An AI Dashboard for Your Online Store
An AI dashboard for your online store should tell you what your AI team did and what needs you. Here is what Pulse and Plans cover, and what they do not.
An AI dashboard for your online store is the single screen that shows what your AI team did across every department and what now needs your decision. When agents run the daily work, the founder's job shifts from doing tasks to reviewing them, so a good dashboard reads like a narrative feed of actions, not a wall of charts.
This is a SoGood post, so read it with that bias in mind. SoGood is an AI co-founder that builds and runs an ecommerce store, and the dashboard described here is ours. We will be honest about where it is thin: this is an oversight and visibility layer, not a business-intelligence suite, and the analytics depth is closer to a lightweight web-analytics tool than to a full reporting stack.
The problem changes when an AI team runs the store
When you do the work yourself, the dashboard you want is a to-do list and a few numbers. You already know what happened today because you did it. The hard part is execution, so the tooling is built around getting things done, the way most guides to starting an ecommerce business with AI frame it.
When an AI co-founder for your ecommerce business runs the daily work, the hard part flips. The agent sourced suppliers, drafted ads, posted to social, and answered an email overnight, and you were asleep. Now the scarce resource is knowing what happened and catching the two things that actually need you.
This is the visibility problem, and it is the real reason an AI-run store needs a different dashboard. A traditional store dashboard answers how is the business doing. An AI-run store dashboard has to answer that plus what did my team do and what is waiting on me. Those last two questions are where most tools go quiet.
Pulse: one narrative feed across eight departments
Pulse is the answer to what did my team do. It is a real-time activity feed that pulls signals from every department, including completed tasks, sent and received emails, ad and traffic metrics, social posts, and knowledge-base updates, then groups them into readable stories.
The synthesis is the part that matters. Raw event logs are noise; forty individual signals across AI agents for ecommerce departments would bury the one thing you care about. Pulse uses an LLM to group related signals into a single narrative story, so launched a new ad set and traffic rose 18 percent reads as one item, not six.
The eight departments behind the feed are CEO, Brand, Tech, Marketing, Sales, Operations, Strategy, and Finance. Pulse does not silo them into eight tabs you have to check; it merges their output into one chronological story stream. You read it the way you would read a sharp status update from a chief of staff.
This is what makes the dashboard feel less like analytics and more like oversight. You are not interpreting charts to guess what your team did. The feed states it in plain language, with the underlying signals one click away if you want the detail.
Plans: the task board behind the feed
If Pulse is the story of what happened, Plans is the board of what is being worked on and what is stuck. It is a kanban with columns for Backlog, Todo, In Progress, Waiting, Done, and Cancelled, filtered by department and priority, with tasks assigned to specific agents.
The Waiting column is the one founders watch. That is where a task sits when it needs a human decision: a supplier replied with a quote, an ad is drafted and ready to publish, an outreach batch is queued. The board makes the handoff explicit, so nothing silently waits on you without showing up somewhere visible, which matters even on a near-automated model like automating dropshipping with AI.
Together, Pulse and Plans split the founder's attention cleanly. Pulse is read-mostly, the narrative of done work. Plans is act-mostly, the queue of decisions. A non-technical founder can run a store by scanning the feed and clearing the Waiting column, which is roughly the workflow we describe for non-technical founders launching without developers.
What the store dashboard actually shows
Beyond Pulse and Plans, the dashboard has the concrete store tabs you would expect, and it is worth being precise about their depth so you are not surprised.
The Commerce tab shows orders, products, and refunds, with an order detail and timeline. It is read-only in the dashboard; the selling and refunding happen in the agent and worker layer, and you watch the results here. The Website tab shows self-hosted Umami analytics: visitors, page views, average session length, top pages, and a geography and device breakdown for the last 30 days.
That analytics depth is the honest limit. It answers the early-stage questions well, but it is not a reporting suite. There is no multi-touch attribution, no cohort analysis, no custom funnel builder, and no enterprise BI. If you need to know which of five touchpoints drove a sale, this dashboard will not tell you, and you should keep a dedicated analytics tool for that.
| Dashboard surface | What it shows | What it does not do |
|---|---|---|
| Pulse feed | Narrative stories of work across all 8 departments | Forecasting, predictive alerts |
| Plans board | Task kanban, agent assignment, Waiting queue | Gantt charts, resource planning |
| Commerce tab | Orders, products, refunds, order timeline | Edit or refund from the dashboard |
| Website tab | Umami visitors, top pages, geo, device, 30-day | Attribution, cohorts, custom funnels |
| Multi-business grid | Every store, live preview, status, 7-day traffic | Cross-store BI rollups |
The table is the fastest way to set expectations. The dashboard is wide, covering many surfaces of a running store, but it is deliberately shallow on the analytics axis. That is the trade you make for having one screen instead of a stack of logins.
Approval lives in the dashboard, and that is the point
A dashboard that runs your store has to be the place where you say yes or no. The risky actions are gated behind human approval, and the dashboard is where that gate sits.
Paid ads are drafted by the agent, including static creatives and AI-video drafts, but they publish only after a human approves them in the dashboard. Supplier sourcing runs a full playbook, find candidates, vet, send an RFQ, negotiate, but it never commits money; the founder makes the call. You can see exactly how the sourcing side works in our piece on whether AI can source suppliers and handle fulfillment for an online store.
This approval flow is why the dashboard is an oversight layer first. It is not just reporting after the fact; it is the control point where founder judgment enters the loop. The agent proposes, the dashboard surfaces the proposal, and you decide.
Running more than one store from one screen
Many SoGood founders run more than one store, and the multi-business view is built for that. It is a grid of every project you run, each card showing a live preview of the storefront, a status pill for live, setting up, or offline, and seven days of traffic.
From the grid you open one store to get its full Pulse feed and Plans board, then jump back. You are not opening a separate analytics login per store or stitching exports together. For an operator overseeing several stores, the consolidation is the value, even though each store's analytics stays at the same lightweight depth.
This is the bundle argument in miniature. SoGood will not beat a dedicated analytics tool on depth for any single store, but it gives you one screen across all of them, which is the same logic founders use when they cannot afford a marketing agency and build an AI stack instead.
CEO chat: the dashboard you can talk to
The dashboard is not only something you read. Each store has a persistent CEO chat, a streaming conversation with the agent where you can ask what happened, request work, or upload an image for context.
In practice, CEO chat and Pulse work as a pair. Pulse tells you what happened without being asked; CEO chat lets you drill in, ask why, or redirect. If a Pulse story says traffic dipped, you ask the agent in chat to investigate and propose a fix, and the resulting task shows up on the Plans board.
This is the part that makes the dashboard feel like working with a co-founder rather than reading a report. The feed, the board, and the chat are three views of the same running business, which is the broader idea behind what an AI co-founder is.
Where this dashboard wins and where it does not
Be clear-eyed about fit. The AI dashboard for your online store wins when your problem is oversight: you have an AI team doing the work and you need to see it, approve the risky parts, and run by exception. It is strong on breadth, narrative clarity, and the approval loop.
It loses when your problem is depth. If you need multi-touch attribution, cohort retention curves, or board-grade BI, the built-in analytics will frustrate you, and you should pair a dedicated tool. SoGood is a bundle that runs the company, not a best-in-class analytics product, and we would rather you know that before you sign up than after. SoGood is priced in tiers: Basic is free, Pro is $29 a month, and Expert is $99 a month, and you can add credit packs on any plan.
The right way to choose is to weigh it against the field, which we do in our roundup of the best AI co-founder platforms of 2026. If the core need is one screen that shows what your AI team did and what needs you, across one store or several, this dashboard fits. If the core need is reporting depth, buy the depth.