LeadLine by Fusion Data Company
Send an applicant or a web lead a link. They talk for five minutes. You get a scorecard with a score on every question you asked, the exact line they said it in, and one sentence explaining why it is not the score above it.
Not a receptionist. It does not answer your phone. It screens the people you send it.
5
minutes on the phone
The length of one screen, start to finish.
60s
from form fill to link sent
The trigger fires the invite; nobody queues it.
100%
of scores carry their quote
Stored with the score, not regenerated later.
0
apps to install
The applicant taps a link in whatever browser they have.
Who it is for
How it works
Five of them, typed and weighted, in the words you would use yourself. Mark the ones that are must-haves.
From the board, from your form, from your ATS, or from a Zapier zap through the API. One POST is enough.
Five minutes in a browser. No app, no account, no scheduling. They can type their answers instead if they would rather.
Every question, every score, the quote behind it, and the recording - in your inbox, your CRM and the board.
The output, before you call anything
Marisol Reyes is invented and so is the depot. The scorecard is not: it is the real layout, the real fields, and the 82 out of 100 is recomputed on this page from the five scores and their weights by the same arithmetic the scorer runs.
The headline feature
The common complaint about AI interviewers is not that the score is wrong. It is that nobody can say where it came from, so it cannot be used to make a decision anyone would have to explain later. Here is the whole mechanism, on the sales page, before you pay for anything.
A yes, a no and a number between one and five are read straight out of the transcript by a rule, at 0.98 confidence, with no model in the loop at all. On a five question set that is usually two or three of the answers decided deterministically.
Open questions go to a model, and it has to return the answer, the score and the exact quote it read them from. No quote, no score. A rule is never overwritten by a less certain model answer.
Scoring one interview twice replaces the previous scores rather than adding to them, so a re-run is idempotent and the row you looked at yesterday is the row you look at today.
What the owner does on Monday
Reads seven scorecards over a coffee and calls two people. That is the entire job the product has.
Monday morning
Seven people talked to the screen between Friday evening and Sunday afternoon. Two are worth a call, one asked for a person and got a booked callback, one failed a must-have and is capped, and one hung up at forty-seven seconds - which is not a screen, is not counted and is not billed.
| Applicant | Role | Screened | Length | Score | Grade | Outcome | What the scorecard says |
|---|---|---|---|---|---|---|---|
| Marisol Reyes | Class B driver | Fri 7:43pm | 4:44 | 82 | A | Booked | Takes the 9:15 Tuesday slot |
| Andre Boothe | Class B driver | Fri 9:02pm | 5:11 | 76 | B | Screened | Licence 3 months, no air brakes |
| Yesenia Cruz | Warehouse lead | Sat 8:11am | 4:28 | 74 | B | Callback | Screened in Spanish, asked for a person |
| Devon Pike | Class B driver | Sat 10:47am | 3:12 | 58 | C | Screened | Vague on the incident question |
| Tobias Lund | Warehouse lead | Sat 1:19pm | 4:00 | 55 | C | Screened | No weekend availability |
| Rey Alcantara | Class B driver | Sun 9:30am | 1:36 | 22 | D | Must-have failed | Cannot work for any employer - capped |
| Halle Osei | Warehouse lead | Sun 2:08pm | 0:47 | - | D | Not a screen | Hung up at 47s - not counted, not billed |
Rules read the yes, the no and the number straight out of the transcript. Only what is left goes to a model, and it has to hand back the line the answer came from. You can always see why somebody got a B.
Type the questions in your own words. Mark each one open, one to five, yes or no, or must-have. Weight the ones that matter. A failed must-have caps the grade no matter how well the rest went.
A signed webhook, a Zapier-shaped payload, a CSV with every answer in its own column, and an email you can forward. Field mappings for HubSpot and Greenhouse are in the docs.
Set the hours you take calls. The interviewer only offers times that are actually open on your calendar, and the confirmation carries a calendar invite.
Set the language on the question set and the whole screen switches: the questions, the email, the page. Anyone who says "talk to a person" gets a callback booked instead, in either language.
POST a name and an email with your key and that person has a link. Wire it to your application form, your ATS, or a Zapier zap in an afternoon.
Pricing
A screen is one completed interview with a scorecard at the end. A call that never connects is not a screen and is not counted. Every plan starts with a 14-day free trial.
One role or one form. About one screen a working day.
$5.96 per screen at the included volume
Hiring for a few roles at once, or screening every web lead that comes in.
$3.99 per screen at the included volume
High-volume hourly hiring, or a sales team that calls every inbound lead.
$3.00 per screen at the included volume
Included in every plan
Secure checkout by Stripe. Nothing is charged until the trial ends. Cancel any time.
Screens past your monthly allowance are added to your next invoice by a person, never charged automatically without telling you.
Open the interview board immediately. Explore fictional candidates, question scores and the quotes behind them. No signup, no card.
