What You'll Get — Drive Preview
Preview the Capgemini preparation material folders and documents before opening the complete materials page.
Click a preview to view it larger. Use the controls to browse or pause the carousel.
Every stage explained in full — what changed from 2026, what's actually tested, worked examples for debugging and AI-assisted coding, sample prompts, and a week-by-week roadmap to prepare.
Online assessment (stages 1-5) → interviews (6-7)
A quick primer before we go stage-by-stage.
From your campus application to the offer letter — nine checkpoints, five of which are online-assessment elimination rounds.
Nodes 2-6 are the Online Assessment. Nodes 7-8 are live interviews, which only happen after all five online stages are cleared.
Not just a headline list — here's exactly what changed round by round, and what it means for how you prepare.
| Area | 2026 Legacy | 2027 Reported | What it means for you |
|---|---|---|---|
| Aptitude / Pseudocode | Standalone pseudocode-heavy technical MCQs | Folded into the Technical Module's "Technical Assessment" section Merged | Pseudocode interpretation still matters, but it's now one of several sub-topics rather than the whole section. |
| AI Knowledge | Not tested at all | 20 of 40 Technical Module questions — LLMs, prompt engineering, RAG, agentic AI New | You now need to study AI concepts the same way you'd study DBMS or OOP — this is graded like any other MCQ topic. |
| Coding Round | 2 questions to code from scratch, 45 min | Split: 1 Debugging Q (~20 min) + 1 AI-Assisted Coding task (~45 min) Split & changed | You're no longer judged purely on writing code unaided — half of this is now about fixing existing code, the other half about directing an AI well. |
| Cognitive Test | 4 games picked from a pool of 24, unnamed | 4 specific named modules: Motion, Grid, Logical Reasoning, Behaviour Standardized | You can practice against a known, fixed set of 4 game types instead of guessing which of 24 you'll get. |
| Behavioural Assessment | Separate 20-25 minute PowerSkills test | Folded into the Cognitive Assessment as its 4th module | One less standalone round to sit through — behavioural questions now come at the end of the cognitive games. |
| English Communication | 30 MCQs, 30 minutes, largely reading/grammar | Full LSRW (Listening, Speaking, Reading, Writing), AI-proctored, ~30 min | Speaking is new and significant — practice speaking answers aloud, not just reading comprehension. |
| Interviews | Technical Interview → HR Interview | Technical Interview → HR Interview (unchanged, per reported material) | No reported change here — prepare as you would have for previous years' interviews. |
Scroll down for a deep dive into each stage — what's tested, what changed, worked examples, and preparation tips.
Listening, speaking, reading & writing
~30 MINAI Literacy + Technical Assessment
~40-50 MINFind & fix a broken code snippet
~20 MINDirect an AI assistant to solve a task
~45 MINGame-based reasoning & behaviour
~20-30 MINThis is the gatekeeper round — every candidate starts here, and nobody moves on without clearing it. An AI system scores your listening, speaking, reading, and writing in real time.
2026's English test was 30 MCQs focused on reading and grammar. 2027 tests full LSRW — Listening and Speaking are new additions, scored by an AI proctor rather than pure multiple choice.
Prompt: "Speak for 60 seconds about a time you solved a difficult problem in a team project."
What's graded: Clarity of structure (situation → action → result), fluency without long pauses, and natural pronunciation — not memorized scripts.
This is where the 2027 pattern diverges most from previous years — half the section is now AI Literacy, not just core CS fundamentals.
2026's technical section was 100% programming logic and pseudocode. 2027 keeps the same 40-question, ~45-minute format but replaces half the questions with AI Literacy — a topic that didn't exist in the syllabus before.
Q: "An LLM gives you an answer with no cited source. What should you do before using it?"
Expected reasoning: Validate the claim against a trusted/authoritative source rather than accepting it at face value — this tests output-validation awareness, not memorized facts.
You're handed broken code and one shot to fix it. It's short, but the snippets reportedly touch advanced topics — this round rewards depth over speed.
Debugging didn't exist as its own round in 2026 — it was implicitly part of the 2-question coding round. In 2027 it's a standalone ~20-minute round focused purely on fixing existing code, not writing from scratch.
The function below is supposed to sum all elements of an array but returns a wrong (too-small) result:
Fix: change i < n - 1 to i < n — a classic off-by-one boundary error, exactly the kind of bug this round tests.
The single biggest departure from every prior Capgemini pattern: you don't write the solution alone — you direct an AI assistant to build it, and you're graded on how well you steer it.
This round simply didn't exist before. In 2026, one of your two coding questions had to be solved entirely unaided. In 2027, this round explicitly expects — and grades — AI usage, replacing "can you code alone" with "can you code well with AI."
Human analyzes the problem & picks an approach
Human gives structured context & constraints
AI produces an initial code solution
Human inspects output, fixes weak logic
Human tests edge cases & correctness
AI Literacy: Do you understand and interpret the problem correctly before prompting?
Prompt Quality: Is your prompt structured and specific, or a vague one-liner?
Problem-Solving: Do you guide the AI step-by-step toward the right approach?
Review & Adapt: Do you catch and correct wrong AI output, or copy-paste blindly?
Four short interactive modules, reportedly delivered on AON's ADEPT-15 assessment platform. No coding here — this measures planning, memory, reasoning, and workplace behaviour.
2026 picked 4 games at random from a pool of 24 unnamed games, plus a separate PowerSkills/behavioural test. 2027 reportedly standardizes this to exactly 4 named modules and folds behavioural questions into the same round.
Track and predict the movement of objects across a grid to identify patterns and future positions.
Use spatial awareness and visualization to recall and solve grid-based pattern challenges.
Identify patterns, relationships and the next logical step in a rule-based sequence.
Situational workplace prompts about your thought process, attitude, and approach to real-life scenarios.
Reaching here means you've cleared all five online-assessment stages. These two rounds are reported as unchanged from previous years.
"Walk me through the database design of your final-year project — why did you choose this schema over alternatives?"
"Tell me about a time you disagreed with a teammate. How did you resolve it?"
How you move from one stage to the next — and what can knock you out.
Suggested pacing — not an official Capgemini plan. Adjust it to your own drive date.
Preview the Capgemini preparation material folders and documents before opening the complete materials page.
Click a preview to view it larger. Use the controls to browse or pause the carousel.
The Capgemini materials page lists preparation packs with actual-question practice, core assessment topics, and interview preparation. Open it to review the full package details and place an order.
2000+ students trust TechKalaA for placement prep — let's get your Capgemini Exceller 2027 prep on track.