Research Agent
agentSearches the live web for what is happening in your market right now: competitors, positioning, the current conversation.
- Reads
- Brand Guide · Industry Intelligence · live web
- Produces
- Research summary, per channel
How it works
Every screen below is a working picture of the product. Scroll, hover, click. Nothing here is a description of a feature we could not show.
Chapter 1
The Brand Guide is filled in once, from scratch or from the files you already have. Every field ends up somewhere in what gets written, and your team reads the same page.
Dhanteras falls on 29 October. The Meera set is finished: polki drops on a fine chain in warm gold, hallmarked 18k, made in our Jaipur workshop. Reserve a fitting this week and collect before the day. Shot on ivory linen, window light. An unhurried piece for every day after.
Hover or tap a field on the left.
Upload product photos and Aura writes each one's Visual DNA: shape, material, finish, colour, how it is presented. From then on it knows what your product looks like, and stops inventing features it does not have. Bulk import brings a whole catalogue in at once; backgrounds can be stripped in place.
Chapter 2
Five questions, then nine AI workers and two approvals. Here is what each one hands to the next.
What this campaign is for. A launch, a festival, a season, a point of view.
Which products or collections it features. Their Visual DNA comes along automatically.
Where it runs. Copy is written natively for each one rather than reformatted.
The visual direction it inherits: your lighting, your palette, your composition rules.
When each post goes live. Then research starts.
Nine workers and two approvals, in this order. Scroll and they stack. Each card shows what that step hands to the next.
Searches the live web for what is happening in your market right now: competitors, positioning, the current conversation.
Turns raw research into an angle: what to say, to whom, on which channel, and why this beats the obvious version.
You read the thinking before a single word of copy is written. Approve it, or reject it with a reason the agents actually receive.
Writes each channel's copy on its own terms, instead of one caption squeezed five ways. The rules in your Brand Guide are rules it cannot break.
Reads every caption for trademarks, copied wording and claims your industry does not allow. Anything doubtful is flagged for the person approving, so you see the risk before you say yes.
Looks at your product photos and writes down what each piece really is: shape, material, finish, colour. A product is read once and remembered, not described again for every campaign.
Art directs the shot. Composition, camera angle, lighting and framing are decided before anything is rendered.
Checks the image brief against your brand rules and banned words before any image is made, when a fix costs nothing.
Makes the images, all in one shared style, so post three still looks like post one.
Checks the finished work against your Brand Guide and sends anything off brand back round before you ever see it.
The finished campaign lands in your approval queue. Nothing reaches a channel until a person says yes.
Chapter 3
Strategy first, so you read the thinking before copy exists. Creative second, so nothing is scheduled you have not seen.
Gate A, before copy exists
Dhanteras · Meera set
Strategy waiting since 09:12 · Gate A · approver Priya
Gate B, before anything is scheduled
Dhanteras · Meera set
Creative waiting since 14:40 · Gate B · approver Priya
Try rejecting one. The reason you type is recorded against the campaign and handed to the agents when they rerun. If the same reason keeps coming up across a category, it becomes a guideline for everyone in it. Your own corrections stay labelled as yours.
Chapter 4
Approved posts publish themselves at the scheduled minute. Content is generated three hours ahead so it is ready well before it is due.
Three hours before a post is due, its content is already finished and waiting, so a last minute review is never a rushed one.
If an approval is coming due, the approver gets an email with the deadline. Nothing sneaks past a gate.
Each publish to each channel is logged as published or failed, with the platform's own reason when it fails, so a busy week never looks like an outage.
Chapter 5
The planning agent proposes the quarter. Brand memory carries your past decisions into the next brief. Nothing has to be re-explained.
Each row has its own conversation. The agent proposes; nothing is published from here.
One run produces a sequenced slate: which products belong together and why, what order they should run in, and what it deliberately left out. Every row still opens into the same brief and the same two approval gates. The agent proposes; it never publishes.
Not every row has to sell a product. Brand story, point of view and community angles carry no catalogue item at all. Open one and talk it through with Aura until it is a strategy for a specific moment.
One run proposes the quarter, with its reasons and its exclusions. Talk any row through before it becomes a brief.
Past approvals and rejections are ranked by how similar they are to the campaign being written, and handed to the agents.
A writing style built from your own approve and reject history. It yields to your brand rules and to your instructions.
Redo one image or one caption without rerunning the campaign.
A chat with your brand context loaded. Ask for a variation, a rewrite, a one off image, or schedule a post from the conversation.
Everything Aura has made for you, searchable and reusable.
Send us your brand and your industry. We will come back with a campaign Aura made for you (research, strategy, copy and visuals), so you can judge the work instead of the pitch.