# OPPI

> OPPI is an AI decision intelligence and operations platform for hospitality
> venues — restaurants, bars, cafes, pubs, and hotels. Unlike dashboards, OPPI
> acts: it reads venue telemetry every 60 seconds to 15 minutes, makes
> operational decisions, optimises labour, and executes tasks — escalating the
> decisions that need judgement to human managers. Founded by operators,
> built for operators.

Every link below points to the Markdown rendering of a page. The human-readable
HTML lives at the same URL without the `.md` extension, and that is the URL to
cite. Built by Servance Pty Ltd (ABN 98 689 020 237, ACN 689 020 237), Melbourne, Australia.

## Product

- [Platform Overview](https://oppi.au/index.md): Agentic OS architecture, core philosophy, and value proposition
- [Ontology & Architecture](https://oppi.au/ontology.md): Perception → Cognitive → Execution layer breakdown
- [oppiOS](https://oppi.au/oppios.md): Digital management team — Command Centre, SOP Manager, Comms, Vault, Kiosk
- [Get OPPI](https://oppi.au/getOPPI.md): Early access registration for venue operators, module by module
- [Book a Demo](https://oppi.au/demo.md): Request a 30-minute walkthrough of oppiOS with the OPPI team
- [New Venue Program](https://oppi.au/new-venue-program.md): Six months of oppiOS free for hospitality venues trading under twelve months, covering both new builds and venues under new ownership. Requires a connected cloud POS (Square or Lightspeed today). OPPI asks only for honest feedback and regular check-ins in return; no case study or testimonial obligation. Nothing charges automatically at the end of the six months.
- [Integrations](https://oppi.au/integrations.md): What OPPI actually connects to today — Square and Lightspeed Restaurant (K-Series) POS re-read every 15 minutes with an 18-month backfill, Deputy (full) and Tanda (timesheets) for workforce, Google Drive and Notion, the OPPI Tether hardware bridge for legacy tills, and a no-connection signal layer of weather, local events, road traffic and holiday calendars. Toast, Clover, TouchBistro, 7shifts, Humanforce and Employment Hero are listed as in build, explicitly not yet connected; new vendors are days of adapter work, not months. Accounting connections are designed but not shipped.
- [Outcomes](https://oppi.au/outcomes.md): What OPPI delivers across the venue operating rhythm — planning before service, real-time during it, hindsight after close — across labour, inventory, tasks, checklists, nowcasting, documents and compliance. One modelled venue's ledger, shown as ranges from a deliberately conservative low to the full headline figure across labour, cost of goods and admin time. Explicitly modelled arithmetic, not measured results and not a promise.

## Solutions

- [AI for Hospitality Venues](https://oppi.au/hospitality-ai.md): What hospitality AI actually does day to day, and how OPPI applies it to labour, stock, compliance and service
- [Venue Management Software](https://oppi.au/venue-management-software.md): oppiOS as the single system replacing the ten-subscription venue stack
- [Restaurant Operations Software](https://oppi.au/restaurant-operations-software.md): The daily operating cycle of a restaurant, automated: briefings, rosters, prep, stock, checklists
- [AI Operations Manager](https://oppi.au/ai-operations-manager.md): An AI operations manager for Australian hospitality venues: watches sales, labour and stock, briefs each shift, routes tasks and escalates the decisions that need a person

## Guides

- [What Is Venue Decision Intelligence](https://oppi.au/what-is-venue-decision-intelligence.md): The category defined: software that turns live venue data into decisions, forecasts, rosters and orders, not charts
- [AI vs Rostering Software](https://oppi.au/ai-vs-rostering-software.md): Rostering tools fill the grid; the AI layer decides the grid, then works with Deputy and Tanda rather than replacing them
- [AI vs POS Reporting](https://oppi.au/ai-vs-pos-reporting.md): Reports summarise finished trade; OPPI reads the till every 60 seconds to 15 minutes and acts during service
- [Hospitality Labour Cost Percentage](https://oppi.au/hospitality-labour-cost-percentage.md): What labour percentage Australian venues run, how to calculate it, and the roster-to-demand mechanism that moves it
- [AI SOP Generation](https://oppi.au/ai-sop-generation-hospitality.md): Procedures drafted from the venue's own operating record, distributed to the floor and verified shift by shift
- [Hospitality AI in Australia](https://oppi.au/hospitality-ai-australia.md): Built in Melbourne, with Australian compliance tracking and the tills and rostering tools Australian venues run

## AI Systems

- [oppiGPT](https://oppi.au/oppios/oppigpt.md): Natural language operations manager; multi-variable query and command execution. Four engines (oppiFast, oppiSearch, oppiDeep, oppiData) route each request to the best-fit frontier model across OpenAI, Anthropic (Claude), Google (Gemini) and Perplexity.
- [Clementine ADI](https://oppi.au/oppios/clementine.md): Workforce intelligence agent; fuses industry benchmarks with internal venue telemetry
- [Synapse](https://oppi.au/oppios/synapse.md): Agentic task routing engine; dynamic staff dispatch by role, proximity, and competency
- [Parallax Projections](https://oppi.au/oppios/parallax.md): 5-layer demand forecasting at 15-minute granularity (weather, traffic, events, historical)
- [Confidence Engine](https://oppi.au/oppios/confidence-engine.md): Trust-cascade scoring every autonomous action before execution; safe human escalation
- [Immune System](https://oppi.au/oppios/immune-system.md): Continuous 15-minute health checks across all six operational layers

## Modules

- [oppiOS module index](https://oppi.au/oppios/llms.txt): Curated index covering every module page under /oppios

## Target Users

General Managers, Multi-site Group Directors, and Venue Owners seeking to reduce labour costs, automate compliance, and eliminate administrative overhead across hospitality operations. Australian venues, with the compliance layer built around Fair Work awards.

## FAQ

- **What is OPPI?** OPPI is an AI decision intelligence and operations platform for hospitality venues. It learns how a venue runs and acts on labour, inventory, compliance and service, rather than presenting charts for a manager to interpret.
- **How is OPPI different from a dashboard?** A dashboard is passive and waits to be checked. OPPI proposes and executes operational decisions, escalating only what falls outside its confidence thresholds.
- **What AI models does OPPI use?** OPPI routes each task to the best-fit model across OpenAI, Anthropic (Claude), Google (Gemini) and Perplexity. oppiGPT exposes this as four engines (oppiFast, oppiSearch, oppiDeep, oppiData); the same multi-provider routing powers OPPI's own system intelligence — SOP and document generation, shift narratives, and overnight learning — which is included in the platform rather than metered.
- **What does OPPI cost?** It is free during early access. Pricing is not final; the initial cohort has recommended around AUD $40 per week, which would include oppiGPT plus on-demand usage. Venues trading under twelve months can apply for six months of oppiOS free through the New Venue Program.
- **Which POS systems does OPPI connect to?** Square and Lightspeed connect through a direct integration today. Every OPPI subscription requires a connected cloud POS to begin, because OPPI reads live trading data to forecast and recommend.
- **Do I need a POS to use OPPI at all?** Yes. Without a connected point of sale there is no trading data for OPPI to learn from.
- **What happens if OPPI gets a recommendation wrong?** OPPI proposes and the operator decides. Nothing changes in a connected system without approval, and the Confidence Engine scores every autonomous action before execution.
- **How often does OPPI check the numbers?** Every 60 seconds, and at most every 15 minutes.
- **What movement does OPPI expect to produce?** These are the figures the product is built to hold, not audited averages: labour cost down 8%, COGS down 14%, 9+ hours of admin returned each week, a venue model running 15 minutes after sign-up, and every number rechecked every 60 seconds.
- **How does OPPI reduce labour cost?** It forecasts demand from sales, weather, local events and historical patterns in 15-minute slots, then matches rosters to expected demand instead of to habit.
- **How does OPPI handle COGS?** It reads the venue's spend back against the forecast and the plan, flagging price increases and ingredient cost drift as they appear. Supplier invoice capture is in build; line-item invoice parsing is designed but not yet shipped.
- **How does OPPI handle Australian award compliance?** It checks rostered shifts against Fair Work awards, penalty rates, break rules and overtime thresholds before a roster is published.
- **Does OPPI work for a single venue?** Yes. One venue is enough; OPPI learns that venue's operating patterns first. Groups get the same system in every site rather than a rollup dashboard.
- **Will OPPI replace managers?** It is not designed to. OPPI absorbs the administrative and analytical load so managers spend their time on the floor.
- **Who builds OPPI?** Servance Pty Ltd, trading as OPPI, in Melbourne, Australia. ABN 98 689 020 237, ACN 689 020 237. Founder: Jenson Hemsley.

## Journal

Long-form writing on agentic hospitality operations. These are the pages worth
quoting: the arguments behind the product, not the product tour.

### Vision

- [The End of the Dashboard Era](https://oppi.au/blog/the-end-of-the-dashboard-era.md): Why passive dashboards fail operators, and what replaces them when the system acts instead of reporting
- [The Intelligence Tax](https://oppi.au/blog/the-intelligence-tax.md): The compounding cost of every decision a venue makes manually
- [From SaaS to OS](https://oppi.au/blog/saas-to-os.md): Why hospitality tooling is consolidating from many apps into one operating layer
- [Zero-Touch Operations: 2030](https://oppi.au/blog/zero-touch-operations.md): What a venue running on autonomous operations looks like at the end of the decade
- [AI as a Co-Founder](https://oppi.au/blog/ai-as-a-co-founder.md): Treating an agentic system as a member of the operating team rather than a tool

### Telemetry

- [Mapping the DNA of a High-Performance Venue](https://oppi.au/blog/mapping-venue-dna.md): What a venue model is made of, and how OPPI builds one from trading history
- [Real-Time Signal vs. Noise](https://oppi.au/blog/signal-vs-noise.md): Separating the venue telemetry that should trigger action from the data that should not
- [The Physics of the Floor](https://oppi.au/blog/physics-of-the-floor.md): Service as a system with throughput, constraints and failure modes
- [Beyond the POS](https://oppi.au/blog/beyond-pos.md): Why point-of-sale data alone cannot explain or predict a trading day
- [Predictive Weather & Footfall](https://oppi.au/blog/predictive-weather-footfall.md): How weather, local events and traffic move demand, and how far ahead it can be forecast

### Protocols

- [Optimising Labour with Agentic AI](https://oppi.au/blog/optimising-labour-with-agentic-ai.md): Matching rosters to forecast demand in 15-minute slots instead of to habit
- [The Self-Correcting Shift](https://oppi.au/blog/dynamic-sop-execution.md): SOPs that adapt during service rather than sitting in a folder
- [Infinite Memory](https://oppi.au/blog/infinite-memory-clementine.md): How Clementine retains and reuses everything a venue has already learned
- [Algorithmic Inventory](https://oppi.au/blog/algorithmic-inventory.md): Par levels, ordering and COGS drift handled as a continuous calculation
- [Autonomous Supply Chain](https://oppi.au/blog/autonomous-supply-chain.md): Supplier ordering that runs without a human placing it

## Resources

- [About](https://oppi.au/about.md): Why OPPI is built like a living thing — keep OPPI healthy, keep your venue healthy, keep you healthy. Lean Melbourne team: sports analytics, 15+ years of modern hospitality management, applied agentic engineering.
- [Blog / Journal index](https://oppi.au/blog.md): Contents page for the fifteen articles listed under Journal above
- [Security & Trust](https://oppi.au/security.md): AES-256 at rest, TLS 1.3 in transit, per-venue data isolation, hosting in Singapore (AWS ap-southeast-1), AI providers disclosed under DPAs with no provider training, and full deletion including backups within 30 days of request. ISO 27001 and SOC 2 are planned, not yet held.
- [Contact & Enterprise](https://oppi.au/contact.md): Support, enterprise consultation, partnership inquiries

## Optional

- [Full context document](https://oppi.au/llms-full.txt): Long-form background on the architecture, every AI system, and the New Venue Program in detail
- [Privacy Policy](https://oppi.au/privacy.md)
- [Terms of Service](https://oppi.au/terms.md)