01Conversational technical
Real conversation to assess fundamentals, applied reasoning and domain depth.
- Voice or text
- Follow-ups that follow
- Clarity of reasoning
Best for: All roles
Powered by CognityHire
The full-stack AI interview platform
The AI interview platform that replaces the entire interview loop.
CogniSense is an AI interview platform that runs the entire loop — conversational rounds, live coding, architecture reviews, production simulations and executive assessment — from screening through to C-suite, with evidence behind every conclusion.
- Enterprise-grade security
- Proctored & fair
- Evidence-backed reports
- Available standalone
- System Design & ArchitectureWhiteboard. Design. Discuss.
- Coding InterviewsWrite. Run. Debug. Discuss.
- Simulation: Debug & FixDebug real issues. Explain the approach.
- Simulation: Pair ProgrammingCollaborate. Build. Solve together.
- Simulation: Incident TriageInvestigate. Prioritise. Communicate.
- Simulation: Code ReviewReview as you would in the real world.
- Leadership & ExecutiveVision, judgement and leadership impact.
- Conversational InterviewsNatural conversation. Real understanding.
- Voice & text interviews
- Interactive canvas
- Real coding environment
- Adaptive AI interviewer
- Proctored & secure
- Structured reports
- Cohort benchmarks
The range
Most interview platforms stop where the hard part starts.
They screen. Screening is the cheapest round to automate and the least valuable one to get right. Everything above it is still done by pulling senior people out of their week and putting them in a room — which is exactly the part that does not scale, and exactly the part where inconsistency costs the most.
Screening
Where most platforms begin and end
Mid-level
Depth beyond whether the tests pass
Senior
Trade-offs, judgement, and defending a design
Principal / Staff
Architecture under real constraints
Director / VP
Systems thinking across teams and quarters
CTO and C-suite
Live discussion at the level the role demands
What occupies that bottom rung: Screening quizzes and Recorded video.
This platform runs all of it. The same system takes a first-round screen and a conversation with somebody who will report to your CTO — because what changes between them is the depth and the scenarios, and both are set per role.
“Explain caching” is not a principal engineer question. The same skill is probed differently at every level — and what a communication score measures for a junior is not what it measures for a CTO.
Screening quizzes — They measure recall. They cannot stand in for a technical panel, and nobody claims they can.
Recorded video — A scheduling optimisation wearing the clothes of an assessment. No adaptation, no probing, no interview — just playback.
Product film — in progress
We are still recording the full walkthrough.
Rather than wait, we will take you through it live — thirty minutes, your own role, and a real interview run end to end. Honestly a better look at it than a film.
Rich interview modes
Built for the full interview loop.
From large-scale screening to principal engineering and C-suite leadership, CogniSense covers the whole spectrum.
- Adaptive and dynamic
- Questions evolve from what the candidate actually says.
- Evidence-backed
- Every response, action and artefact is captured as evidence.
- Seniority-aware
- Depth and complexity move with the level of the role.
- Consistent and fair
- The same standard for every candidate, every time.
- Secure and proctored
- Integrity signals and audit trails, with a person reviewing.
Mix and match rounds to build the loop your role actually needs.
Interview rounds
Every type of interview.All in one platform.
Every round you currently run with people — at every seniority level, with the depth and adaptability you would expect from your best interviewer.
02Coding interviews
Write, run and debug with live discussion and iterative problem solving.
- Live coding environment
- Multi-language
- Edge cases and complexity
Best for: Engineers, developers
03System design & architecture
Design real systems on an interactive canvas, defended as they are drawn.
- Live canvas
- Architecture diagrams
- Trade-offs, in depth
Best for: Senior and staff engineers upward
04Simulation: debug and fix
Real broken code, real logs — diagnose the cause, fix it, and explain both.
- Broken-code scenarios
- Logs and diagnostics
- Fix, test, explain
Best for: Engineers, SDE II upward
05Simulation: code review
Review a pull request as a senior would, with an author who pushes back.
- Pull-request review
- Quality and security
- Constructive feedback
Best for: Senior engineers upward
06Simulation: incident triage
Lead a live production incident while the picture keeps changing underneath.
- Real incident scenarios
- Logs, metrics, alerts
- Mitigate, then root-cause
Best for: SRE, DevOps, engineers
07Simulation: pair programming
Build together — as a peer, as a junior to guide, or as an expert who disagrees.
- Shared environment
- Collaborative problem solving
- Communication and iteration
Best for: All engineering levels
08Leadership & executive rounds
Strategic, ambiguous, cross-functional scenarios for senior and executive hires.
- Strategy and decisions
- Stakeholder judgement
- Vision and impact
Best for: Tech leads, directors, VPs, CTOs, C-suite
Adaptive. Comprehensive. Human-caliber.
- Voice & text
- Canvas & coding
- Proctored & secure
- Evidence-backed
Live discussion
The round changes while the candidate is still in it.
In the coding, architecture and simulation rounds a real conversation runs alongside the work — and new scenarios arrive part-way through, chosen from what the candidate has already said and built.
Because adaptability is not something a candidate can tell you about convincingly. It is something you either watch them do or you do not. A fixed exercise, however hard, only ever measures preparation.
- “Traffic is now ten times what you designed for.”
- “The primary datacentre is offline.”
- “We have been asked to make this multi-region.”
- “The read-to-write ratio has just inverted.”
- “Now do it in-place.”
- “Handle a stream of updates instead of a batch.”
The brief
Step 1 of a round.Their approach
Step 2 of a round.A constraint arrives
Introduced liveStep 3 of a round.They adapt, or they do not
Step 4 of a round.Something breaks
Introduced liveStep 5 of a round.Recovery, under discussion
Step 6 of a round.
- Prepared answers stop helping
- A rehearsed solution survives the opening question. It does not survive the constraint that arrives four minutes later.
- You see the recovery, not just the answer
- How somebody reacts when their design no longer works is closer to the job than whether their first design was right.
- Difficulty is configurable, and has no ceiling
- The same round runs for a mid-level engineer and for a CTO. What changes is the depth it goes to and the scenarios it introduces — set per role, before anyone sits it.
Adaptive interviewing in action
The interview adapts to the candidate.
Two candidates. Two interviews with almost nothing in common — different questions, different depths, different routes. One bar, applied to both.
Ground every interview covers
Introduction
Context and warm-up
Core concept
Base question
Deep dive
Probing questions
Application
Real-world problem
Adaptability
A new twist
Wrap-up
Summary and close
Not a running order. Every answer decides what happens next — press harder on the same thing, drop back to find where the understanding stops, or move on. Two candidates can cover the same ground and take entirely different routes through it.
Candidate A
Strong performer
4.6/ 5
Answer quality by area covered, out of 5
- 1
- 2
- 3
- 4
- 5
- 3.8
- 4.2
- 4.7
- 4.6
- 4.8
- 4.5
- Good grasp. Probed deeper.
- Strong understanding. Advanced follow-ups.
- Applied well. Complex scenario introduced.
- Handled the twist. Pressed on trade-offs.
Across the interview
How it adapted
- Asked deeper follow-ups
- Increased question complexity
- Introduced real-world scenarios
- Tested trade-off thinking
Excellent adaptability
Consistently high across every stage
Candidate B
Average performer
2.3/ 5
Answer quality by area covered, out of 5
- 1
- 2
- 3
- 4
- 5
- 2.6
- 2.3
- 2.1
- 2.4
- 2.0
- 2.4
- Basic understanding. Simpler follow-up.
- Concept unclear. Rephrased and broken down.
- Struggled to apply. Structured guidance given.
- Difficulty adapting. Easier twist to reassess.
Across the interview
How it adapted
- Simplified follow-up questions
- Rephrased for clarity
- Provided hints and scaffolding
- Adjusted scenario difficulty
Limited adaptability
More development needed before this level
What this shows
Answer quality is read as the interview runs, and the path changes to pull the best signal out of whoever is sitting it.
Read as it happens
Every response evaluated in the moment.
The path changes
Where it goes next follows the answers.
Right level of challenge
Not wasted on the strong, not punishing the rest.
Better signal
More to go on, and less noise around it.
Scores and commentary are illustrative. The horizontal axis shows ground covered, not a fixed running order.
Interview workflow
A complete, end-to-end interview journey.
From invitation to hiring recommendation — structured, adaptive and evidence-driven at every step, with a person in the loop before anything counts.
01Interview created
Define the role, the rounds, what is being measured, and the template it runs from.
Create interview
RoleSenior Backend EngineerTypeTechnical + System designTemplateBackend Engineer v2Create
02Candidate invited
A secure link, their choice of slot, and a preview of what the session involves.
CogniSense
You are invited to an interview
Senior Backend Engineer
Start interview
010203Candidate prepares
Guidelines, a system check and a walkthrough — so nothing on the day is a surprise.
System check
- BrowserOK
- MicrophoneOK
- CameraOK
- ConnectionGood
04Consent and identity
Identity confirmed, environment checked, and consent taken before anything is recorded.
Before we begin
- IdentityOK
- EnvironmentOK
- ConsentOK
- Integrity levelSet
030405The interview runs
Adaptive across every round the role needs, in voice or text, at the candidate's choice.
- Conversation
- Coding
- Design
- Simulation
- Behavioural
- Leadership
06Evidence captured
Every interaction stored as evidence — not as somebody's notes written up afterwards.
Captured
- Voice, videoYes
- CodeYes
- DiagramsYes
- SimulationYes
050607Organised against the role
Findings arranged against what this role was defined to need, dimension by dimension.
By dimension
08A person reviews it
A recruiter or hiring manager reads the evidence, adds their own, and validates the read.
Reviewer
Senior Engineering Manager
Strong on failure reasoning. Agree with the read.
Validated
070809The report is produced
Summary, dimension scores, evidence excerpts, transcript and a recommendation with reasoning.
4.3Overall,
by dimension- Summary
- Evidence
- Recommendation
10The decision is made
By your team, on evidence — with the rationale sitting underneath it if anyone asks.
Recommendation
Strong hire
Rationale attached. Confirmed by a person.
0910- 11
Export it anywhere
Circulate it, archive it, or push it straight into your ATS.
- DOCX
- CSV
- API
Consistent and fair
Every candidate held to the same standard, on the same criteria.
Faster, without cutting corners
More interviews in less time, at the same depth for each one.
Evidence, not opinions
Decisions rest on what was said and built, not on who remembers what.
What lands afterwards
Two reports. One of them is theirs.
Every score carries the candidate’s own words underneath it, and the transcript sits alongside. Nothing is summarised in a way you cannot go and check.
For your committee
Senior Backend Engineer
- Systems reasoning
- Trade-off judgement
- Failure thinking
- Code quality
- Communication
Evidence · Failure thinking
“I would fail the reads over to the replica and accept the staleness for a few minutes.”
Round 3 · System design · 00:21:14 — transcript attached
For the candidate
Their own feedback
Strongest area
Reasoning about failure and recovery under pressure.
Where it fell short
Cost implications of the multi-region design.
Worth reading next
Consistency models and replica lag budgeting.
Sent whether or not they move forward. Most of the industry sends nothing.
Non-negotiable
AI assists. People decide.
- A person decides
- No rejection on an automated score. Ever.
- Evidence or it does not count
- Every score carries the candidate's own words.
- Bias measured, not assumed
- Continuous bias audit; disparities flagged for review.
- Demographics never score
- Collected optionally, used only for audit.
- Integrity informs, never disqualifies
- A signal for a reviewer. Consent taken up front.
- Candidates can challenge it
- They can request the report and dispute a finding.
Two ways in
CogniSense is one of the two pillars CognityHire is built on.
Which means you can have it without the rest. Nothing about the platform requires you to buy recruitment from us.
Buy the platform
Subscribe to CogniSense on its own. Your team, your templates, your hiring managers — we run none of it.
For teams whose hiring already works, and whose interviews do not.
Buy the search
We run the whole loop, with CogniSense as the assessment layer inside it. You get shortlists, not software.
For roles you would rather hand over than staff.
How it is priced
You buy interview hours, not seats.
A subscription to a pool of hours you draw down as you hire. What it costs depends on the mix inside that pool — which is why there is no single number on this page.
A pool might look like this
100 interview hours
- 10hC-suite
- 20hSenior / principal
- 40hMid-level
- 30hEarly career
Text-basedOne rate for the pool above
Voice-basedA higher rate for the same pool
The seniority you interview for
A C-suite conversation and a first-round screen are not the same hour of work, and they are not priced as though they are. The blend inside your pool is what sets the rate.
Voice or text
Voice interviews cost more to run than text, so they carry a different rate. You choose per role, and most programmes run a mix of the two.
Straight answers
The questions that come up before a demo.
Can we customise what an interview measures?
Yes — that is the normal way to run it. Each role is defined by its own interview template covering which formats are included, what is being measured and how deep it goes. Templates are reusable across similar roles and versioned, so an interview from six months ago remains interpretable against the definition it was run under.
Can our interviewers overrule the platform?
Always, and on any individual finding rather than only on the overall result. The platform produces evidence and a recommendation with its reasoning attached; the decision is your team's. Rejecting someone on an automated score with no human in the loop is not a supported use of the product.
Does it handle coding and system design, or only conversation?
Conversation, live coding, system design on a drawing canvas, targeted knowledge checks, and scenario simulations such as debugging a failing service or triaging an incident — interleaved within a single session rather than run as separate tests.
How do you stop candidates from cheating?
Several weak signals are combined into an integrity read, and that read is given to a reviewer rather than used to disqualify anyone automatically. Candidates are told in advance what is captured and can opt for a lower-integrity format instead. We deliberately do not publish which signals we use.
Is candidate data secure, and what happens to it?
Collection is limited to what the assessment needs, consent is taken before anything is recorded, access is role-based and logged, and candidates can request their report and challenge what it says. Demographic data, where collected, is used only to audit outcomes and never as a scoring input.
Can we use the platform on its own?
Yes. It is used inside our own searches, and it is also adopted independently by teams who run their own hiring and want the interview stage to be structured and defensible. Nothing about it requires you to buy recruitment from us.
Who is it a poor fit for?
Teams wanting a quick keyword screen or a short behavioural quiz, and anyone hoping to automate the hiring decision itself. It earns its keep on technical and judgement-heavy roles, at enough volume that consistency matters — and it makes hiring managers better rather than absent.
See it run against a role you are actually hiring for.
Thirty minutes. Bring the role you are struggling with and we will run a real interview end to end — the candidate side, the report, and the evidence behind it.