Internal dashboards have a dirty secret: they’re almost never built by the person who needs them.
The sales manager who wants a pipeline health view has to ask an engineer to build it in Retool. The engineer is underwater. The dashboard is delayed by three weeks, then built slightly wrong, then never updated when requirements change. Eventually the manager goes back to copy-pasting data into a spreadsheet.
AI changes this. You can now describe the dashboard you want in plain English and have it running — connected to your real data — in under 10 minutes. No SQL. No Retool. No engineering ticket.
Here’s how.
What “AI-built dashboard” actually means
Before we get into the how, let’s be precise about what this is and isn’t.
An AI-built dashboard is not a pre-made template you fill in. It’s also not a chart builder where you manually connect fields. It’s an AI agent that:
- Reads your data sources — your CRM, your billing tool, your database, your spreadsheet
- Understands what you want from a plain English description
- Builds a live interface showing the metrics and views you described
- Updates automatically on the schedule you set
The output is a shareable link to a live dashboard that reflects current data. Not a PDF. Not a static screenshot. A live view that’s current every time someone opens it.
What kinds of dashboards can you build?
Sales and revenue:
- Pipeline health by stage, rep, and deal age
- MRR, ARR, new revenue, churn by cohort
- Inbound lead volume and conversion by source
- Rep activity: calls, emails, meetings logged
Engineering:
- Open PRs by repo, reviewer, and age
- Deploy frequency and change failure rate
- Sprint velocity and issue burndown
- Incident history and mean time to resolution
Support:
- Ticket volume by category and priority
- SLA compliance and breach rate
- First response time and resolution time by rep
- CSAT trends by category
Marketing:
- Campaign performance by channel (spend, CTR, CPC, ROAS)
- Content pipeline status
- Email performance: open rate, click rate by campaign
- SEO keyword tracking and traffic by page
Operations / Finance:
- Invoice status and overdue amounts
- Expense categories vs. budget by department
- Hiring pipeline stage distribution
- Vendor contract renewal dates
If you can describe what you want to see, the agent can build it.
Step-by-step: Building a pipeline health dashboard in 10 minutes
Let’s do this concretely with a sales pipeline dashboard as the example.
1. Open Shogo Studio (1 minute)
Go to studio.shogo.ai. Free account, no credit card.
2. Start from the right template (1 minute)
For a sales pipeline dashboard, start from the Sales Pipeline template. For revenue metrics, start from Revenue Tracker. For engineering metrics, start from Git Insights.
Templates are pre-configured agents with the core logic already defined. You’re customizing, not building from scratch.
3. Connect your data source (2 minutes)
Click “Connect HubSpot” (or Salesforce, or Google Sheets, or whatever your data lives in) and authenticate with OAuth. Read-only access is fine for most dashboard use cases — you’re pulling data, not writing it.
No database credentials. No API keys. No webhook configuration. OAuth handles authentication with one click.
If your data lives in a spreadsheet (which is fine for many use cases), you can connect Google Sheets the same way — one OAuth click and the agent reads your sheet directly.
4. Describe what you want to see (3 minutes)
This is the key step and it’s simpler than it sounds. Write it the way you’d explain it to an analyst you just hired:
“I want a dashboard showing:
- Total pipeline value by stage (as a funnel)
- Number of open deals per rep, sorted by total value
- A list of deals that haven’t had any CRM activity in the past 7 days
- MRR from new deals closed this month vs last month
Refresh this daily. Data should be current as of this morning.”
That’s your spec. The agent builds the dashboard based on this description plus whatever it finds in your connected HubSpot account.
Tips for good dashboard descriptions:
- List each panel or section you want separately
- Mention groupings (“by rep,” “by stage,” “by week”) explicitly
- Include time ranges (“last 30 days,” “this quarter vs last quarter”)
- Note what order or sorting matters (“sorted by deal size descending”)
5. Review and refine the first run (2 minutes)
Hit “Run Now” to generate the first version of the dashboard. Look at what it built and check:
- Are the right deals/records showing?
- Are the numbers accurate against what you’d expect?
- Is anything missing?
- Are the groupings and labels readable?
If something needs adjustment, describe the change: “Add a filter to exclude deals marked as ‘on hold’” or “Change the deal age calculation to use days since last activity, not days since created.”
Most dashboards are right within 1–2 iterations.
6. Set the refresh schedule and share (1 minute)
Configure how often the dashboard should update:
- Daily — good for operational dashboards managers check each morning
- Weekly — good for review-cycle dashboards (sprint reviews, weekly reports)
- Real-time / hourly — good for live monitoring (support ticket queue, live deal activity)
Then share: copy the dashboard link and post it in your team’s Slack channel, add it to your notion page, or bookmark it in a browser. It’s a real URL that reflects current data every time someone opens it.
Common questions
What if my data is in a spreadsheet, not a CRM?
That’s fine. Connect Google Sheets via OAuth and describe your dashboard based on the column names and data structure in the sheet. The agent reads the sheet and builds the dashboard from whatever’s there.
Can I add my own custom metrics that aren’t pre-built?
Yes. Describe the metric in the agent prompt: “Calculate a deal health score as follows: 100 minus (days since last activity × 5), capped at 0.” The agent computes it.
Can multiple people see the dashboard?
Yes. The shared link is accessible to anyone you give it to. If you’re on a Team plan, you can manage access control.
What if I want the dashboard in Slack instead of a web link?
Configure the agent to post a summary or screenshot to a Slack channel on its schedule instead of (or in addition to) the web dashboard.
Can it update data as well as read it?
For write operations — creating tasks, updating CRM records — you need to enable write permissions on the OAuth connection. This is available but turned off by default for safety.
Why this is better than Retool (for most use cases)
I’ll be direct about the trade-off.
Retool is better if you need: rich interactive components (forms, edit-in-place tables, action buttons), pixel-perfect design control, or a developer who wants to own the code. Full comparison here →
Shogo is better if you need: a dashboard that builds itself, stays current automatically, can be created without a developer, and doesn’t require maintenance when data structures change. Most internal dashboards fit this description.
The Retool workflow is: write the query → build the UI → connect the data → maintain it when things change → hire a developer when you want something new.
The Shogo workflow is: describe what you want → it builds it → it stays current.
Templates to start with
| Dashboard type | Starting template |
|---|---|
| Sales pipeline | Sales Pipeline |
| Revenue / billing | Revenue Tracker |
| GitHub / engineering ops | Git Insights |
| Support tickets | Support Desk |
| Marketing campaigns | Ad Campaign Dashboard |
| CRM health | CRM Dashboard |