Interview Setup options explained, with examples
What every Interview Setup option in the LastRound AI desktop app changes about your answers, with an example for each: resume, domain, job description, the hands-on experience toggle, your instructions, About you, behavioral stories, answer style, STAR, metrics, auto-submit, the 61 interview languages and the confirm screen.
Last reviewed 2026-09-13
About the examples on this page
The example candidate
| Field | Value |
|---|---|
| Name | Priya Sharma (made up) |
| Resume says | Backend Engineer at Northwind Payments, 4 years. Python, Django, PostgreSQL, Redis, AWS (EC2, S3). "Cut checkout API p95 latency from 800 ms to 250 ms." |
| Interviewing for | Platform Engineer. The job description asks for Kubernetes, Terraform and GitHub Actions. |
| Not on the resume | Kubernetes, Terraform and GitHub Actions. |
How your settings reach an answer
- Interview category and Specific domain: the field the interview is in.
- Job description, the Present this role's skills as hands-on experience choice and Your instructions.
- In Code mode only: the coding round type and programming language.
- Your resume.
- About you, your answer style, length and tone, the STAR and metrics switches, and your behavioral stories.
- Honesty rules: never invent employers, dates, team sizes or numbers that are not in your background.
- A running note of what has already been said in this interview, so later answers stay consistent.
- Interview language, added last as a hard rule: the whole answer is written in that language.
| If these disagree | This wins |
|---|---|
| Resume vs About you | The resume. About you only fills gaps the resume leaves. |
| Your instructions vs Style, Length or Tone | Your instructions. |
| Your instructions vs the honesty rules | The honesty rules. Instructions shape how you come across, not the facts of your history. |
| Your instructions vs Interview language | Interview language. To get answers in another language, change Interview language rather than writing "answer in Spanish". |
Changes apply from your next answer. An answer that is already on screen is not rewritten.
Resume
What happens: The AI answers as you, in the first person, using the employers, projects, tools and results in your resume. It treats the resume as the truth about your history: it will not claim a company, project or number that is not in it. Up to the first 6,000 characters of extracted text are used, so keep your most relevant experience near the top.
Example answer with Priya's resume loaded
Interviewer asks: Tell me about a performance problem you fixed.
"At Northwind Payments our checkout API had drifted up to about 800 milliseconds at p95. I profiled the slowest requests and found two PostgreSQL queries doing most of the work, so I added the missing index and cached merchant settings in Redis. That brought p95 down to around 250 milliseconds."
One exception: screenshot assessment answers
Interview category and Specific domain
What happens: The AI is told this is, for example, a Kubernetes Engineering interview in Cloud Computing & Infrastructure, and to use that field's terminology and best practices in every answer. The same question gets a different angle depending on the domain you pick.
Interviewer asks: How would you handle a sudden spike in traffic?
Example: domain set to Backend Development
"First I'd make sure the hot paths are cheap: cache read-heavy responses, keep database queries indexed, and push slow work like emails and reports onto a background queue so request threads stay free. Then I'd add rate limiting so one client can't take everyone down."
Example: domain set to Kubernetes Engineering
"I'd rely on a Horizontal Pod Autoscaler driven by CPU or request rate, with sensible requests and limits so scheduling is predictable, and a cluster autoscaler so new nodes arrive when pods can't be placed. Readiness probes keep traffic off pods that are still warming up."
Job description
What happens: Answers are steered toward this role. The AI uses the posting's vocabulary, leans on the tools, platforms and responsibilities it names, and when a question could go several ways, picks the angle this role cares about. This is what makes an interview JD-targeted.
Interviewer asks: What does good deployment automation look like to you?
Example: no job description
"Every change should go through the same automated path: tests on every pull request, a build that produces one artifact, and a deploy that can be rolled back in one step. At Northwind we deployed the Django services to EC2 that way."
Example: Platform Engineer job description pasted
"For a platform team I'd want pipelines defined as code in something like GitHub Actions, infrastructure changes planned and reviewed through Terraform, and deploys to Kubernetes rolled out gradually with automatic rollback on failed health checks. At Northwind I built the test-and-deploy flow for our Django services, and the same principles carry over."
Notice the second answer names the job description's tools but does not claim Priya has used them. That is the next setting.
Present this role's skills as hands-on experience
| Setting | What happens |
|---|---|
| Off (default) | Skills missing from your resume are explained, not claimed. The AI explains the skill confidently from fundamentals and links it to the closest thing you have really done, without saying you lack experience and without inventing a history. |
| On | Those skills are answered as work you have done: first person, concrete and confident. It still never invents named employers, dates, team sizes or exact metrics that are not in your resume. It describes the work, not a fake CV line. |
Interviewer asks: Have you worked with Terraform?
Example answer with the toggle Off
"Terraform lets you declare infrastructure as code, keep state so it knows what already exists, and review a plan before anything changes, which makes environments repeatable. The closest thing I've done is managing our AWS setup for the Django services, where keeping EC2 and S3 configuration consistent between staging and production was exactly the problem Terraform solves."
Example answer with the toggle On
"Yes. I've used Terraform to define our AWS environments as modules, so staging and production came from the same code with different variables. I kept remote state locked so two people couldn't apply at once, and every change went through a reviewed plan before apply, which caught a few risky changes before they reached production."
Only turn it on if you can back it up
Your instructions
What happens: Your instructions are followed for every answer in the interview. They override the general style settings (Style, Length, Tone) but never the honesty rules, and never the interview language.
Example
Instructions: "Lead with my checkout latency work whenever performance comes up. The interviewer is the Head of Platform. Do not mention Northwind by name."
Interviewer asks: How do you decide what to optimise first?
"I measure before I touch anything. On a checkout API I owned, p95 was around 800 milliseconds, and profiling showed two queries were most of it, so that's where I started. Fixing those and caching the settings they read brought it to about 250. For a platform team I'd apply the same rule to build and deploy times."
The answer leads with the latency story, pitches it at a platform leader, and says "a checkout API I owned" instead of naming the employer.
About you
| Field | What happens when you set it |
|---|---|
| Full name | The AI knows your name, so an answer to "Priya, tell us about yourself" sounds like you. |
| Current role | Your title today. "So you're a backend engineer now?" gets an answer that matches it. |
| Years of experience | Fresher (0 years) to 20+ years. Sets how senior the answers sound: a fresher talks about projects and learning, 10 years talks about trade-offs and leading others. |
| Target role | The title you want. With Platform Engineer set, "Why this role?" is answered as a move toward platform work, not as another backend job. |
| Target company (optional) | Answers can reference the company when it is relevant, for example why its scale or product appeals to you. |
| Industry | Technology, Finance, Healthcare, Retail, Education, Consulting or Other. Frames examples for that sector. |
| Skills | Headline skills you want mentioned even if the resume buries them. Add "Incident response" and it can come up in an on-call question. |
| Key achievements | Free text, for example "Cut checkout p95 from 800 ms to 250 ms; led a 3-person migration". These are the real figures the Include metrics and numbers switch draws on. |
Behavioral stories
| Field | Write about |
|---|---|
| Leadership experience | A time you led a team or project. |
| Challenging situation | A difficult problem you solved. |
| Teamwork example | A successful team collaboration. |
| Problem-solving approach | How you break down complex problems. |
Example
Leadership experience: "Led a 3-person effort to move our jobs from cron on one EC2 box to a queue with retries. Wrote the plan, split the work, ran the cutover on a weekend. No missed jobs after the switch."
Interviewer asks: Tell me about a time you led a project.
"Our scheduled jobs all ran from cron on a single EC2 machine, so one bad deploy could silently skip payouts. I proposed moving them onto a queue with retries, wrote the plan, and split the work across three of us. We ran the cutover on a weekend with the old jobs still in place as a fallback, and after the switch we didn't miss a single job."
Two or three sentences per story, with the outcome, is enough. Leadership, challenge and teamwork stories are sent to the AI with every answer.
How the AI answers: Style, Length and Tone
| Setting | What happens |
|---|---|
| Style | Concise (default): direct, no filler, nothing beyond what was asked. Balanced: the answer plus brief supporting detail. Detailed: comprehensive, with examples and depth. |
| Length | Short, 30 to 60 seconds aloud (default): 2 to 4 sentences, one paragraph. Medium, 1 to 2 minutes: 4 to 8 sentences in two short paragraphs. Long, 2 to 3 minutes: 8 to 14 sentences over two or three paragraphs with an example. Outside Code mode, Length is applied as a hard limit. |
| Tone | Formal, Professional (default) or Conversational. Professional is what most panels expect; Conversational reads more naturally in startup or culture-fit rounds. |
Interviewer asks: What is a Kubernetes readiness probe?
Example: Concise, Short, Professional
"A readiness probe tells Kubernetes when a pod can take traffic. Until it passes, the pod is kept out of the Service's endpoints, so users never hit an instance that is still starting up."
Example: Balanced, Medium, Conversational
"It's basically Kubernetes asking a pod, are you ready for traffic yet? Until the check passes, the pod stays out of the Service, so nobody gets routed to something that's still loading config or warming a cache."
"It's different from a liveness probe, which restarts a pod that has hung. I'd point the readiness check at an endpoint that actually confirms the database connection is up."
STAR structure for behavioral answers
What happens: Behavioral answers follow the Situation, Task, Action, Result order, so the story has a clear start, what you were responsible for, what you did, and how it ended. Technical questions are not forced into that shape.
Interviewer asks: Tell me about a time something broke in production.
Example: STAR off
"We once had checkout timing out during a sale. I found a slow query, added an index and a cache, and it recovered. Since then I always load-test before big launches."
Example: STAR on
"During a sale weekend our checkout API started timing out (Situation). I was on call, so it was on me to get payments flowing again (Task). I traced it to two slow PostgreSQL queries, added the missing index and cached merchant settings in Redis (Action). Timeouts stopped within the hour, and p95 settled around 250 milliseconds, down from 800 (Result)."
The labels in brackets are added here to show the structure. The real answer reads as normal speech.
Include metrics and numbers
What happens: The AI works the numbers from your resume and Key achievements into answers. It does not make numbers up, so this only helps if your resume or achievements contain them.
Interviewer asks: What impact did your work have?
Example: metrics off
"I made our checkout API much faster, which cut timeouts and made the payment flow noticeably smoother for customers."
Example: metrics on
"I brought checkout API p95 latency down from 800 milliseconds to 250, more than three times faster, which is what stopped the timeouts during peak traffic."
No figures on file?
Auto-submit AI response
| Setting | What happens |
|---|---|
| Off (default) | Nothing is generated until you ask. When the interviewer finishes, press ⌘/Ctrl + Enter or click Answer. You choose which questions get an answer. |
| On | An answer starts on its own each time the interviewer stops talking. Very short turns such as "okay" are skipped. If the next question ends while an answer is still streaming, it is queued and answered as soon as the current one finishes. |
Example
The interviewer says, "Thanks, that makes sense. So, how would you design a rate limiter for a public API?" and pauses.
Off: the transcript shows the question and waits. You press ⌘/Ctrl + Enter when you are ready.
On: the answer starts streaming as soon as the pause is detected, with no key press.
Which to choose
Interview language
What happens: You choose one language per interview. The interviewer's speech is transcribed in that language, and every answer comes back in that language. Languages are never mixed and never auto-detected: in an English interview where the interviewer drops in a few Hindi words, you still get English answers. Technical terms, product names and code keep their usual form. The setting applies to Live, Code and screenshot answers alike.
Example: Interview language English
Interviewer asks: Can you explain what a load balancer does?
"A load balancer spreads incoming requests across several servers, so no single one is overloaded and traffic keeps flowing if one goes down."
Example: Interview language Spanish
Interviewer asks: ¿Puedes explicar qué hace un balanceador de carga?
"Un balanceador de carga reparte las peticiones entre varios servidores, así ninguno se satura y el servicio sigue funcionando si uno falla."
| Where | When to use it |
|---|---|
| Onboarding, last step | The first time you set up the app. |
| Settings → General → Interview Language | To change your usual language. |
| Interview Setup, This interview | To set it for the interview you are about to start. |
Pick the language the interviewer actually speaks
All 61 interview languages
Available in the desktop app. The browser and mobile copilots are not covered by this list.
- English
- Afrikaans
- Arabicالعربية
- ArmenianՀայերեն
- Assameseঅসমীয়া
- BelarusianБеларуская
- Bengaliবাংলা
- BosnianBosanski
- BulgarianБългарски
- CatalanCatalà
- Chinese (Cantonese)粵語
- Chinese (Mandarin)中文
- CroatianHrvatski
- CzechČeština
- DanishDansk
- DutchNederlands
- EstonianEesti
- FinnishSuomi
- FlemishVlaams
- FrenchFrançais
- Georgianქართული
- GermanDeutsch
- GreekΕλληνικά
- Gujaratiગુજરાતી
- Hebrewעברית
- Hindiहिन्दी
- HungarianMagyar
- IndonesianBahasa Indonesia
- ItalianItaliano
- Japanese日本語
- Kannadaಕನ್ನಡ
- KazakhҚазақ
- Korean한국어
- LatvianLatviešu
- LithuanianLietuvių
- MacedonianМакедонски
- MalayBahasa Melayu
- Marathiमराठी
- MongolianМонгол
- Nepaliनेपाली
- NorwegianNorsk
- Pashtoپښتو
- Persianفارسی
- PolishPolski
- PortuguesePortuguês
- Punjabiਪੰਜਾਬੀ
- RomanianRomână
- RussianРусский
- SerbianСрпски
- SlovakSlovenčina
- SlovenianSlovenščina
- SpanishEspañol
- SwedishSvenska
- Tagalog
- Tamilதமிழ்
- Teluguతెలుగు
- Thaiไทย
- TurkishTürkçe
- UkrainianУкраїнська
- Urduاردو
- VietnameseTiếng Việt
Confirm interview details
What happens: When you press Start (or any shortcut that starts an interview), a Confirm interview details screen appears first. It shows exactly what the AI will use. Nothing is billed until you confirm.
| Row | What it shows |
|---|---|
| Interview type | Live interview or Coding interview. |
| Domain | Your category and domain. If no domain is selected, you cannot start from this screen. |
| Candidate | Your name, target role and target company from About you. |
| Resume | The resume file name the AI will answer from. |
| Job description | The start of the pasted job description, or Not added. |
| Your instructions | The start of your instructions, or None. |
| Coding round | Coding interviews only: round type and programming language. |
| Interview language | The language for transcription and answers. |
| Answers | Automatically when the interviewer stops, or when you press Answer. |
Click Edit details to go back and fix something, or the green Details are correct button to start the interview.
Example
Yesterday Priya interviewed for a Backend Developer role. Today it is the Platform Engineer interview, but she forgot to change the setup. The confirm screen shows Domain: Software Development · Backend Development and yesterday's job description. She clicks Edit details, switches the domain to Kubernetes Engineering, pastes the new posting, and starts with the right profile, before a single credit is used.
"I always interview for this same profile"
Tick I always interview for this same profile (don't ask me again) on the confirm screen if you only ever interview for one profile. A warning explains the trade-off: every interview will start straight away with these details, so if you interview for a different role or company you must update Interview Setup before you press Start.
- It comes back on its own. The screen reappears automatically if any of these change: interview type, category, domain, name, target role, target company, resume, job description, your instructions, coding round type or language, or interview language.
- Answer mode does not count. Switching Auto-submit on or off changes when answers appear, not what they say, so it does not bring the screen back.
- You can turn it back on. In Settings → General → Interview Language, the Confirm interview details before each interview row has a Turn back on button.
Ready for a real interview?
Still stuck?
Signed in? Open a ticket under Account → Support in the app. Otherwise use the contact form or email. Tell us what you tried, which page of this guide you were on, and your operating system or browser.