The Internet Broke Product Reviews and Shop Avorio Wants to Build a Trust Layer

After careers building infrastructure at Meta, Microsoft, Google and AWS, engineers Anshuk Chakraborty and Debatri Mitra are turning their attention to a surprisingly difficult computing problem: figuring out what people actually think about the things we buy.

There was a time when buying a pair of headphones online seemed relatively straightforward.

You searched. You compared a few models. You looked at the star ratings, skimmed several reviews, and made a decision.

Today, that same purchase can feel like an investigation, and Shop Avorio is changing that.

Search results are packed with affiliate roundups. Social feeds contain influencers who may have been paid to recommend what they are holding. Marketplace reviews can range from genuinely useful to suspiciously enthusiastic. Reddit may offer candid opinions, but finding them means digging through sprawling threads. YouTube can be invaluable, assuming you are willing to watch several 18-minute videos before discovering whether the headphones actually survive a run in the rain.

And then there is artificial intelligence, which has made producing convincing product copy cheaper and easier than ever.

The problem confronting online shoppers is no longer a lack of information.

It is too much information, with too little certainty about which information deserves to be trusted.

Real and honest product reviews is the problem Avorio is trying to solve.

The shopping technology platform, co-founded by engineers Anshuk Chakraborty and Debatri Mitra, is designed to sit between the consumer and the increasingly chaotic universe of online recommendations. Instead of asking shoppers to bounce between retailer listings, Reddit posts, YouTube videos and social feeds, Avorio gathers opinions from across those sources, attempts to remove sponsored and AI-generated material, and turns the remaining signals into a simpler assessment of a product.

On its website, Avorio describes the result as a combination of authentic reviews, summarized pros and cons, a “trust score,” and price comparison. A shopper can paste a product URL or describe what they are looking for in ordinary language. Avorio says its system then searches sources including Reddit, YouTube and social media, filters material it considers sponsored or artificial, and surfaces the conclusions alongside available prices. The company says it has analyzed more than 50,000 products and that roughly 12 percent of the reviews it processes are filtered as sponsored.

It is an ambitious proposition because Avorio isn't really trying to build another shopping search engine.

It is attempting to build something more elusive: a trust engine.

The founders' backgrounds make their choice of problem particularly interesting.

Chakraborty has spent more than 15 years working on large-scale software systems, according to a biography provided for this article, with experience at AWS, Google and Meta. At Meta, where he serves as a senior staff software engineer, his work has included improving the efficiency of large advertising systems.

His career has also centered on a problem that increasingly defines modern computing: how to make enormous technical systems optimize themselves.

Chakraborty has worked with machine-learning models and teams of AI agents designed to automate infrastructure optimization and accelerate the development of AI features. He holds a master's degree in computer science from the University of Illinois Urbana-Champaign, with a focus on machine learning.

Mitra comes from a similarly infrastructure-heavy world.

During more than a decade in software engineering, she has specialized in systems designed to keep large applications reliable and responsive. According to her supplied biography, her work at Microsoft included redesigning core infrastructure supporting Microsoft 365 services, including systems used by Microsoft Copilot.

She has also worked on generative-AI platforms and autonomous agents intended to help engineering organizations diagnose and resolve software problems. Mitra earned a master's degree in electrical engineering and computer science from the University of California, Irvine.

These are not, in other words, founders whose professional lives began in fashion, retail merchandising or influencer marketing.

They are systems engineers.

And that distinction helps explain Avorio.

To a consumer, the modern product-review ecosystem might look like a media problem: too many advertisements, too many dubious recommendations, too many voices competing for attention.

To an infrastructure engineer, it can look different.

It looks like corrupted data.

The signal exists. It is simply buried in noise.

For years, technology companies have tried to make shopping frictionless. One-click checkout removed steps from purchasing. Recommendation engines reduced the work involved in discovering products. Social media transformed friends, creators and strangers into a vast distributed sales force.

Generative AI is accelerating that trend again.

But every reduction in friction creates another question: Who or what is influencing the decision?

The web's original promise was that access to more information would make consumers better informed. Consumers now encounter product information within an ecosystem where reviews, advertisements, affiliate links, recommendations, algorithms and entertainment frequently overlap.

Avorio's bet is that this confusion creates an opening for a new category of intermediary.

Its interface is deliberately uncomplicated. A user can ask for something like sweat-resistant running headphones below a certain price or provide the URL of a product already under consideration. The platform says it gathers opinions from multiple sources, extracts recurring positives and negatives, filters questionable material and searches retailers for better prices.

There is also a browser extension intended to make Avorio available directly on product pages, and the company offers an iPhone app. The browser tool is designed to display analysis without requiring the shopper to leave the page they are viewing.

That last detail may be more consequential than it sounds.

Consumer technology succeeds not merely when it can answer a question, but when it can answer the question at precisely the moment someone needs it.

For Avorio, that moment is the few seconds before a purchase.

Of course, introducing software to solve a problem partially created by software raises an obvious challenge.

How does an algorithm decide whom to trust?

Distinguishing authentic enthusiasm from paid promotion is difficult even for humans. Sponsorships can be clearly disclosed, subtly disclosed or not disclosed at all. A creator may genuinely love a product while also receiving compensation for discussing it. A Reddit recommendation could come from an ordinary customer, a brand representative or an account operated specifically to manufacture buzz.

AI-generated content creates another layer of complexity.

Text produced by modern language models is increasingly difficult to distinguish from human writing based solely on style. And as synthetic content gets better, systems designed to detect it risk becoming part of an endless technical contest between generation and detection.

Avorio therefore faces the same problem confronting search engines, social networks and AI companies more broadly, trust is not a binary classification problem.

A piece of content cannot always be neatly stamped, authentic or fake.

What a platform can do is assemble signals.

That is where the founders' experience with large-scale infrastructure becomes relevant. Systems engineering is often less about obtaining one perfect piece of data than about combining imperfect signals, assigning confidence, detecting anomalies and producing something reliable enough to use.

Avorio applies that philosophy to shopping.

The company says it cross-checks multiple sources rather than depending on a single review ecosystem. Its product pages turn that material into a trust score and summarize frequently mentioned strengths and weaknesses.

The goal is not necessarily to prove that every opinion is genuine.

It is to make a better judgment than a shopper could reasonably make while standing in a store aisle or staring at 37 browser tabs.

There is, however, an important paradox at the heart of any company attempting to become an arbiter of commercial trust.

Eventually, consumers must trust the arbiter.

Avorio itself participates in the economics of online commerce. The company states that some links on its platform are affiliate links and that it may receive a commission when users make purchases. It also says those commissions do not influence a product's trust score or ranking.

That disclosure matters.

Affiliate commerce is not inherently incompatible with objective recommendations, but it creates precisely the kind of incentive structure Avorio is asking users to scrutinize elsewhere on the internet.

For the company, transparency will therefore be more than a legal footnote. It may become a central product requirement.

If Avorio grows, users will likely want to know not merely what its system recommends but why. They may want to see which sources contributed to a score, why certain material was rejected, how commercial relationships are separated from ranking systems and how conclusions change when new evidence appears.

In an AI-mediated world, explainability can become part of the user interface.

That may ultimately be one of Avorio's biggest opportunities.

For decades, e-commerce condensed consumer sentiment into one extraordinarily powerful symbol: five stars.

The system worked because it was simple.

But simplicity depended on an assumption, that the people assigning those stars were basically who they appeared to be.

That assumption is becoming harder to maintain.

The next generation of shopping tools may therefore look less like review databases and more like intelligence systems. Instead of showing consumers 3,000 comments and asking them to decide, software will read those comments, compare them with videos and discussions elsewhere, identify suspicious patterns and report back.

In that world, the competitive advantage may no longer belong to the platform with the most reviews.

It may belong to the platform that can best determine which reviews matter.

That is the larger technological bet behind Avorio.

Chakraborty and Mitra have spent their careers working on systems most consumers never see: infrastructure, optimization layers, distributed services and AI tools operating beneath enormous applications.

Now they are trying to build another invisible layer.

Only this time, it sits underneath a decision nearly everyone makes.

Should I buy this?

The answer used to come from a star rating.

Avorio is betting that the future requires something considerably more sophisticated.