Cerebre: controlled decisions
Cerebre is our platform for data analytics, predictions, and decision support in regulated workflows. It helps teams make consistent decisions and shows why a decision was made.
Built for regulated work: secure, compliant, and easy to explain. It combines LLMs with guardrails from traditional AI models, then lets teams apply scoring, routing, and review paths without changing application code.
Core capabilities:- Data analytics, predictions, and scoring
- LLMs with guardrails from traditional AI models
- Thresholds, caps, multipliers, and scoring weights
- Routing and escalation rules
- Checks for completeness, missing data, and required documents
- Decision reasons and explanation text
- Version history and approvals before changes go live
- Testing on historical cases before rollout
- Forecasting and simulation for staffing, cost, and performance planning
Cerebre in use
The examples below show production use in insurance and cross-industry meeting support, plus analytics work on remuneration and call-center planning.
The impact figures below use simple operating assumptions. Specific assumptions are shown with each example.
Pricing is indicative and is confirmed after a brief initial workshop.
Production: Policy amendment decision support (insurance)
What's the problem:
What Cerebre does:
What's the impact:
Typical engagement:
Proven in practice:
Notes:
- After a claim, insurers need to reassess policies and decide on premium increases, remediation actions, or policy termination.
- Decisions rely on multiple rules and signals such as claims history and risk metrics.
- No clear portfolio view across household members.
- Inconsistent application of underwriting guidelines, with limited transparency for customer communication and audit.
What Cerebre does:
- Cerebre supports consistent, data-driven policy decisions by combining internal data, rules, and predictive models.
- It pulls policy, claims, premium, and household data from production systems, evaluates risk using configurable logic, and recommends actions such as premium adjustments, remediation, or policy exit.
- Each decision includes a clear rationale, supporting data, and structured outputs for review and communication.
What's the impact:
- Estimated annual impact of about CHF 180k-315k (assumes 450 cases per year, CHF 400-700 improvement per case through pricing, remediation, or risk reduction).
- Consistent application of underwriting and risk policies.
- Improved portfolio quality and loss-ratio steering.
- Reduced manual case-by-case interpretation.
- Transparent, defensible decisions for customers and audit.
Typical engagement:
- Initial workshop (data, rules, inputs): ~1-2 weeks
- Prototype (models, logic, workflow): ~2-6 weeks
- Optional validation on historical cases: depends on data availability
- Pricing: starting from CHF 0 (setup), ~CHF 25k/year
- Next step: share a sample decision workflow or dataset. We'll assess feasibility and estimate potential impact.
Proven in practice:
- Consistent decision-making across underwriting teams.
- Better portfolio steering through pricing and risk selection.
- Reduced reliance on manual interpretation of rules.
- Clear reasoning for internal review and customer communication.
- Ability to test threshold and policy changes before deployment.
Notes:
- Cerebre can run standalone or combined with Myriad for document extraction, analytics, and end-to-end decision support.
Production: Remuneration and incentive modeling (insurance & banking)
What's the problem:
What Cerebre does:
What's the impact:
Typical engagement:
Proven in practice:
- Remuneration logic is often fragmented across spreadsheets, legacy tools, and operational systems.
- Spreadsheet-based models do not scale reliably and make it difficult to test changes on real production data.
- Leadership lacks clear visibility into how model changes impact costs, payouts, and incentives before rollout.
What Cerebre does:
- Cerebre ingests and structures production data across roles, salary components, and bonus schemes, and rebuilds the existing remuneration logic as a baseline model.
- Teams can simulate changes by adjusting targets, weightings, thresholds, and caps, and instantly see the impact on cost base, payout distribution, and incentive structures.
What's the impact:
- Estimated annual impact of about CHF 125k-250k (assumes CHF 250k-500k improvement potential from optimized remuneration models).
- Clear visibility into cost and incentive tradeoffs.
- Faster and more reliable scenario testing than spreadsheets.
- Decision-ready outputs for executive and board-level discussions.
- Stronger alignment before model rollout.
Typical engagement:
- Baseline model (rebuild existing logic): ~4-8 weeks
- Scenario modeling and iteration: ongoing during redesign phase
- Pricing: ~CHF 40k setup, ~CHF 20k/year (depends on number of roles, remuneration scenarios, complexity)
- Next step: share some anonymized remuneration sample data and describe two scenarios. We'll assess feasibility and estimate potential impact.
Proven in practice:
- Reliable replication of existing remuneration models.
- Rapid scenario testing to support strategic redesign.
- Strong analytical foundation for executive decision-making.
- Used as backbone for large-scale remuneration reviews.
Call-center demand forecasting and FTE planning
This model was implemented in a client project and can be deployed as a reusable planning system.
What's the problem:
What Cerebre does:
What's the impact:
Typical engagement:
Proven in practice:
What's the problem:
- Customer demand fluctuates seasonally, making it difficult to plan staffing levels in advance.
- Teams must balance service levels, backlog, utilization, and cost, often without a clear, unified view.
- This leads to over- or under-staffing, inconsistent service, and reactive decision-making.
What Cerebre does:
- Cerebre was used to combine historical demand, staffing, backlog, and service data into a unified planning model.
- It forecasts customer demand across periods, queues, and seasonal patterns, and translates these forecasts into FTE requirements under different assumptions.
- Teams were able to simulate scenarios such as hiring, outsourcing, overtime, or automation and see the impact on service levels, backlog, and cost.
What's the impact:
- Estimated annual impact of about CHF 125k-210k (assumes 3-5 FTE optimization and about CHF 85k loaded cost per FTE).
- More accurate FTE planning ahead of seasonal peaks.
- Clear tradeoffs between capacity, cost, and service levels.
- Reduced over- and under-staffing, with more predictable operations and workload distribution.
Typical engagement:
- Data model and initial forecasts: ~2-4 weeks
- Capacity planning and scenario layer: ~1-2 months
- Model tuning: depends on data quality and seasonality
- Pricing: ~CHF 40k setup, ~CHF 10k/year
- Next step: share sample operational data (demand, staffing, service levels). We'll assess feasibility and estimate potential impact.
Proven in practice:
- Improved staffing decisions ahead of demand peaks.
- Better balance between service level targets and cost.
- Supported redesign of call-center capacity and staffing mix.
Production: Automated meeting minutes and summaries
What's the problem:
What Cerebre does:
What's the impact:
Typical engagement:
Proven in practice:
Notes:
- Teams spend significant time capturing meeting notes, action items, and formal minutes.
- Important objectives are often missed during the meeting, and post-meeting documentation is time-consuming and inconsistent.
- Many organizations also require structured minutes aligned with internal or customer-specific templates.
What Cerebre does:
- Cerebre generates a live transcript during meetings and highlights which agenda points or objectives have not yet been addressed.
- Immediately after the meeting, it produces structured minutes and summaries aligned to the organization's preferred format.
- The transcript is used only during processing and is not stored.
What's the impact:
- Estimated annual impact of about CHF 8k-15k per user (assumes about 10 meetings per week, 15-30 minutes saved per meeting, and CHF 55-60 per hour).
- Reduced time spent on note-taking and minute drafting.
- Immediate availability of structured meeting minutes.
- Better coverage of objectives during meetings.
- More consistent and professional documentation.
Typical engagement:
- Setup (templates, objectives, structure): ~1-2 weeks
- Pricing: CHF 3-5k setup, then from CHF 5 per meeting hour (typically CHF 100-200 per user/month)
- Requirements: lightweight client installation on user devices
- Optional: external microphone for improved audio quality
- Next step: run a live demo on one of your meetings. We'll assess fit and expected impact.
Proven in practice:
- Immediate generation of structured minutes after meetings.
- Real-time support for agenda coverage during discussions.
- Works across industries and meeting formats.
- Reduces post-meeting administrative workload.
Notes:
- Data is processed within Switzerland.
- Audio is not stored; transcripts are deleted after summary generation.