Coding, system-design, and behavioral rounds in one shared, secure workspace — recorded, transcribed, and scored. Hiring teams get a rigorous, defensible process; coaches get a full back-office. Self-hosted on your own cloud.
One session, three surfaces — code, whiteboard, and shared notes. Click any to enlarge.
The same real-code workspace powers structured hiring interviews and dedicated one-on-one interview coaching — with the workflow each side actually needs.
Assess real engineering, not whiteboard trivia — and run the whole loop around it, from the first screen to the decision.
Run a coaching business end-to-end: sell sessions, book them, assign work, track progress — and give clients an AI interviewer to practice with between sessions.
Candidates, their interview loops, and the people who run them — a recruiter who keeps the pipeline moving, interviewers who only interview, and a committee that decides with every scorecard in front of it.
A candidate directory, and a loop per role: every round in order — scheduled, run, scored — and what each one was. It updates live as the team works.
What needs doing today, every interview across the team, and who is free to run the next one. Move an interview, hand it to another interviewer, or call it off — everyone in it is told.
The recruiter sets rounds up; interviewers run the ones they are given. A recruiter never sits on the interviewer side, so the questions, reference solutions and marking schemes stay with the people who need them.
The recruiter runs the first conversation — scored on motivation, fit and logistics — and, once the loop is decided, the call that tells the candidate how it went, with every interviewer's notes to hand.
Every finalized scorecard on the loop in one table, dimension by dimension, with each interviewer's summary — and how many hints each round took.
Record the decision once, with its rationale, and email the candidate that it was made — with the outcome, if your organization shares outcomes. Hiring events stream to your ATS by webhook.
A full interview, spoken out loud, with an AI interviewer — in the same room your sessions use, with questions from your own bank, follow-ups, hints when someone is stuck, and feedback at the end. Prefer to type? It works as a text chat too. Always labeled as an AI, never passed off as a person.
Run it like the real thing, where the interviewer assesses and helps only as an interviewer would — or as mentorship, with hints, explanations, and a better approach after each attempt.
Coding, system design, behavioral, or both. It asks questions from your bank, watches the code and test runs, and reads the whiteboard as the design takes shape.
Talk and it talks back, out loud — interrupt it, speed it up, follow along in subtitles. Or turn the voice off and it is a text chat. Held in twelve languages, eight of them spoken.
Clients say what they are preparing for — the company, the role, the topics — and the session is chosen and pitched to it. The same brief reaches coaches on their bookings.
Every session ends with the rubric filled in and a written recap of what went well and what to work on, kept in the client's history.
Price an AI session as a fraction of a coach's, include AI sessions in packages, or make them free. Booking times no coach can take become bookable with the AI — always the client's choice.
Once a client hands in their homework, they can go over it with the AI mentor — it reads what they handed in and talks it through. Never before, so it never does the homework for them.
A client can share any AI session with their coach — recap, scorecard and conversation — so practice between sessions shows up where the coaching happens. Off until they choose.
It picks questions the person has not seen yet that fit what they are preparing for, raises or lowers the next one by how the last went, and the mentor remembers what earlier sessions covered.
As a mentor it works alongside: a comment next to the line in question, a sketch on the whiteboard, and — when the bank has nothing on a topic — a fresh question, run against its own tests first and held for an administrator before anyone else is asked it.
Every model call is metered: a monthly usage report by model, purpose and session, and an optional budget that stops new sessions once it is spent. A fast model holds the conversation; a stronger one writes the assessment.
Optional, and off until you turn it on. It runs on a frontier model with your own key — or on a local model, such as Ollama, if nothing may leave your infrastructure.
Senior and manager sessions are offered as practice, marked as indicative rather than calibrated.
The same AI that runs practice sessions works for your interviewers, coaches and hiring committee — following along, drafting, screening — and never makes the call.
For the interviewer or coach alone: it hears the conversation, follows the code, test runs and whiteboard, and speaks up only when something is worth a look — the next hint, a claim the code contradicts, a topic not yet covered.
Type it a question during the session — what to probe next, whether that complexity claim holds. It knows the earlier rounds' summaries, never their scores, so it helps without anchoring.
Send a candidate a first round the AI runs, taken whenever suits them within a window. They are told plainly an AI conducts it and that people decide; its assessment is advice, kept out of the side-by-side unless someone adds it.
After a round, the AI drafts the rubric from the recording, the code and the notes. It fills only what is blank; the interviewer edits, and the scorecard stays theirs.
By default, a notice tells candidates an AI assistant may be used in the interview; an AI screen always says so before anything starts. Who gets the assistant is an administrator's choice, person by person.
Nothing the AI writes records a decision or moves a candidate along. It suggests, drafts and screens; the interviewers score and the committee decides.
The parts of a coaching business that are not the hour itself — selling it, scheduling it, getting paid for it, and keeping a client's preparation on track.
Every coach has a page to share — photo, rating, experience, and what they offer at what price. People who sign up from it land back on it.
One price formula across focus, company, and level, with bundle discounts and one-off prices. Referral links, campaigns, and bonus sessions for growth.
Two-way Google Calendar sync, booking windows, and a coach who can't make it hands the session to another, so the client keeps the slot.
Coach pay rates by coach and offering, payouts recorded as they happen, and an optional no-show policy with fees and compensation.
An intake on goals and timeline, the onsite dates on the program, an agenda and recap for every session, messages with the coach, and skill-by-skill progress charts.
Download a homework submission or a session as one package — the marking scheme, the tests, the transcript, and the work — for a careful review on your own time.
slate isn't only for coding. Switch a session to a built-in whiteboard for system design, or a shared notes doc for behavioral rounds — all recorded and replayable. On every surface, the problem brief, signals, and follow-ups sit alongside, visible only to the interviewer — never the candidate.
A full IDE in nine languages, running in a sandbox with real diagnostics and completion.
A collaborative whiteboard to diagram architecture and data flow while you talk it through.
A shared doc, in sync for both sides. Queue several questions per session and capture answers as you go.
A single, low-latency workspace — no plugins, no local setup, nothing for the candidate to install.
A full Monaco editor with shared cursors and instant sync. Both sides type in the same file with zero conflicts.
Every run happens in a fresh, throwaway sandbox with no network access — candidate code never touches your infrastructure.
Diagnostics, hover, and completion for all nine languages — the IntelliSense engineers expect in their own IDE, sandboxed per language.
Talk over built-in audio, with optional live captions generated in the browser, then replay the whole session — code, whiteboard, audio, and edits in sync, with seek and search.
Every recorded session is transcribed automatically, in any spoken language, right inside your deployment.
Custom rubrics with per-dimension scoring, an optional blind reveal, and automatic grading against your test cases.
Programs, homework, scheduling, packages and credits — a complete back office for running interview coaching as a business.
Curate problems once, with typed function signatures — slate generates idiomatic stubs and test harnesses for every language automatically.
Sign in with Google, GitHub, or Apple (or email + passkeys). Stream session and hiring events to your ATS and tools via signed webhooks.
Either side can turn a camera on for a more personal session — live only, never recorded — in tiles you can move and resize.
Share a screen when the round needs a real tool — an iOS or Android app built in Xcode or Android Studio — with screenshots kept in the replay.
Staged problems reveal one constraint at a time, each invalidating the last answer — the follow-up ladder a strong interviewer runs, built in.
Candidates code in the language they'd actually use on the job.
Executing a stranger's code is the hardest part of an interview platform. slate treats it as the defining constraint — not a feature bolted on later.
A recruiter sets the interview up from the candidate's loop for one of your interviewers, who picks the questions from your bank. Nothing to install for either side.
Meet in the shared editor, whiteboard, or notes with audio on. The candidate writes and runs real code in the sandbox while you follow along and take notes.
Fill the rubric — or start from an AI draft — and auto-grade the code. The committee sees every scorecard side by side, with the recordings and transcripts behind them.
The same platform runs across a Kubernetes cluster for scale, or as one self-contained instance you can stand up in minutes. Config, not code, switches between them.
Each organization gets its own brand — logo, colors, fonts, and a custom domain — and can send transactional email from its own domain. Great for agencies and multi-brand coaching businesses.
A Kubernetes cluster on GCP (GKE) or AWS (EKS) for scale, or a single Linux machine running the all-in-one image with Postgres, Redis, and S3-compatible storage. We'll help you plan the rollout on the demo call.
No. Recordings, audio, transcripts, and code stay on your servers, and candidate code executes in sandboxes with no network access. The one exception is the AI features, which are optional: they use a frontier model by default, so the sessions they work on go to that model's provider — or run them on a local model such as Ollama, and nothing leaves.
Yes, if the organization turns it on. Coach and client then know each other only by pseudonym — across the coach directory, bookings, the session itself, and every email — so a client can prepare without their name, or their employer's, reaching the coach.
No — it gives clients more practice between the sessions that matter. Clients always know when they are talking to an AI, it never takes a booking without the client choosing it, and you decide what it costs, or whether it is free.
No. It assists the interviewer, drafts a scorecard for them to edit, and can run a screening round whose assessment is kept apart as advice. Interviewers score and your committee decides. Candidates are always told when an AI conducts a round, and by default that an assistant may be used in the others.
Pricing depends on your deployment and team size. Request a demo and we'll give you a straight answer for your situation.
A single-node instance stands up in under an hour. A production cluster deployment — including SSO, custom domains, and branding — is typically days, not weeks.
Yes — every branding surface is per-organization: logo, colors, fonts, a custom domain, and email sent from your own domain.
A 30-minute live walkthrough: we run a mock interview together — code, whiteboard, and notes — then replay it with the transcript and scorecard, and talk through how a deployment would look on your infrastructure.
We're onboarding hiring teams and coaching businesses now. Book a personalized walkthrough and we'll help you plan a self-hosted rollout.
or email hello@slatecareers.com