A conversational analyst with read access to all your behavioural data and gated write access to your store — every change previewed, approved, logged and revertible.

DynoAgent is the difference between a dashboard and an answer. A dashboard makes you know where the number lives before you can ask for it. DynoAgent takes the question.

It has read access to everything DynoWeb tracks — clicks, scrolls, frustration signals, journeys, conversions, traffic sources, performance — plus your Shopify catalogue. And it can write: rewrite a product description, create a discount, generate an image, publish a page. Every write goes through an approval gate first.

DynoAgent conversation answering a data question inline.

Asking about your store

Roughly two dozen analysis tools sit behind the chat, and the agent picks and composes them. So questions can be shaped how you actually think, not how a report is filed:

"Which products have the most traffic and the lowest conversion?"
"Where are visitors rage-clicking most this week?"
"Compare this week's conversion rate to last week."
"What's my best-selling product with the worst UX signals?"
"Give me a quick health check of my store."
"Which traffic source actually makes money?"

The last two are composite — a health check spans engagement, frustration, conversion and performance; the traffic question joins channel data to attributed revenue. Composing those is the point of the agent, because they are the questions a dashboard cannot have a page for.

The tool surface also covers cohort comparison, funnel queries, scroll and form analytics, element-level metrics, trend and pattern analysis, period comparison, segment session lookup, storefront error queries, churn-risk and pricing analysis, campaign planning, and an insight memory that lets it recall what you established in an earlier conversation.

Making changes

You ask for something

"Rewrite the description for my four worst-converting products, benefit-first."

It drafts, using your data and your voice

Not generic AI copy. The rewrite is informed by what shoppers actually do on that page — what they click, where they stop, what they seem to be looking for — and written in your store's voice as captured by Brand DNA.

You see the exact change first

The proposed before → after is shown in the conversation. Nothing has touched your store yet.

You approve or deny

Per change. Nothing is written until you say so.

It is logged, and revertible

Every executed action lands in the action history with before-and-after snapshots, and any one of them can be undone in a click, restoring the prior value.

The approval gate is not a setting

There is no autonomous mode to switch on. Writes are queued as pending actions and executed only after you approve them, by design. An agent with unsupervised write access to a live storefront is a bad idea regardless of how good the model is.

What it can write

Area	Capability
Products	Descriptions, SEO title and meta description, variant prices, bulk updates across a collection
Collections	Titles and descriptions
Pages	Create and update — About, FAQ, policy pages, landing copy
Blog	Create and update articles across your blogs
Discounts	Create codes, including ones tied to a SmartNudge
Images	Generate product lifestyle shots and marketing banners, previewed in chat before anything is added
SmartNudge	Create, update, preview, publish and A/B test interventions end to end

Bulk operations are batched — "rewrite every description in the Summer collection" is one instruction, not forty — and are a Custom-plan capability.

Searching outside your store

DynoAgent can search the web in the same conversation as your store data, so competitive and keyword questions do not require a second tab:

"What are competitors charging for something like this?"
"What keywords are trending for this product type?"

The value is the join: external context and your own conversion data in one answer.

DynoAgent approval card with before and after diff.

Usage and cost

Actions are metered, and the meter is tiered by what an action costs to run:

Reads — most analysis queries — are the cheap tier
Writes — anything that mutates your store — cost more
Model-heavy work — long generation, deep reasoning — draws on a separate monthly AI budget rather than the action count

Daily action allowances: 500 on Free, 1,500 on Pro, 5,000 on Custom. Free and Growth carry a $1/month AI budget; Growth and Pro can add more as a paid add-on (+$5 and +$10 respectively), and Custom includes a $40 monthly credit. Free is limited to basic queries; writes need a paid plan.

Conversations export as Markdown or CSV.

Limits

It answers from your tracked data, so it inherits that data's gaps. If a page has little traffic, the agent will say so rather than invent a trend — but a confident-sounding answer about a thin dataset is still thin.
Generated copy is a draft. It is grounded in behaviour and voice, which makes it a good draft. It is not a substitute for reading it before approving it, particularly on claims, sizing, materials and anything regulated.
Web search returns what is published, not what is true. Competitor pricing found this way is a snapshot of a public page.
Revert restores DynoWeb's recorded prior value. If something else changed the same field in between — you, another app, a theme update — reverting sets it back to what DynoWeb captured, which may not be the newest value.
It is not a scheduler. DynoAgent acts within a conversation. Recurring automation lives in the cron-driven features (SEO Autopilot, Impact, performance audits), not here.

DynoAgent