Everything we shipped in July — and the one idea behind it
Six things went live at Ocula this month. Before the list, the thing that connects them.
Whoever has the richest, most authoritative product data wins in agentic commerce. Not the best-designed PDP, not the biggest ad budget — the deepest, most trustworthy data. That's what LLMs read when they answer a shopping question, what Google reads when it decides whether your product deserves a place, and what a buying agent reads when it compares your listing against three others.
Everything below serves that. Some of it makes your data richer. Some makes it more authoritative. Some makes sure a person is still in charge while agents do the volume.
Document Links: the field almost nobody fills
There's an attribute in the Google Shopping feed called document_link. It tells Google where the manufacturer's own PDF lives — the spec sheet, the manual, the datasheet — and pulls it straight into the feed.
It is one of the strongest authority signals available to a retailer. It's a document the manufacturer produced, not marketing copy you wrote about yourself, and both Google and LLMs weight that difference heavily. When an assistant is asked whether a TV supports 4K/120Hz on HDMI 2.1, a manufacturer spec sheet is the kind of source it trusts.
Almost nobody populates it. Not because teams don't know it exists — because finding the correct PDF for every single SKU, by hand, across a catalogue of tens of thousands, doesn't scale. So the field stays blank. For you, and for every competitor you're listed beside.
The enrichment agent now captures manufacturer document links while it researches each product, as first-class data points rather than an afterthought. The link is attached to the product, visible under Document Links on the PDP, and flows into your feed with everything else. A blank optional field becomes a working link across the entire catalogue, with no manual sourcing.
Taxonomy Boost: performance past the PIM ceiling
Taxonomy has been the most painful conversation we've had with customers this year, and the reason is structural. Your taxonomy lives in your PIM, and PIMs are rigid by design — that rigidity is the point, it's what keeps your data consistent. But it constrains what you can express about a product, and the detail and nuance that gets lost is exactly the detail that makes a product findable.
The instinct is to fix it in the PIM. That means a project, a data team, months, and a migration nobody wants to own.
So we didn't touch the PIM. Your PIM stays the source of truth. Instead the agent looks at your actual enrichment opportunities and expands the taxonomy only where content is being held back — adding new attributes and new values where they'd change performance, and leaving the rest alone.
The analogy Danny uses in the video: it pushes performance past what the speedometer says, without touching the car's specs. Same car, new limit.
The part that matters to whoever owns your data quality bar: everything the agent adds is badged and flagged. Data quality teams review what's flagged and edit anything on the spot, or by telling the agent in the chat. Every expansion is attributed, timestamped and reversible in one click. Delete a category with children and you have to type its name to confirm. The agent proposes; your team disposes.
Agentic commerce attributes, now on by default
This one is live on every dashboard today. No waitlist, no setup, no request to your account manager.
It takes everything the market says about your products — editorials, reviews, Reddit threads, YouTube reviews — and turns it into structured data: highlights, Q&As, use cases and differentiators, ready for your PDPs and your feeds.
Why it earns its place: during the World Cup, What Hi-Fi published "the best TVs for the football tournament". Ocula catches that, and the editorial becomes a differentiator on the PDP — brightest panel for daytime kickoffs, most HDMI 2.1 ports for multi-device match nights. Products carrying that moment-relevant content got more visibility from LLMs. Not because the TV changed. Because the content was fresh when the world was asking.
That's the mechanism worth internalising. Relevancy is a moving target, and the products that win a moment are the ones whose content was updated for it. Plug this into your trading calendar — your most-talked-about products, your key retail moments — and let relevancy do the work.
Every tab is generated from real market signals, and every claim is editable or removable, so you stay in control of what ships to the PDP. Pick any SKU, run it, done.
Export Builder: one collection, every destination
You've enriched the data and generated the content. Now it has to land somewhere — and your PIM, your feed manager and your e-commerce platform each want their own format.
The Export Builder handles the shape-shifting. Start from a template or from scratch. Pick your fields, rename, reorder, resize. Mix static and multi-source columns. Wide layout for feeds, one row per product. Long layout for your PIM, attribute-value pairs with blanks dropped so it loads straight in. Live preview shows exactly what you'll get before you commit.
Then save the format once, at company level. Next time it's one click.
The point isn't the file formats — it's who can produce them. One collection exports into as many formats as you have destinations, with no ticket to anyone technical. The person who owns the content owns the export.
Agent Plans: scale the work, keep control
The question behind nearly every conversation we have right now: everything today is agents doing more work on their own, so how does a team let that happen without losing the wheel?
Agent Plans is our answer. Before anything runs, the agent scopes the job and shows you which products are in, which rules apply, in what mode, and exactly what will be produced. Nothing generates until someone approves it.
Inside a Plan, enrichment and copywriting can belong to different owners, so handovers are clean and it's never ambiguous who's holding a stage. You run a sample and inspect it before committing spend — which matters, because the expensive mistake in agentic workflows isn't a bad output, it's a bad output produced ten thousand times. If something's off, fix the rule and rerun only the products that need it. Enrichment gets signed off before copywriting starts, so every stage builds on approved work.
The result is that the team agrees what the agent will do before it does it, catches issues before they cost anything, and big jobs run as one connected flow instead of a series of disconnected runs nobody can audit.
More scale and more control. Not one or the other.
User profiles and the audit trail
Optimising product content touches different people. Data experts on taxonomy and attributes. Copywriters and merchandisers on brand voice. Seniors who approve, juniors who operate. For a single product, several hats are needed — and you need to know who's wearing which.
The new user profiles section covers admins and members, with every session timestamped. The audit trail answers the question every team eventually asks: who changed that rule, who generated the final copy, who exported what to which channel — and when. Open any rule and you can see its version history, what changed between versions, and who made the change.
On handovers — say a content review — you notify the right person directly from the platform, and ownership of a stage moves with it. No chasing over email.
This section is new and it'll keep growing with the enterprise requirements customers bring us. If there's something your security or compliance review needs to see, tell your account manager — that's how this list gets written.
The goal doesn't change: the right people working with the right agents, clean handovers, and always a trail of what happened and when.
Richer data, more authoritative data, and a team that stays in control of both. That's the whole thesis, and it's why these six shipped together rather than separately.
🎥 Watch the full Ocula Product News episode here
-
Ocula Product News — Episode 01 · full transcript
0:00 — Welcome and the product flywheel
Welcome to our first product update — a lot has shipped, and it all serves one idea. And that idea is: to win in agentic commerce, you need the richest, most authoritative product data. Everything we're about to show you exists to serve that single purpose. This is the product layer flywheel, and it never stops moving. You enrich the data, generate content, activate it into every channel — and because the market keeps talking, you refresh and go again. Always on.
0:29 — How we ship updates
And this is how you're going to get updates from us — every release sorted into these buckets. The four you just saw, plus the fifth: governance. Permissions, access control, SOC 2, security, privacy — everything enterprise teams need to run agents with confidence.
0:46 — Document Links
Starting with a secret. Danny — tell them about the field nobody fills— In your Google feed there's an attribute called Document Link. It's how Google ingests the manufacturer's own PDF — the spec sheet, the manual — straight into their feed. It's one of the strongest authority signals available… and almost nobody populates it. Why? Because sourcing the right PDF for every single SKU by hand just doesn't scale. So that's why it stays blank — for you, and for everyone else. Now as you can see, the enrichment agent researches each product. It captures the manufacturer's document links as first-class data points. Here's the document for this particular product— Hold on. Why is your demo dark? Oh — dark mode's live. Just go to Settings, click on Appearance, and you can pick your side. So if you click here — that's the actual manufacturer's spec sheet. All of this happens fully automatically. A blank optional field just became a working link across your entire catalogue — zero manual work. That's AI discoverability the rest of the market is leaving empty.
1:52 — Taxonomy Boost
Data quality teams and PMs will love us for the next one. Danny, why has taxonomy been the most painful conversation with every customer this year? Because the taxonomy lives in your PIM — and PIMs are rigid by design. It constrains the data and you lose a lot of the detail and the nuance for the product to really perform. And now? So here's where we give your taxonomy a major boost. It pushes performance past what the speedo says — without even touching the car's specs. Your PIM stays the source of truth. We just give you the turbo boost you need to win the drag race with your current car. The agent looks at your actual enrichment opportunities and expands the taxonomy only where the content is being held back. New attributes, new values — everything it adds is badged and flagged. So the CSV ping-pong just… becomes this screen? This screen — Data Quality teams can come in, review what's flagged, edit anything right there, or just by telling the agent in the chat. Okay, but the people who own that taxonomy — data teams, PMs — they've spent years building it. Why would they let an agent near it? Because they keep full control while boosting discoverability. Every expansion is attributed, timestamped, reversible in one click. Whoever owns the data-quality bar sees the entire audit trail. And deleting a category with children? You type the name to confirm. The agent proposes — your team disposes. Bottom line: your PIM stays the source of truth, your team stays the owner — but your content finally performs past the old ceiling.
3:26 — Agentic commerce attributes
This one's mine, because it's my favorite. Agentic commerce attributes are now on by default. For everyone. Live on your dashboard today. Yes — no waitlist, no setup. It's just there. Quick recap: this takes everything the market says about your products — editorials, reviews, Reddit, YouTube — and turns it into structured data. Highlights, Q&As, use cases, differentiators. Straight into your PDPs and your feeds. Here it is on a real SKU. Every tab generated from those market signals — and every claim editable and removable. You stay in control of what ships to the PDP. Pick any SKU, run it, and it's done. Here's why I love it. During the World Cup, What Hi-Fi publishes "best TVs for the football tournament." Ocula catches it — and that editorial becomes a differentiator on the PDP: brightest panel for daytime kickoffs. Products carrying that moment-relevant content got more visibility from LLMs. Not because the TV changed — because the content was fresh when the world was asking. So plug this into your trading calendar — your most-talked-about products, your key moments — and let relevancy do the work. It's on your dashboard right now — so just go and press the button.
4:36 — Export Builder
You've enriched the data, you've generated the content — now it has to actually land somewhere. PIM, feed manager, e-comm platform — every one wants its own format. So this is the Export Builder. You start from a template or from scratch. Pick your fields — rename, reorder, resize. Mix static and multi-source columns. Wide layout for feeds — one row per product. Long layout for your PIM — attribute-value pairs, blanks dropped, loads straight in. Live preview shows exactly what you get — and you save the format once, at company level. Next time it's just one click. One collection, exported into as many formats as you have destinations. No tickets to anyone technical — anyone can do this.
5:19 — Agent Plans
Governance time. Danny, the question behind every conversation I have: everything today is agents doing more work on their own. How does a team let that happen without losing the wheel? That's exactly what Agent Plans is. Let the agents scale the work and keep people in control. Before anything runs, the agent scopes the job and shows you: which products are in, which rules apply, in what mode, and exactly what will be produced. And here's the key — nothing generates until someone approves it. Inside the Plan: enrichment and copywriting can belong to different owners, so handovers are clean. You run a sample and inspect it before committing spend. If something's off — fix the rule, rerun only the products that need it. And enrichment gets signed off before copywriting starts, so every stage builds on approved work. So the team agrees what the agent will do before it does it, catches issues before they cost anything, and big jobs run as one connected flow. More scale, more control. Not one or the other.
6:18 — User profiles and audit trail
Optimising product content touches different people — data experts on taxonomy and attributes, copywriters and merchandisers on brand voice, seniors who approve, juniors who operate. For a single product, different hats are needed. Correct — and that's why you need to know who's wearing which. This is the new user profiles section. Admins and members, every session timestamped. The audit trail answers the question every team eventually asks: who changed that rule, who generated the final copy, who exported what to which channel — and when. And on handovers — say a content review — you notify the right person directly from the platform. No chasing over emails. And this section is new — it'll keep growing with the enterprise requirements you're bringing us. The goal doesn't change: the right people working with the right agents, clean handovers, and always a trail of what happened and when.
7:10 — Wrap-up
That's everything for this update. It's all live — written update with links below. Reach out to your account manager for questions, and we'll see you in the next one.