Scattered candidate data
Candidate information can live across resumes, spreadsheets, inboxes and separate recruiting tools.
In active development · built module by module
Find the Right People Faster with AI
From screening to hiring, Scoreboad helps you build high-performing teams with the power of AI.
Screen candidates, manage interviews, evaluate applicants, and keep your hiring workflow connected from one workspace.
Overview
Last 30 daysNew candidates124
Screenings completed38
Interviews today4
Overdue applications3
Hiring pipeline
Senior Backend Engineer- Applied124
- Shortlisted46
- AI Screening Call28
- Video Interview9
- Hired2
The challenge
Great candidates shouldn't get lost in a fragmented hiring process.
Sourcing, resumes, screening, interviews, evaluation and candidate communication each tend to end up in a different place. The work of hiring then becomes the work of keeping those places in sync.
Repetitive screening
Reading every application and working out which ones fit the role is slow, and it is the same work every time.
Interview coordination
Scheduling rounds, sending invites and keeping candidate conversations organised adds real overhead.
Inconsistent evaluation
Feedback is hard to compare when each interviewer records it somewhere different, in a different shape.
One process, many tools
A single hire can touch a job board, an inbox, a spreadsheet, a calendar and a chat thread before anyone decides.
Six parts of one hire — resumes, applications, screening answers, interview times, interviewer feedback and the decision itself — each sitting in a different tool.
The solution
One connected workspace for modern hiring.
Scoreboad brings candidate screening, interviews, evaluation and the hiring workflow into one place, and uses AI to take the repetitive reading and summarising off your team — so the time goes into the decision instead of the admin around it.
Scoreboad — one workspace
- Resumesin Candidate record
- Applicationsin Applications
- Screening answersin Screening report
- Interview timesin Interviews
- Interviewer feedbackin Evaluation
- The decisionin Pipeline
The Scoreboad platform
Everything you need to manage hiring in one workspace.
Jobs, candidates, AI screening, interviews, evaluation and the hiring pipeline are six views of the same records — not six tools that have to be kept in step. Pick one to see it.
Open a role with its own hiring stages, requirements and screening questions.
Jobs
4 open- Senior Backend EngineerRemote · Full-time
- Product DesignerBengaluru · Full-time
- Data AnalystRemote · Contract
- Support LeadPune · Full-time
Candidates
248 on file- A. SharmaBackend Engineer · 7 yrs
- M. IyerPlatform Engineer · 5 yrs
- R. NairProduct Designer · 6 yrs
- S. BoseData Analyst · 3 yrs
Resume match
A. Sharma · Senior Backend Engineer- Match scoreRequirements met, weighted by the job
- Strong matchesGo, PostgreSQL, distributed systems at scale
- GapsNo Kubernetes experience stated on the resume
- Needs verificationTeam-lead scope is claimed but not evidenced
Interviews
This week- A. SharmaVideo · Tomorrow, 10:00
- M. IyerPhone · Yesterday, 15:30
- R. NairOn-site · Thursday, 11:00
- S. BoseVideo · Monday, 09:00
Evaluation
A. Sharma · Senior Backend Engineer- Resume matchScored against the job's requirements
- AI Screening CallStructured report from the call
- Video InterviewPanel feedback, two interviewers
- VerdictA person decides — the panel only gathers
Hiring pipeline
Senior Backend Engineer- Applied124 in stage
- Shortlisted46 in stage
- AI Screening Call28 in stage
- Video Interview9 in stage
The flow
From application to hire.
One path through the process, with every stage writing to the same records.
Create a job
Requirements, stages and screening questions in one place.
Find candidates
Bulk resume intake, a public application form, or added by hand.
AI screening
Resumes read, structured and scored against the role.
Interview
Rounds scheduled, invites sent, feedback captured.
Evaluate
Every signal on one panel, for a person to weigh.
Hire
Offer, onboarding checklist, documents collected.
AI-powered recruitment
Turn candidate information into hiring insight.
Scoreboad reads candidate information, checks it against what the role actually requires and surfaces the signals worth a recruiter's attention — so the repetitive part of candidate screening stops being a person's job.
Resume intelligence
Resumes are read and structured automatically, then held for review before anything is written to a candidate.
Candidate screening
Requirements checked in code, judgement calls left to the model, and one score that cannot contradict either.
AI evaluation
One paragraph on where an application stands, derived strictly from its own recorded fields.
Recruiter review
Every AI output lands on a screen before it counts. No stage moves itself.
AI screening calls
Let every candidate have a structured first conversation.
Scoreboad can run the first-round call for you — the questions your team wrote, asked the same way every time, captured as a transcript and returned as a structured report against the application. A recruiter reads it and decides what happens next.
- Start
- Connect
- Ask
- Respond
- Understand
- Review
A screening call is queued against one application. Calling is disconnected until somebody turns it on, and every attempt is counted against a cap — so a candidate cannot be dialled repeatedly by a loop nobody is watching.
The first thing the candidate hears is that the call is automated and may be recorded, and that they can stop it. That line is built by the same function for every call and checked before the provider is contacted; there is no setting that removes it.
The questions are the ones your team wrote on the job — not questions a model invented on the call. Every role carries its own list, asked in the order you set.
Answers are captured as a transcript against the application, with the recording kept beside it. Nothing is treated as usable until the consent flag on the call is true — a database rule, not a code convention.
The conversation becomes a structured report: interest, expected pay, notice period, whether the location works. Anything the model was unsure of is flagged as uncertain rather than guessed at.
A recruiter reads it and corrects what is wrong. What the AI originally said is kept unchanged alongside, so the edit is always visible — and the candidate was told on the call that a person would be reviewing it.
Attempt 1 of 3
TranscriptIllustrative
- AssistantHello, am I speaking with A. Sharma? I'm an automated assistant calling on behalf of Northwind about your application for the Senior Backend Engineer role. Before we start: this is an automated call and it may be recorded so a recruiter can review it. If you'd rather not continue, just say so and I'll end the call and arrange for a person to contact you instead.
- CandidateThat's fine, go ahead.
- AssistantThank you. This will take about five minutes. I'll ask a few short questions about your experience and availability.
- AssistantWhat is your current notice period, and when could you realistically start?
- CandidateI'm on thirty days, and I could probably start the first week after that.
- AssistantThe role is remote with occasional travel to Pune. Does that work for you?
- CandidateRemote is what I'm looking for. Travel now and then is no problem.
- AssistantThat's everything — thank you for your time. A recruiter from Northwind will review this and be in touch about next steps.
Screening report
- Interest level
- High
- Notice period
- 30 days
- Location accepted
- Accepted
- Expected CTC
- Not stated on the callFlagged uncertain
- Availability
- From the first week after noticeAI said “Immediately available” · corrected by R. Menon
Reviewed by R. Menon · the candidate was told on the call that a recruiter would read this.
Questions you wrote
Each job carries its own screening questions, asked in the order you set. Up to twelve per role.
Transcript and recording
Every call is captured against the application, and nothing is usable until the candidate has been told and has agreed to carry on.
Reports your recruiter can correct
Interest, notice, pay and location come back as fields. What the AI said is kept beside what your recruiter changed it to.
Human + AI
AI does the work.You make the decisions.
Scoreboad helps recruiters move faster without removing human judgment from the hiring process.
AI
- AI analyzesReads resumes, scores matches, summarizes calls.
- AI organizesKeeps the pipeline current and surfaces what is stuck.
- AI assistsDrafts messages and briefs inside approved wording.
You
- You reviewEvery AI output lands on a screen before it counts.
- You decideStage moves, offers and rejections are yours alone.
- You hireThe judgement stays with the person making it.
Candidate workspace
Everything about a candidate, in one place.
One record per person, carrying every application they hold — the parsed resume, the screening report, each interview round and the evaluation that draws on all three. One candidate, one complete picture.
A. SharmaBackend Engineer
- Stage
- Applied
Activity
- AppliedCareer page
- Resume parsedAI
- Parsed fields reviewedR. Menon
- Match calculatedCode + AI
- Screening report reviewedR. Menon
- Interview scheduledR. Menon
- Interview feedback submitted2 interviewers
- Stage changedR. Menon
Profile
- Current role
- Backend Engineer
- Current company
- Northwind Labs
- Location
- Pune, IN
- Total experience
- 7 years
- Expected salary
- Not stated
- Notice period
- 30 days
Two applications on file. The record is the person, not the application.
Resume
- Resume
- sharma-backend.pdf · parsed
- Skills found
- Go · PostgreSQL · Kafka · AWS
- Experience
- 3 roles extracted
- Proposed change
- Notice period 60 → 30 days
Parsed fields are proposed as a diff. A person applies them.
Screening
- Call
- Completed · 4m 51s
- Interest level
- High
- Location accepted
- Accepted
- Expected CTC
- Flagged uncertain
Report reviewed by R. Menon. The AI's original answers are kept alongside.
Interviews
- Round 1
- Video · Completed
- Feedback
- 2 of 2 submitted
- Round 2
- Director Round · Scheduled
- Brief
- Generated from the resume and round 1
Feedback is captured in a fixed shape, so two interviewers can be compared.
Evaluation
- Resume match
- 91 / 100
- Screening
- High interest, notice confirmed
- Interview
- Strong on systems, light on Kubernetes
- Verdict
- Needs Review — set by R. Menon
Three sources on one panel. The verdict is recorded against a person.
Activity
- Applied
- 12 days ago
- Resume parsed
- 12 days ago
- Parsed fields reviewed
- 11 days ago
- Screening report reviewed
- 6 days ago
Append-only. Every entry names who did it and cannot be edited afterwards.
One record per person, not one per application. Contact details, current role, experience, expected salary and notice period live here — and stay here when the same person applies for a second job.
Resumes are parsed into structured fields, and the proposed values are shown as a diff against what is already on the record. A person applies them; nothing is written to a candidate by a model alone.
The screening call's report lands on the application: interest, notice, expected pay, whether the location works. What the AI said is kept beside what a recruiter corrected it to.
Rounds, their mode and their outcome, with structured feedback captured per interviewer — so two people's opinions can actually be compared rather than read.
Every signal on one panel: the resume match, the screening result and the interview feedback, with the strengths and concerns each contributed. The verdict is a person's.
An append-only record of what happened and who did it. Every line below is a real event type this product writes — it is the audit log, not a marketing summary of one.
Candidate profiles
One record per person, deduplicated on contact details, carrying every application they have ever held with you.
Resume intelligence
Resumes parsed into fields and proposed as a diff, so what changes on a record is always something a person approved.
Candidate evaluation
Match score, screening result and interview feedback on one panel, with the strengths and concerns each one contributed.
Activity history
An append-only log of every change, naming who made it. It cannot be edited after the fact, by anyone.
Hiring pipeline
See every candidate move from application to decision.
One board per job, with every application on it and every card aged against the target you set for its stage — so what has stalled is visible before somebody asks.
Senior Backend Engineer
2-day target
3-day target
- M. IyerPlatform Engineer1 day in stage
3-day target
- K. ReddyBackend EngineerDay 3 of 3
7-day target
- S. BoseData Analyst2 days in stage
7-day target
- R. NairProduct Designer4 days over
- P. DasBackend Engineer6 days over
No target — terminal stage
31 rejected · 6 withdrawnTerminal outcomes are counted here, not given columns.
- Applied
- Shortlisted
- Screening
- Aging
- Moved
- Hired
Every application lands in the same first column, however it arrived — career page, bulk upload, referral or added by hand.
A resume that clears the job's passing mark moves to Shortlisted. One that does not is flagged rather than deleted, and the flag can be cleared.
An automation can start the screening call when someone reaches this stage. `move_to_stage` and `start_screening_call` are real actions — the board moves on its own only where you have said it should.
Every card is aged against its stage's target — two days in Applied, three in Shortlisted, a week in the interview rounds. Past three quarters of the target a card is at risk; past the target it has breached.
A person moves the card, from a menu on the card itself. There is no drag and drop, deliberately: a dropdown works with a keyboard and on a phone, and it cannot fire from a mis-drag.
Hired is a stage; rejected and withdrawn are exits, counted at the foot of the board rather than given columns. A board full of finished work stops being a work queue.
Stage SLAs
A target per stage, with a card marked at risk at three quarters of it and breached past it. Defaults apply until you change them.
Deliberate moves
Stage changes come from a menu on the card, so they work with a keyboard, work on a phone, and cannot happen by accident.
Automations you set
Rules can move a card, start a screening call or notify a recruiter — and an approval step where the action is one you would want to see first.
Candidate applications
Make it easy for candidates to enter your hiring workflow.
Every job gets a shareable application form. What a candidate types is saved before anything else happens, becomes a real candidate record, and what the resume adds to it waits for a person to approve.
Application
Senior Backend Engineer
Shared as a link · no account needed
An illustration of the real application form. It is read-only and submits nothing.
In Scoreboad
What exists now
In the order the submission creates it
- Form responseSaved — 6 answers
- Resume filesharma-backend.pdf · filed
- CandidateNew record — no duplicate found
- ApplicationSenior Backend Engineer · Applied
- Resume parseSkills, 3 roles, education
- Proposed changeAwaiting review — nothing written yet
A recruiter reviews the proposal and decides what lands on the record.
- Apply
- Resume
- Saved
- Created
- Parsed
- Proposed
Every job gets its own application form, created with the same thirteen fields and shared as a link. No account, no login — a candidate opens it and types.
A resume is required, and it is the only file the form accepts. Custom questions can be added to any form; a custom FILE upload cannot, because a form that collects someone's documents and drops them is worse than one that never offered.
The answers are written down before anything that can fail — before the file is stored, before a model is called. Someone has typed for five minutes on a phone; a failure after this point loses a link, never their submission.
The candidate and the application are created from what they typed, and matched against everyone already on file by normalised email and phone. This happens whether or not the resume could be read.
Then the resume is parsed. A scanned PDF with no text layer, a provider outage, a rate limit — all normal, and none of them fail the application. The file is simply filed unparsed.
What the parse found is queued as a proposal, not written to the record. An existing candidate is never overwritten — not by the model, and not by the form either. A recruiter decides what lands.
A form for every job
Thirteen fields by default, from name and email through to notice period, with your own questions added on top.
A link, not a login
Share one link. Candidates never make an account, and the public endpoint is rate-limited rather than left open.
Answers become the record
Typed answers map straight onto candidate fields, and the person is matched against everyone on file by normalised email and phone.
Nothing lands unreviewed
What the resume adds is queued as a proposal. An existing candidate is never overwritten by a form or by a model.
Recruitment automation
Keep hiring moving, even between conversations.
Rules run on the events your process already produces — a call finishing, an application entering a stage, a week going by. They can move a candidate, start a call or send a templated update, and the ones that matter wait for a person first.
RuleAfter a screening call, move strong candidates on
- Fires when the voice provider reports a finished call. Nobody is signed in for this one, so the engine runs it with a service client and only the actions it can perform itself are available.
- Conditions are read off the application at the moment the trigger arrives. A rule that does not match simply does not run — it is not an error and nothing is logged against the candidate.
- This rule requires approval, so it does not act — it proposes. The run parks and the actions are snapshotted, so what somebody approves is exactly what they read, even if the rule is edited in between. A proposal nobody answers expires after seven days, and lapsing is recorded separately from being rejected.
- A wait is a whole number of minutes, from one minute to thirty days, measured either from the other actions or from a scheduled call or interview. A pending wait is cancelled if the application changes stage — the reason for it has gone.
- The stage change is recorded against the application with the rule as its actor, so the activity log shows what moved this person and why.
- Sent from an approved template. Email is disconnected until somebody connects it, and the candidate's opt-out is checked before anything leaves.
- The same candidate the rest of this page follows. The card's stage is what the rule changed; everything else on the record is untouched.
Run logIllustrative
- 09:41Screening call completed
- 09:41Rule matched — 2 conditions
- 09:41Awaiting approval — actions snapshotted
- 09:52Approved by R. Menon
- 10:22Wait finished — 30 minutes
- 10:22Stage changed to Video Interview
- 10:22Candidate email sent
Two actions, one approval, and a named person on the line that mattered.
- Triggered
- Matched
- Proposed
- Approved
- Waited
- Acted
- Complete
Nine things can start a rule, and this is one of them: a screening call finishing. The voice provider calls back, so there is no signed-in user behind it — the engine knows that and limits itself to what it can do on its own.
Conditions are checked against the application as it stands. A rule that does not match does not run, and that is a non-event — nothing is logged against the candidate and nothing is retried.
This rule requires approval, so it does not act. It proposes, and the run stops here. Nothing has been sent, nothing has moved, and nothing will until a person says so.
A recruiter reads the proposal and approves it. What they approve is a snapshot taken when it was raised — editing the rule in the meantime cannot change what their click authorises. Left unanswered, a proposal lapses after seven days rather than firing late.
Then the wait runs, on a scheduler rather than a timer in somebody's browser. If the application changes stage while it is pending, the wait is cancelled — the thing it was waiting for has already happened.
The stage moves and the candidate gets a templated update. Both are recorded against the application with the rule named as the actor, so the history says what changed this and why.
One rule, end to end: an event nobody was present for, a gate a person had to open, and two actions recorded under the rule's name. Automation that keeps the process moving without deciding anything.
Nine real triggers
From an application being created to one sitting in a stage too long. Each one says where it is dispatched from, so you can tell whether your rule will actually fire.
Approval before action
A rule can be made to propose rather than act. The actions are frozen when the proposal is raised, so approving it cannot authorise something you did not read.
Waits that get cancelled
A minute to thirty days, measured from the actions or from a booked call. If the candidate moves on while a wait is pending, it is dropped.
One scheduler for every timed rule
Every time-based rule drains from one scheduled sweep. No timers in a browser tab, and no second clock to disagree with the first.
Integrations
Connect the tools your hiring workflow already uses.
Five connections, and every one of them is off until you turn it on. Each says plainly what stops working if you disconnect it — because the honest answer is usually less than you would expect.
ScoreboadOne hiring workspace
Five connectionsAll off by default
Pick one to see what it does and, more usefully, what stops working if you turn it off.
- Google Calendar — Creates interview invites and Meet links.
- Voice calling — Places automated screening calls to candidates.
- AI provider — Powers resume parsing, matching, summaries and drafting.
- Email — Delivers notifications and reminders outside the app.
- WhatsApp Business — Sends candidate messages over WhatsApp, alongside email.
Off until you connect it
Nothing reaches a candidate until an integration is connected on purpose. A fresh workspace sends nothing to anyone.
Credentials you cannot read back
Keys are encrypted before they are stored and the column is revoked, so they cannot be read back out — not through the app, and not by a query.
It tells you what breaks
Before you disconnect anything, the product lists exactly which features stop and which carry on regardless.
The manual path still works
Every integration is assistance, not a dependency. Without the AI provider connected, every manual workflow keeps working.
Analytics
Turn hiring activity into clear, actionable insight.
See what is happening across your hiring workflow and understand where it slows down — built from the records your team already creates, and honest about what the numbers can and cannot tell you.
ActivityIllustrative
- Application receivedSenior Backend Engineer
- Resume parsedA. Sharma
- Screening call completed4m 51s
- Screening report reviewedR. Menon
- Interview scheduledVideo · round 1
- Interview feedback submitted2 interviewers
- Evaluation recordedNeeds Review
- Stage changedDirector Round
Applications
124
Hired
2
Median time to hire
24d
Screening completion
68%
Recruitment funnel
Each step counts applications that ever reached it, not those sitting there now.
- Applied124100% from previous
- Shortlisted4637% from previous
- AI Screening Call2823% from previous
- Video Interview97% from previous
- Hired22% from previous
Time in stage
Median days, from closed stage visits only. A stage nobody has left yet shows no figure.
Applied2d
Shortlisted4d
AI Screening Call3d
Video Interview6d
Director Round—
Hire rate by source
A source with too few candidates shows its raw counts instead of a percentage.
Career page
3 of 74
Referral
2 of 18
Agency database
Too few to rate
A recruiter reads it. Analytics says where the process is slow and which sources are worth the effort. It does not rank candidates, and it decides nothing.
- Activity
- Structure
- Analytics
- Decision
Every step already leaves a record — an application arriving, a call finishing, feedback going in, a stage changing. Nobody has to fill in a spreadsheet for any of this to exist.
Those events are already structured, because each one was written by the screen that produced it. Grouping them is counting, not guessing — there is no parsing step between what happened and what is reported.
Which is what makes the figures defensible. The funnel counts everyone who ever reached a step rather than who is sitting there now, time in stage uses only visits that have ended, and a source with too few candidates shows its raw counts instead of a rate.
And then a person reads it. Analytics here tells you where the process is slow and which sources are worth the effort. It does not rank your candidates, and it does not decide anything.
Recruitment analytics
Applications, hires, median time to hire and screening completion, filtered by date range, job, client or recruiter.
Hiring workflow visibility
Median days in each stage, counted from visits that have actually ended — so a slow step shows up as a number rather than a feeling.
Candidate pipeline insights
Where applications come from, how far each source gets, and what the funnel loses between one step and the next.
From activity to insight, your hiring workflow stays connected. Explore the platform
Who it's for
Built for the people who make hiring happen.
Whether you are running a desk of client roles, hiring for your own team, or sitting one interview a month, it is the same workspace — you just open different parts of it.
Agency recruiter
Many clients, many roles, and submissions to track.
- Take the requirement
- Search candidates already on file
- Bulk-add new resumes
- Screen and shortlist
- Submit to the client
- Track feedback turnaround
The one setting that genuinely changes the product. In agency mode there are client companies, submissions to them, and a measured feedback turnaround; an in-house workspace has none of those surfaces at all.
In-house talent team
Fewer roles, deeper process, more stakeholders.
- Open the role with its stages
- Route inbound applications
- Match and screen
- Coordinate interview panels
- Record structured evaluations
- Hand off to onboarding
Fewer roles and more people to satisfy on each one. Everything client- facing is switched off, so the workspace is the pipeline, the panels and the records they write to.
Hiring manager
You do not want a recruiting tool. You want a decision.
- See only your roles
- Read the shortlist with reasons
- Take the interview with a brief
- Leave structured feedback
- Approve the offer
There is no hiring-manager role in the product, and that is the point: a hiring manager does not administer anything. They take an interview with a brief and leave feedback in a fixed shape, so two opinions can be compared rather than merely read.
Candidate
They never signed up for this product, and never have to.
- Apply from a public link
- Take a screening call with a disclosure
- Open the coding round from a link — no install, no login
- Receive interview details
- Upload documents on hire
A candidate never opens this product, and never makes an account. They get a link to apply, a call that says what it is, a coding round that needs no install, and an opt-out that works without a login.
People, Intelligence, Opportunity
Security and trust
Built with trust at every step of the hiring workflow.
Hiring decisions affect people's livelihoods, so the guardrails are part of the architecture rather than a policy page. Everything below is something the product does today — not a roadmap, and not a badge.
Every request proves who sent it — a session, a signed link or a signature.
- Sessions are HTTP-only cookies, refreshed at the edge on every request.
- Every route handler resolves the tenant from that session — the five that do not are public by design, and each is authorised by a signed token or an HMAC signature instead.
- Candidate-facing links need no account at all, which is why they are signed rather than guarded.
Hiding a control is cosmetic, so the rule lives in three places.
- Four roles, enforced in the interface, in the API and in the database — not one of the three on its own.
- Role changes are database triggers rather than application code: only an Owner may grant or revoke Owner, and an organisation cannot be left without one.
- Every change is logged with both the old value and the new one.
A workspace cannot see another one, even if it asks precisely.
- The tenant comes from the session. The active-workspace cookie is only a hint, re-checked against real memberships on every request — a forged value gets the default workspace, not access.
- An id belonging to another workspace returns not found rather than forbidden, so nobody can use the difference to work out which records exist.
- Around twenty database triggers stop a correctly-owned row from pointing at another workspace's parent.
The links sent to candidates store no secret at all.
- Each one is a signature over a single row id, with nothing written down — so a database copy contains no working links, and there is no token column to leak.
- A bad signature, a missing key and a malformed link all fail identically and in constant time, because telling them apart would let somebody probe which records exist.
- One version bump revokes every link and QR code a form has ever issued, still without storing a secret.
What happens to anything a model produces
- AIAI outputA structured result, never a free-text answer
- CodeValidatedShape and figures checked before anyone sees it
- PersonReviewedOn a screen, beside what it was derived from
- PersonDecisionRecorded against the person who made it
The principles that hold across the product
Human review, by construction
AI output is validated and shown to a person before it reaches a trusted record. There is no path that skips that step.
Candidate consent
Screening calls open with a disclosure that cannot be switched off, and every automatic message carries a working opt-out.
Isolation in the database
Each organization's rows are separated by row-level security in PostgreSQL rather than by a filter in the application.
An append-only audit log
Who changed what, and when — including who enabled an automation and who overrode a candidate's opt-out.
Encrypted integrations
Provider credentials are encrypted at rest and readable only through a privileged server path.
Keeps working when the AI does not
If an AI provider is unavailable the manual workflow keeps running. You lose the assistance, not the ability to recruit.
No certification is held, and this section claims none. The awkward version of that answer is in the FAQ below. Read the FAQ
Use cases
See Scoreboad in action across the hiring workflow.
From the first application to the final decision, Scoreboad connects the people, information and workflows involved in hiring. Pick the situation that sounds like yours.
High-volume candidate screening
Two hundred applications against one role, and the reading is the same work every time — extract the facts, check them against the requirement, decide who is worth a call.
- Application
- Resume parsed
- Requirements checked
- Match score
- Review
What you end up withThe reading is done and the reasons are on the record. A person still decides who moves.
Interviews that can be compared
Two interviewers, two sets of notes, two different shapes. Comparing them afterwards is guesswork, and the round that mattered is the one nobody wrote down properly.
- Round scheduled
- Brief generated
- Feedback captured
- Evaluation
- Verdict
What you end up withFeedback in one shape per round, on the application, beside everything else known about the person.
One record per person, not per application
The same person applies twice, six months apart. Without a single record, the second application starts from nothing and the first conversation may as well not have happened.
- Candidate
- Resumes
- Applications
- Screening
- Activity
What you end up withOne record carrying every application, every resume and an append-only history of what changed and who changed it.
Knowing what has stopped moving
Nothing is obviously wrong, and three people have been sitting in the same stage for a fortnight. The board looks busy either way.
- Stage target
- Days in stage
- At risk
- Breached
- Someone acts
What you end up withEvery card aged against the target you set for its stage, so a stall is a number on a column rather than something you notice late.
A process that survives the team growing
What two people held in their heads stops working at five. The steps are still the steps; they are just no longer in anybody's head.
- Job stages
- Screening questions
- Rules
- Approval
- Audit log
What you end up withThe process written down where it runs — per-job stages and questions, rules that can wait for a person, and a log of everything that happened.
These are product scenarios, not customer stories. Scoreboad is early and has no customers to name yet — so there are no logos here, and nothing on this page claims otherwise.
Resources
Resources for better hiring.
Practical product resources and hiring insights for teams building a more connected recruitment workflow. Everything listed here is published and readable now.
Capability guides
One page per part of the hiring process, written as what it does rather than what it is called.
Source Every requirement, candidate and application in one place.
Open a role, bring candidates in the way they actually arrive — a career page, a bulk folder of resumes, a referral, a manual add — and keep every application tied to the job it was for.
Understand Read every resume, without reading every resume.
Resumes arrive as PDFs and Word files written to no standard at all. They are parsed into structured profiles, then scored against the role with deterministic checks first and semantic matching second — so the reason for a score is always inspectable.
Screen First-round screening calls that actually happen.
An AI voice agent runs the first screening conversation, then the call becomes a structured report a human can act on — availability, notice period, expectations, and the answers to the questions you set.
Decide A pipeline that tells you what is going wrong.
Move applications through the stages that role actually uses, with service-level tracking that surfaces the candidate nobody has touched in nine days. Schedule interviews with a prepared brief, and run the technical round in the product instead of over a screen share.
Close From offer to first day, without a spreadsheet.
Share a shortlist with the client and track how long feedback actually takes. Once someone is hired, the document checklist opens itself and chases what is missing, so onboarding is not an email thread.
Operate Measure it, automate it, and know who changed what.
Analytics over the whole funnel with the filters that matter, automation rules for the repetitive moves, candidate and internal communication from one template library, and an append-only audit log behind all of it.
Answers
The questions a careful buyer asks before they trust a product with candidate data.
AI safety Where the model stops
The fixed pipeline every AI feature runs through — raw data, AI, a structured output, validation, human review, and only then a business action.
Use cases Which hiring problem is yours
Five situations, each pointing at the part of the product that answers it. Product scenarios rather than customer stories, and it says so.
FAQ The awkward questions, answered
What the product does not do yet, what it does not hold, and what it costs — including the answers that are not flattering.
Still to come: a blog, hiring guides, product documentation. None of them exist yet, so none of them are linked.
Questions
Frequently asked questions.
What is Scoreboad?
One workspace for hiring: jobs, candidates, applications, resume screening, first-round calls, interviews, evaluation, onboarding and analytics — with AI assisting at each step instead of sitting in a separate tab.
How does Scoreboad use AI?
For the repetitive reading and writing: parsing resumes, scoring them against a role, summarizing screening calls, drafting briefs and messages. Each one is a named function with a defined input and a validated output, not an open-ended chatbot.
Does AI make the final hiring decision?
No, by design. AI parses, matches, summarizes and drafts. A person reviews that output and makes every consequential decision, and AI output is never written straight to a trusted record.
How does candidate screening work?
A resume is parsed into structured fields and scored against the role's requirements, with the reasoning shown beside the score. Optionally a structured first-round voice call runs and its answers are captured on the candidate's record. A recruiter reviews both.
Can recruiters review and override AI results?
Yes, and that is the normal path rather than an escape hatch. Parsed fields are proposed, not applied; scores are advisory; generated text is editable before it goes anywhere.
How does Scoreboad handle candidate information?
Each organization's data is isolated by row-level security in the database. Files sit in private storage behind short-lived signed links, recipients are masked in logs, and an append-only audit log records who changed what.
Do candidates need an account?
No. Applying, taking a screening call, opening a coding round and unsubscribing all work without a login, authorised by signed links instead.
What happens if an AI service is unavailable?
The manual workflow keeps working. AI failure is treated as an expected condition throughout rather than an outage — you lose the assistance, not the ability to recruit.
Do you hold SOC 2 or ISO 27001 certification?
Not today, and we would rather say so than imply otherwise. The architecture is built to the practices those audits look for — tenant isolation at the database layer, encrypted credentials, an append-only audit log and least-privilege access — but no certification has been completed.
Pricing
What Scoreboad costs today.
Nothing, because there is nothing to charge for yet. There is no billing in the product — no plans, no card, no trial running down — and that is worth saying plainly rather than leaving you to hunt for a price list that does not exist.
Today
An account costs nothing, and takes no card
- There is no payment step, because there is nothing in the product that could take a payment.
- No trial is running, so nothing expires and nothing needs cancelling.
- No plan limits either — not because they are generous, but because there are no plans to limit anything.
- Everything described on this page is in the product now. What is not built yet is said so on the page that describes it.
When there is a price
When pricing is set, it will be published here
- Pricing is unpublished because it is undecided, not because it is being withheld until you ask.
- When it exists it will appear on this page with the figures on it, not as a form that promises a quote.
- Anyone already using the product will be told before anything changes for them.
Build your next team with Scoreboad.
Bring your hiring workflow into one intelligent workspace.