Your customers store data in your product. They want to see it as charts, tables, and reports. This happens to every B2B SaaS company between Series A and Series B — the dashboard requests start rolling in and never stop.
You have three paths: build a reporting system in-house, embed a traditional BI tool, or let AI generate dashboards for each customer on demand. This post walks through the third path — the one that takes about a week to ship and doesn't require a dedicated analytics engineering team.
Why your customers keep asking for dashboards #
They're making decisions using the data trapped in your product. Revenue by quarter. Tickets by priority. Inventory by location. Whatever your domain, they need to slice it differently than your one-size-fits-all UI allows.
When you don't give them reporting, they export to spreadsheets and build it themselves outside your product. That's the first step toward churn — they're living in Excel instead of your app.
The longer you wait, the more spreadsheet workflows calcify. And once a customer's decision-making happens outside your product, they start asking why they're paying for it.
Three ways to add customer-facing dashboards #
Build in-house #
Two engineers. Six months minimum. You'll need a charting library, a query engine, per-tenant data filtering, export functionality, a dashboard designer UI, and ongoing maintenance as customers request new chart types.
The build vs buy math rarely works. You'll spend $300k+ in year one and still have a backlog of dashboard requests.
Embed traditional BI (Metabase, Luzmo, etc.) #
Faster to start. You pick a vendor, embed their dashboards via iframe or SDK, and configure per-customer data filtering. But here's the catch: you still build every dashboard.
Customer A wants revenue by region. Customer B wants it by product line. Customer C needs a completely different metric. Suddenly you're maintaining 50 dashboard templates and the backlog looks the same as building in-house — just with a vendor in between.
Pricing compounds too. Per-seat or per-MAU models punish you as customers scale.
AI-powered (Gigacatalyst) #
You embed once. Customers build their own dashboards by describing what they need in plain English. The AI generates interactive reports using their actual data.
No per-dashboard engineering. No template maintenance. Per-tenant pricing that scales with your customer count, not their usage.
This is the approach we'll walk through.
How to add Gigacatalyst in practice #
Step 1: Define your API catalog #
Tell the AI what data it can access. You list your existing API endpoints — the ones that return customer data — and describe what each one does.
Your coding agent can generate this catalog from your existing routes in one pass. No new APIs to build. No data warehouse to configure. The AI reads from the same endpoints your frontend already uses.
Step 2: Embed the builder #
Drop the SDK into your frontend. You configure:
- Allowed origins — which domains can load the embed
- Auth mode — inherited (uses your existing session) or standalone
- Branding — your logo, your colors, or fully white-label with no Gigacatalyst branding visible
It's a script tag and a container div. Your frontend engineer ships it in a morning.
Step 3: Multi-tenancy works automatically #
Each customer gets their own isolated space. The AI only accesses data that customer's session can reach. You don't configure per-tenant filtering manually — it inherits from your existing access control.
Customer A sees Customer A's data. Always. There's no shared state, no accidental cross-tenant leakage.
Step 4: Customers start building #
A customer clicks the builder icon inside your product. They type: "Show me all sales in the last 90 days, grouped by region."
The AI reads their data through your approved APIs, generates an interactive dashboard, and shows it instantly. Charts work. Lists are clickable. They can drill into any data point.
Then they keep going: "Add a section showing my top 5 customers by revenue." Done. New section appears.
Step 5: Save, share, iterate #
After building a dashboard, your customer saves it to their account. They share it with teammates in one click. Teammates can view it or fork their own version.
No tickets. No engineering time. No waiting two sprints for a report.
What this looked like for UpKeep #
UpKeep is a Series B company (Y Combinator W17) that sells maintenance management software. Their customers — facilities managers, maintenance teams — needed different dashboards for different operational views.
They integrated Gigacatalyst in one week. Over 1,000 customers now use it daily. It's the highest-retention part of their app with 500 DAU in the first month.
Their CEO Ryan Chan said it was "absolutely transformational" to their customers and their business. The key: customers build exactly what they need instead of waiting for UpKeep's team to build it for them.
What your customers actually get #
This isn't a chatbot that returns text summaries. Customers get real, interactive applications:
- Interactive charts — bar, line, pie, area, scatter. All clickable, all filterable.
- Data tables with drill-down — click any row to see details. Sort, search, paginate.
- PDF and CSV export — one click to download any view as a formatted report.
- Follow-up refinement — "also highlight overdue items" adds to the existing dashboard.
- Save and share — bookmark any dashboard, share with teammates, access from a personal library.
These are AI-powered dashboards that adapt to each customer's specific needs. Not pre-built templates that force everyone into the same view.
What you DON'T have to build #
This is the part engineering leaders care about most. Here's what leaves your roadmap entirely:
- No chart library — no D3 configs, no Recharts components, no visualization bugs
- No dashboard designer UI — no drag-and-drop editor to maintain
- No per-customer templates — no "Enterprise Customer A wants this layout" tickets
- No query engine — the AI generates queries against your existing APIs
- No export system — PDF and CSV generation is built in
- No sharing infrastructure — save, share, permissions all handled
Your engineers integrate once and move back to your product roadmap. Customer-facing dashboards stop being an engineering problem.
The product adoption effect #
When customers build their own dashboards, they use your product more. Not because you forced a feature on them, but because the analytics match their exact workflow.
We see ~90% adoption rates across our customer base. Compare that to pre-built dashboards (typically 20-40% adoption) where most reports don't match what individuals actually need.
Higher adoption means higher retention. Higher retention means better NRR. The analytics feature becomes a growth engine, not a cost center.
Frequently asked questions #
How long does integration take?
About one week for most teams. Day 1-2: catalog your APIs. Day 3-4: embed the SDK and configure auth. Day 5: QA and ship. Some teams finish in three days.
Does it work with our existing auth?
Yes. Inherited auth mode means the embed uses your existing user session. No separate login, no second identity system. Your customers don't know it's a different product under the hood.
Can we white-label it?
Completely. Your branding, your colors, your domain. Customers interact with something that looks and feels like a native part of your product. You can remove all Gigacatalyst branding.
What data sources does it support?
Snowflake, REST APIs, spreadsheets, and any endpoint that returns structured data. If your product has an API (it does), Gigacatalyst can read it.
What if a customer asks for something the AI can't build?
The AI only accesses your approved API catalog. If the data doesn't exist in an approved endpoint, it says so clearly. No hallucinated numbers, no fabricated data. It's constrained by design.
How is pricing structured?
Per-tenant. You pay based on how many of your customers use the feature. End-users within each customer account are unlimited. No per-seat charges, no session fees, no MAU surprises.
Ship customer-facing dashboards this week #
The gap between "customers want reporting" and "we shipped reporting" doesn't have to be six months anymore. One integration, one week, and your customers build their own analytics inside your product.
Book a demo and see how a customer builds their first dashboard in under two minutes.