Turn every service into evidence for better decisions.
An AI decision-support platform in development for independent restaurants, pubs and bars. Forecast demand, compare costed options and build a record of what works for your venue—with the manager always in control.
Bridging Shift Floor Intuition with Decision Evidence
Hospitality operators face relentless volatility every service. PassLedger AI is built to record decisions, capture context, and measure what actually works.
The Operational Problem
In busy restaurants and bars, managers make dozens of high-stakes operational choices under intense pressure: calling in extra staff, cutting shifts early, reconfiguring dining sections, or pushing perishable inventory. Yet these decisions remain disconnected from bookings and sales records. Knowledge stays trapped in individual heads—and vanishes whenever managers move on.
Our Decision-Learning Approach
Forecasting alone doesn’t solve service friction. PassLedger AI focuses on the decision loop: presenting transparent operational choices, recording the manager's actual decision and override reasons, and measuring the outcome against expected baselines. Over time, the venue builds structured, durable evidence of which interventions genuinely succeed.
Transparent Operational Choices
Costed options with explicit assumptions, trade-offs, and confidence ranges rather than black-box instructions.
Manager Approval & Overrides
The human manager always approves, modifies, or declines recommendations. Override rationales are recorded to inform future models.
Measured Outcomes vs Baselines
Post-service debriefs compare realised covers, labour hours, and turnover against counterfactual baselines.
Preserved Venue Knowledge
Institutional operating wisdom survives team turnover, ensuring incoming general managers inherit hard-won service playbooks.
Deval Karnik
Founder · Hospitality Operations & Product Definition
Deval is a London-based hospitality professional and Supervisor / closing manager at The Orange, Cubitt House, Belgravia. His perspective is grounded in live service reality: running entire shifts, coordinating high-volume covers (up to 250 covers per shift), leading 6–7 team leaders, floor planning, kitchen-to-floor coordination, wine upselling, staff training, and end-of-shift reconciliation.
Context note: References to Deval’s employer describe his operational background on the floor. They do not imply that Cubitt House or The Orange is a customer, commercial partner, or endorser of PassLedger AI. Deval is not a software or machine-learning engineer. PassLedger AI is a working brand; name clearance and company structure remain to be confirmed.
Historical Prototype Evaluation & Research Evidence
Empirical findings from initial offline prototype diagnostic work
Evidence Disclaimer: Forecast error metrics (MAPE) reflect offline retrospective model validation and do not constitute an accuracy guarantee. These findings do not establish proven live savings, production operational performance, or paying customer endorsements.
Nine Purpose-Built Modules for Hospitality Operations
From basic data hygiene to intervention learning and multi-site views, each module is designed around real shift constraints. Clearly labelled by development stage.
Data Intake & Data Health
Imports CSVs and, as developed, authorised API data. Detects missing dates, duplicate transactions, capped reservation exports, and inconsistent service records. Excludes unnecessary personal guest information.
Demand, Footfall & External Signals
Forecasts covers, walk-ins, sales, and 15-minute pressure windows using historical service data, bookings, local events, and weather forecasts. Displays uncertainty ranges alongside clear driver explanations.
Decision Ledger & Daily Brief
Presents service forecasts and two to four operational options. Records Accept, Adjust, or Ignore decisions, captures manager override reasons, logs the actual implemented actions, and pairs them with post-shift outcomes.
Notice-Aware Labour Engine
Compares staffing options using configurable notice windows, decide-by deadlines, and potential late-change costs. (Guaranteed-hours exposure tracking is a planned expansion. Does not provide legal advice or statutory compliance claims).
Service Twin & Consequence Engine
Simulates service under different staffing options using arrivals, venue capacity, dwell time, and staffing assumptions. Compares labour cost, service pressure, and lost-cover risks. (Detailed kitchen & table-turn modelling is future work).
Perishable-to-Plate
Links ingredient expiry risk to preparation changes, daily specials, front-of-house upselling, and future prep ordering. Evaluates and measures sell-through, avoidable food waste, margin impact, and kitchen prep pressure.
Cellar Capital Release
Identifies dormant wine and bar stock tied up in storage. Evaluates by-the-glass listings, curated food pairings, staff incentive targets, price adjustments, and supplier returns against sales velocity and profit margin.
Intervention Learning & Peer Priors
Builds on matched-comparison estimates with venue-specific effect learning, A–D evidence confidence badges, and anonymised comparable-venue priors. Proactively withholds strong recommendations whenever empirical evidence is weak.
Multi-Venue & Live Integrations
Multi-site group overview dashboards, role-based access permissions, monitored automated connectors, and direct API integrations with leading hospitality enterprise systems across 1–15 sites.
Planned Operational Interface Features
A tool built for busy shift managers on mobile and tablet devices during service setup:
The Six-Stage Operating Loop
How PassLedger AI integrates into daily hospitality shifts to transform raw service data into actionable, venue-specific evidence.
Predict
Validate venue history, reservation books, local calendar events, and weather forecasts to project the upcoming shift’s demand curves with clear uncertainty ranges.
Explain
Translate underlying statistical signals into plain-English demand drivers: why covers are anticipated to peak, which sections face pressure, and what assumptions are applied.
Recommend
Present two to four feasible, costed operational choices (e.g., call in extra runner, flex section rota, push specials) with transparent trade-offs and evidence strength.
Decide
The manager always retains the final decision. Choose to Accept, Adjust, or Ignore recommendations, and record the rationale and actual implemented action.
Measure
Capture actual covers served, actual labour hours clocked, realised turnover pace, beverage sales velocity, and avoidable waste after the service wraps up.
Learn
Compare observed outcomes against expected baselines and update future recommendations. Over time, the model adapts to your venue's unique kitchen and floor dynamics.
Statistical Rigour & Correlation vs Causality
In live restaurant operations, a favourable outcome following an operational action does not, by itself, prove that the action caused the improvement. External factors—such as unexpected walk-ins, weather shifts, or unrecorded events—routinely influence service performance. PassLedger AI uses matched baselines and cautious evidence grading to avoid false attribution.
Proposed Pilot Journey
A phased, low-friction pathway from initial qualification to live staffing validation.
Initial Qualification
Quick review of venue type, site count (1–15 sites), existing POS and reservations systems, and key operational pain points.
Data-Readiness & DPA
Audit of data hygiene, export compatibility, and execution of a robust Data-Processing Agreement (DPA) prioritizing data minimisation.
Setup & Historical Diagnostic
Backtesting models against historical exports to calculate venue baseline error and identify historical rota mismatch windows. Provided free of charge.
Live Staffing Pilot
Daily decision briefs delivered prior to shifts. General managers review, accept, or override staffing options in real-world service.
Outcome Review & Adoption
Comprehensive review of measured outcomes, manager override trends, and staff adoption. Potential conversion to ongoing subscription where appropriate.
Transparent, Venue-Based Pricing
Predictable monthly pricing designed for independent operators and small multi-site hospitality groups.
FOUNDING VENUE — PILOT
For forward-thinking operators joining early stage live validation.
- Forecast
- External signals
- Decision brief
- Structured feedback
CORE
Essential decision ledger for single-venue restaurants and bars.
- V1/V2 forecasts
- Daily brief
- Options
- Decision ledger
OPERATIONS
Dynamic rota analysis and perishable inventory optimisation.
- Everything in Core
- Notice-aware labour engine
- Service Twin
- Perishable-to-Plate
INTELLIGENCE
Advanced cellar capital release and cross-venue peer priors.
- Everything in Operations
- Cellar Capital Release
- Evidence dashboards
- Peer priors
MULTI-SITE
Tailored multi-unit deployment for groups operating 2–15 venues.
- Group dashboards
- API integrations
- Onboarding
- Support
Features will be introduced in stages. A setup fee may apply to complex multi-site deployments; no fixed setup fee is specified.
Clear, Transparent Answers
Everything you need to know about our approach, data handling, and pilot process.