Enterprise pilot
Supper connects to your entire data stack, builds a semantic model around how your business actually measures itself, and gives everyone accurate answers — in plain language, in seconds, without asking the data team.
This document covers what makes Supper different, how the 90-day pilot is structured with your dedicated Forward Deployed Analyst, and what your team should expect at each stage.
The accuracy layer
Every question is routed through a semantic model built specifically for your company — encoding your definitions, your business logic, and your data relationships before a single number reaches a user.
Every answer runs through your metric definitions and business logic — not generic SQL. Your canonical definitions, applied consistently to every question from every user.
Anyone on the team asks questions the way they'd ask a colleague. No SQL, no filters, no rebuilds. Answers from live sources in around ten seconds.
No seat limits. Everyone who needs data access gets it. Your data team's time shifts from fielding recurring requests to strategic work.
Any analysis becomes a scheduled workflow — delivered to Slack or email, validated every run. The weekly report that took hours becomes something that just shows up.
Ask anything. Full conversation memory. Follow-ups work naturally.
Always-current dashboards. Click any metric to ask what drove it.
Turn any analysis into a scheduled workflow. Validated every run.
Supper's data layer in Claude, ChatGPT, or any AI tooling via MCP.
A three-phase plan
Your data team is active in Supper by the end of month one. Month two is content production and capability validation. Month three opens to broader user cohorts with FDA support running continuously. Your dedicated Forward Deployed Analyst leads every step — your team's required input is a few hours of review and approval, not a sprint.
Your FDA connects every data source, validates syncs, and generates the first-pass semantic model automatically — overnight, before day two. Days two through four encode your specific business logic with your data team. By day five your first users are asking real questions on live data; by day 30 the model is signed off and your data team is confident in answer accuracy.
Your FDA produces a full content library — executive summaries, repeatable workflows, live dashboards, and presentation-ready outputs — addressing your team's critical questions. This content is reviewed iteratively to surface any remaining logic gaps and identify the formats and features that work best before broader rollout. The first user cohort and testing criteria are finalized by day 60.
The initial cohort is onboarded with 1:1 training by day 65. Broader cohorts follow from day 66. Supper's automated features deliver content proactively and reactively based on user interests. Your FDA reviews user questions on demand, collaborating with your data team as a conduit for anything complex. A usage report — questions answered, time saved, workflows running — is prepared for the day 90 review.
Supper does not store or copy your data. We connect to your existing sources and query them in place — read-only, with permissions enforced at query time across every user and every agent call.
Every Supper customer gets a dedicated FDA — a full-time Supper employee with a senior data background. They lead all technical setup, build your semantic model, and stay on as your ongoing data expert throughout the engagement.