

Reddit threads on Clay complaints, what a "self-healing" email waterfall actually costs to run, why UK and Ireland outbound gets worse data than US outbound. New note roughly every other week — drop your email and we'll send the next one.