Central was born inside an ecommerce agency that got tired of fixing the same problem.

Over a decade of building shops, running ads, fighting feeds, and writing product copy for our own brands and our clients, we kept hitting the same wall: nothing in commerce works without trustworthy product content - and nothing in the stack was actually built to produce it. So we built it.

Ian Schäfer Founder · Central
"The team that invested in product content always won. The hard part was that it was always the hardest thing to do well."
Editorial still life · the world commerce now operates in

It started as an agency problem

For over a decade, we've been building, scaling, and running ecommerce. Our own projects alongside client work, through sixa AG, the digital enablement agency I founded in Switzerland.

We have built brand stores, marketplace operations, B2B shops, and multi-channel rollouts across consumer electronics, premium kitchen, fashion, sports, and industrial parts. We have lived inside WooCommerce, Shopify, Shopware, Magento, Akeneo, Channable, Google Merchant, Amazon Seller, plus generations of updated product feeds.

The pattern was always the same: every project, every brand, every channel. The same fight over product data.

The lesson we kept relearning

In every project, the team that invested in product content won. Not the team with the prettiest design. Not the team with the biggest ad budget. The team that gave the buyer, and the algorithm, something factual to chew on.

The compounding was undeniable:

Google SearchRich, specific product pages outranked thin, templated ones.
Google AdsBetter quality scores cut CPCs without touching the bid.
Meta AdsCreative built from real product facts beat generic catalog ads.
On-site ConversionPages that answered buyer questions converted noticeably better.

Investing in product content was the single highest-leverage thing we did. And it was always the hardest to do well at scale, in quality, and in time.

The ad volume problem

By 2024, with the latest advancements of Claude / GPT and image models like Flux, we were churning out hundreds of ads a month for our brands and clients. Text, image, video, multi-variant, across Meta, Google PMax, and email.

It only worked when there was fact-based content to build on. Real specs. Verified compatibility. Sourced claims. The honest truth: that base was rarely there. So creative teams improvised. And improvised content doesn't compound. It plateaus, gets reported, gets rejected, and burns budget.

We started asking a different question: what if the ad input came from a verified product record, not an incomplete content brief?

The channel sprawl problem

The other side of the same coin: distribution.

Pushing products into price search engines, Google and Bing product feeds, marketplaces. That was a constant drain on resources and an ongoing investment of time. Different attributes per channel. Different rules per category. Different update cadences. Different validators.

So the work was either done shallow, with data optimized for the most important channel and not specifically for the rest, or not done at all. Half the channels that could have driven revenue were sitting there, unconfigured, because nobody had three weeks to map attributes.

Every missing channel was a missed touchpoint. A missed sales channel. A missed conversion. Quietly, every month.

The decision

In late 2025 we made the call. We were going to build the thing we kept wishing existed.

central · founding spec

Hold the product truth. Feed every surface from it.

  • One source of product truth per SKU, gathered from the shop, supplier files, manufacturer specs, certification databases, and source-tagged.
  • Fact-based data flowing into pages, feeds, ads, and AI agents. Not prompts. Not guesses.
  • Channel-specific output generated automatically from the verified record. Google one shape, Amazon another, ChatGPT a third, Perplexity a fourth.
  • AI-first, but with verification as the gatekeeper. AI drafts. The verifier decides what ships.

That's Central.

what we refuse to do

The story produced four commitments.

These are not abstract values. Each one is something we tried the other way first, and watched it not work.

01 · drafting vs. shipping

AI can draft. Verification decides what ships.

The model is a useful drafter. The decision to publish belongs to a verifier that compares sources, scores confidence, and shows operators what they're approving.

02 · verified shipping

Only verified claims ship.

Unsupported claims move into review with their source state visible, so teams can approve content with evidence instead of guesswork.

03 · automation by default

The machine does the work. Humans see only the exceptions.

Central runs the full gather-verify-enrich-publish loop without asking. If a claim is verifiable and sources agree, it ships. Operators are surfaced only when the machine genuinely cannot decide.

04 · readable by all

Product facts should be readable by people, channels, and AI.

Schema.org, JSON-LD, x-central extensions. Every output carries provenance and confidence so any consumer can trust it.

where central is going

Central today is product content. Tomorrow it is agentic-commerce readiness.

Perfect product feeds are enough to justify the company. The longer-term reason we built it is bigger than that.

Commerce is shifting underneath us. Customers are starting to ask ChatGPT, Claude, Perplexity, and other answer machines for product recommendations instead of opening ten browser tabs. AI agents are beginning to search for the best solution for their user, and increasingly, complete the purchase on their behalf.

That shift changes what a brand has to be ready for.

01 · the user journey

"Browse, compare, decide" collapses into "ask, accept."

Most steps the storefront used to own (search, comparison, shortlist, decision) now happen inside someone else's interface, before the buyer ever reaches you.

02 · the storefront

UI/UX matters less. The underlying record matters more.

Whatever the agent decides to surface comes from your data, not your design. The shop is still useful, but it stops being the place the decision is made.

03 · the consumer of your data

Marketplaces, search, ads, and one new consumer: the AI agent.

Every existing channel still wants product content. The agent is just the newest, fastest-growing, and least forgiving one. And it reads the same record.

04 · what wins

Catalogs that are complete, sourced, structured, and trusted.

The brands that win the next decade will have product data an AI agent can recommend confidently. Everything else (pricing, brand, design) only gets a vote if the data clears that bar first.

central's bet

We don't think agentic commerce replaces ecommerce. We think it reshapes it.

The product record stops being a back-office detail and becomes the surface the customer actually meets. Central is built for that world. The same record that powers your product content today is the layer that prepares your catalog for agentic commerce tomorrow.

sixa AG
the sidestoryCentral was founded by Ian Schäfer, who also founded sixa AG, a Swiss digital enablement agency for brands, webshops, and marketplace sellers. sixa is where the pattern matching for Central happened. Central is independent, but sixa ran the experiments that proved it had to exist.
sixa.ch

We are not the first to notice that product content is broken. We may be the first to treat it as an operating system problem instead of a copywriting problem, because that is the problem we actually had.

If you have a catalog that frustrates you and a buyer base that is starting to ask AI agents the questions you used to answer on your product pages, we built Central for you. Paste a shop URL. We will show you what the record could look like.

Ian Schäfer Founder, Central · ian@central.to
Read the manifesto ↗ About sixa AG ↗
see it

The fastest way to understand Central is to put one product through it.

Paste a shop URL or upload a catalog. We gather the missing facts, verify every claim, and turn one record into pages, feeds, schema, widgets, and AI-readable data.

Try with your shop URL Talk to us ↗