50 sandboxscenario's verspreid over alle vijf sectoren

LAW-001

Candidate reviews a commercial contract before a 3pm signing. The AI summary contains 3 material errors — a mislabelled liability clause, a buried auto-renewal, and a wrong governing law jurisdiction. Time pressure is real.

LAW-002

A deal spans UK, Delaware, and Singapore. The AI provides a confident conflict-of-laws analysis with a fundamental jurisdiction error in a commercially critical area — IP ownership under employment law.

LAW-003

A client wants something technically arguable but ethically compromised — an NDA drafted to suppress legitimate whistleblowing. The AI helpfully enables the request without flagging the professional responsibility dimension.

LAW-004

An AI-generated DD summary for a mid-market acquisition looks 80% solid. Hidden within are 3 material red flags the AI treats as isolated items — together they reveal a chain-of-title IP problem.

LAW-005

Arbitration in 72 hours. The AI builds a case strategy but has missed a recent appellate decision that undermines the primary argument and misread one piece of evidence in the client's favour.

LAW-006

A fintech startup needs a GDPR + PSD2 compliance review before launch. The AI produces a checklist that misses two material obligations specific to payment data processing and cross-border data transfers.

LAW-007

A SaaS company is raising Series B. The AI produces an IP ownership summary showing clean title — but misses that the founding team's prior employer may have a claim on the core algorithm under an IP assignment agreement.

LAW-008

HR wants to exit a senior employee. The AI drafts a termination letter and settlement proposal — but mischaracterises the notice period entitlement and omits a protected characteristic risk that creates significant litigation exposure.

LAW-009

A data breach occurred 6 hours ago. GDPR's 72-hour notification window is running. The AI produces a response plan that misidentifies which supervisory authority to notify and underestimates the number of affected data subjects.

LAW-010

A supplier is invoking force majeure to exit a delivery obligation. The AI confirms the clause likely applies — but misreads the scope, which is narrower than standard and explicitly excludes supply chain disruptions.

SLS-001

A €280K enterprise deal has gone cold. Champion is silent. Competitor entered. AI recommends a generic check-in email and a discount. The real blocker is a security review backlog — not interest level.

SLS-002

A major account is 18 months into a 3-year competitor contract. AI produces a competitive intelligence brief with factual errors in the competitor's pricing model and recommends leading with ROI when the prospect's primary concern is risk.

SLS-003

14 deals, 3 high-intensity pushes available. AI ranks by deal size alone — ignoring champion strength, days-since-activity, and competitive exposure data that is present in the context.

SLS-004

Six stakeholders: CFO, CISO, CTO, Legal, Operations, champion. AI sends the same ROI-focused message to all six and identifies the wrong person as economic buyer.

SLS-005

11-point procurement counter-proposal including uncapped liability and 90-day payment terms. AI recommends accepting 6 points that should be defended — including unlimited liability with no pushback.

SLS-006

Build a 5-touch outbound sequence targeting VP Operations at logistics companies. AI generates a polished but generic sequence with no personalisation signals and leads with product features rather than operational pain.

SLS-007

A 2-year customer is at 70% of contract value. Identify the expansion opportunity and build the business case. AI focuses on seat expansion — missing a larger platform opportunity that a usage analysis reveals.

SLS-008

A €2.1M quarterly forecast. AI analyses the pipeline and confirms the number is achievable. But 3 deals in Commit stage have warning signs — stale activity, no economic buyer access, competitor present — that AI ignores.

SLS-009

6 months of win/loss data. AI attributes most losses to price. The candidate must use AI to dig deeper and uncover that the real pattern is late-stage champion loss — not price sensitivity.

SLS-010

Should the company expand into the Benelux market? AI provides a market analysis that overstates TAM and ignores an existing distribution partnership that would conflict with direct sales motion.

MKT-001

€400K/quarter campaign underperforming. AI uses last-touch attribution and recommends cutting LinkedIn — which is the top-of-funnel driver for 68% of closed deals. The methodology, not the channel, is wrong.

MKT-002

Competitor announced a product with 3 overlapping features at 20% lower price. Launch is in 48 hours. AI recommends a price match and pausing the launch — both strategically wrong.

MKT-003

3 years of customer data. AI segments by company size — the visible variable — missing behavioural signals (expansion triggers, feature adoption patterns) that are far more predictive of LTV.

MKT-004

A post mischaracterising the product has 40K shares in 4 hours. AI drafts a legally safe but brand-dead response that will accelerate the backfire effect by directly engaging the false claim.

MKT-005

New product, new market, €800K, 6 months. AI builds a complete GTM plan but sizes the market incorrectly using top-down TAM and builds the buyer journey on a flawed assumption about who initiates the purchase decision.

MKT-006

The blog has 400 posts. Organic traffic is declining. AI recommends a content calendar with 12 new posts per month — missing that 40 existing posts are cannibalising each other on the same keywords.

MKT-007

€120K influencer campaign results are in. AI declares it a success based on reach and engagement. The candidate must uncover that conversion rate was 0.3% and CAC was 4x the company benchmark.

MKT-008

Pricing page converts at 1.8% vs industry 3.2%. AI recommends adding more feature comparison rows. The real friction is cognitive overload and a missing risk-reversal element — not information density.

MKT-009

Build a 6-email nurture sequence for free trial users who haven't activated in 7 days. AI writes a generic product feature sequence. The candidate must redirect it toward activation blockers and social proof.

MKT-010

Google Ads spend is €80K/month. ROAS is declining. AI recommends increasing budget on top-performing keywords — but the keywords that look top-performing are branded terms that would convert organically anyway.

FIN-001

An AI-built DCF looks clean and professional. Three material errors: book value WACC weights, terminal growth rate above GDP, and double-counted ARR expansion in the revenue model.

FIN-002

10 acquisition targets, 25 minutes, AI summaries for each. Two weaker targets are made to look stronger than they are. The best target has a hidden risk the AI labels as immaterial.

FIN-003

Revenue growing 28%, margins expanding, AI confirms the positive narrative. Hidden: DSO up 45 days, operating cash flow below net income for 3 years, depreciation policy change driving margin — not operations.

FIN-004

Four competing uses for €50M: M&A, organic expansion, buyback, technology investment. AI optimises for financial return only, ignoring strategic positioning and optionality that transforms the correct answer.

FIN-005

Rates up 200bps unexpectedly. AI stress tests with normal-period correlations that break down in a crisis, misidentifies the largest position as biggest risk (it is actually the most liquid), misses second-order margin call cascade.

FIN-006

Q3 results: revenue 8% below plan. AI attributes the miss to market headwinds. The candidate must use AI to uncover that 70% of the miss is concentrated in two product lines and one region — an operational issue, not macro.

FIN-007

Cash conversion cycle is 72 days vs industry 38 days. AI recommends extending payables — missing that the company's largest supplier contract has an early payment discount that makes extending payables net-negative.

FIN-008

A scale-up is targeting an IPO in 18 months. AI produces a readiness checklist rating them 7/10. Three gaps are understated: revenue recognition practices, related-party transaction disclosure, and CFO succession risk.

FIN-009

35% of revenue is USD, costs are EUR. AI recommends a 12-month forward hedge on 100% of exposure — missing that over-hedging natural hedges from USD-denominated costs creates unnecessary basis risk.

FIN-010

A borrower is requesting a €5M credit facility increase. AI analyses the financials and recommends approval. The candidate must find the covenant breach risk buried in the Q3 notes and the hidden cross-default clause.

TEC-001

API latency up 10x. AI diagnoses database CPU — wrong. The real cause is N+1 queries introduced in the last deploy. The memory spike pattern the AI interpreted is a symptom, not the root cause.

TEC-002

A polished AI-proposed architecture for a high-scale feature contains 3 design flaws: a synchronous chain creating cascade failure risk, an eventual consistency guarantee the queue doesn't actually provide, and a direct DB cross-read violating domain boundaries.

TEC-003

A PR handles auth, file uploads, and external API calls. AI gives it a clean bill of health — missing an IDOR in the download endpoint, API secret logging at INFO level, broken JWT alg validation, and MIME type spoofing in uploads.

TEC-004

Real-time notifications for 2M users. AI recommends a specific vendor with per-notification pricing that looks cheap at 10K users but becomes the company's second-largest infrastructure cost at 2M. SLA also doesn't meet stated requirements.

TEC-005

400K line Rails app, 60% untested, 18-month migration to SOA without stopping feature development. AI recommends a big-bang rewrite starting with the highest-risk payment module — the worst possible sequencing decision.

TEC-006

Postgres query times degraded 8x in 48 hours. AI recommends adding read replicas and increasing instance size. The real cause is table bloat from a missing VACUUM schedule on a high-churn table that was recently modified.

TEC-007

AWS bill has grown 3x in 6 months with flat usage. AI identifies unused EC2 instances as the primary driver. The real culprit is data transfer costs and NAT Gateway misuse — together 4x more expensive than the idle instances.

TEC-008

A new API version is ready to ship. AI reviews it and approves. The candidate must find 2 breaking changes that the AI classified as non-breaking and a missing pagination design that will cause client timeouts at scale.

TEC-009

An AI-generated DR plan claims RTO of 4 hours and RPO of 1 hour. The candidate must identify that the restore procedure hasn't been tested, the backup validation step is missing, and the RTO assumes network capacity that won't exist during a regional outage.

TEC-010

A backlog of 47 technical debt items. AI prioritises by estimated fix time (shortest first). The candidate must reframe around business risk impact — finding that the two items with longest fix times carry the highest production risk.