For chains running 5–200 locations
Most analytics tools tell you what already happened. Insights proposes the move with the EBIT case attached, shows the math, and routes any employee-affecting action through human approval before it ships.
For chains past the point where built-in POS reports stop being enough — and before the stage where you'd build your own data warehouse. Insights sits on top of the POS and stack you already run, and turns it into proposed moves the C-suite can approve in one place.
Try the loop on a question
The assistant here is grounded in the Insights handbook — EBIT drivers, the role agents, GDPR Article 22 routing — and can call live tools to qualify your fit. It will refuse to invent prices or features. Walkthrough booking still happens through the form on /contact when you're ready.
Grounded in the handbook · live tools: get_pricing_scope, list_pos_connectors. Insights itself is in Phase 0 — internal validation against Čili-shaped data is the current focus.
The decision loop
A rearview-mirror BI tool answers the question once. Insights closes the loop — explain, predict, recommend, approve, measure — so each recommendation can be checked against what actually happened, and each next decision is informed by the last.
What moved the number, and why.
What happens if you change X — with confidence intervals.
The move with the best EBIT outcome — never executed without approval.
Named reviewer signs off. Required for any employee action — GDPR Article 22.
Was the prediction right? Compare the forecast to the actual result.
Rearview-mirror tools
Show the report. You read, decide, act, hope you measured the right thing. Next quarter, repeat. You never find out whether the call was right.
Self-driving navigation
Closes the loop in days, not quarters. Every recommendation is tied to the forecast it was based on, so you can check whether the call was right — not just whether the chart moved.
What's different
Reporting tools draw the chart. Insights does the math, runs the simulation, drafts the move, and queues it for approval.
Capability 01
Role-specialised AI agents (CFO, CMO, COO, CPO) propose cost reductions and operational shifts in the language of the role. Every action a person feels — schedule, price, hours — is held in an approval queue. GDPR Article 22 requires a named reviewer for any employee-affecting decision, and we ship it that way by default.
Capability 02
Most "campaign performance" reports compare a before-period average to an after-period average and call the difference ROI. That comparison ignores natural variance and produces false positives. Insights runs Welch's t-test against control and reports Cohen's d effect size — peer-reviewed math that tells you whether the lift is real and how big it actually is.
Example output
Capability 03
Stop forecasting from last year. Type in the EBIT you need this year, and Insights works the P&L backward — the turnover you need, month by month with seasonality, to land it. You see the gap before the year starts, not in the December review.
Example calculation
Capability 04
Ask Insights questions the way you'd ask a senior analyst — "why is location 12 underperforming?", "what happens if we raise lunch combo 5% at the mall sites?". Get back a real answer, the data behind it, and — for forecastable questions — a projection with a confidence interval. No SQL, no dashboard navigation, no waiting until Monday.
Example query
Four role-specialised agents
Insights runs role-specialised AI agents — CFO, CMO, COO, CPO — on a shared model. Each answers in the vocabulary of its role, with permissions and approval routing scoped to that role.
EBIT · margin · forecast
Finance vocabulary. Models price and mix scenarios; reconciles forecast vs actual at month/quarter close.
campaigns · ROAS · uplift
Runs campaign analysis with control groups. Calls a campaign uplift real or noise — and stands behind the math.
service · throughput · waste
Compares service time, prep time, error rate, waste against the network median. Surfaces where ops drift opens up.
menu · mix · pricing
Menu engineering with margin and demand elasticity. Recommends mix shifts when input cost moves.
Employee-affecting recommendations route through a separate human-approval gate — see below.
Human in the loop
Insights does not autonomously change a guest's price, an employee's schedule, or anything that touches a person. Every recommendation lands in an approval queue with the data behind it, the predicted impact, and the alternatives.
For employee-affecting actions — schedules, hours, performance flags, retention measures — human approval is not a setting. It's required by GDPR Article 22 and we ship it that way by default. Operators have an AI trust deficit; "fully autonomous" is the wrong promise to make. We don't make it.
Frictionless integration
Insights consumes a narrow slice of the data a chain typically warehouses — the events, prices, costs, shifts and forecasts that move EBIT. The rest can keep living wherever it lives. Two ways in: file uploads or a direct data connector.
Path A
Drop the exports your finance and ops teams already pull weekly. Insights' pluggable importer maps your columns to the canonical model — no schema work, no ETL pipeline.
Path B
When the loop matters in near-real time, connect Insights directly to your POS data. R-Keeper is live in production today through a managed data integration; other POS connectors are scoped per-customer.
No data warehouse required.
Lean ingestion, EU-resident, tenant-isolated. Your source data stays where it lives.
How to get it
Insights is not sold as a standalone analytics product. It is the decision layer of the Fooodo platform — the same operating system that runs your QR menu, ordering and payments. Take Fooodo, and the data flows in automatically. The role agents start scoring every shift against EBIT, gross margin and labor — using the same numbers your locations already report.
No percentage-of-revenue fee. No success tax. No bolt-on contract.
The decision-loop architecture is live in internal validation against Čili-shaped data. We are onboarding a small number of design-partner chains through 2026. If you run 5–200 locations and want to shape the rollout, get in touch.
What you get on day one
Enterprise track available for 150+ locations · Custom POS · SSO · DPA
From any AI client
Every tenant ships a remote Model Context Protocol server. Connect once with OAuth — your CFO and analysts ask the same questions from whichever AI they already use, against your own data, with tenant isolation and a full audit trail on every action.
MCP · 2025-06-18
OAuth 2.1 + Dynamic Client Registration · dark by default per org · enable via support
Frequently asked
BI tools end at the dashboard. Insights closes the loop: every recommendation is tied to the forecast it was based on, so you can check whether the call was right. It also runs the marketing-ROI math (Welch / Cohen) and the reverse-P&L workflow, and routes any employee-affecting action through the human approval that GDPR Article 22 requires. Those are not features BI tools cover.
No. Insights uses a pluggable importer: drop CSV/Excel exports or wire a direct connector to your POS data. Insights only consumes the slice of data that moves EBIT — events, prices, costs, shifts, forecasts. No Snowflake required on your side.
R-Keeper is live in production. Any other POS works via the CSV/Excel importer; real-time connectors for other systems are scoped per-customer and quoted on demand. Custom connectors are part of the Enterprise tier.
Two phases: ingestion (CSV importer or direct connector), then agent enablement once enough data has flowed to fit forecasts. Exact timeline is gated by POS connector availability and the cleanliness of historical data, and is scoped on the demo.
Two reasons. Legal: GDPR Article 22 requires human review of any decision that significantly affects an employee — schedules, hours, performance, retention. Auto-applying those is non-compliant in the EU. Operational: restaurant operators have an AI trust deficit for good reasons. A "fully autonomous" promise is the wrong one to make. AI proposes; people approve; the loop stays auditable.
EU region. Tenant-isolated. Your data trains forecast models for your tenant only — never pooled to train models other tenants use. Architecture and security details: /docs.
No. Insights is the decision layer of the Fooodo platform — bundled with the QR menu and ordering system at no separate cost. There is no percentage-of-revenue fee, no success tax, no bolt-on contract. If you run Fooodo, you get Insights.
See Insights against your numbers.
30-minute walkthrough with your CFO or COO. We'll model one decision in your data and run it through the full loop.