Search that reads the sentence, not the keyword.

Port2Port is South Africa’s leading online wine and spirits marketplace. In August 2026 Pivot Partners shipped a search box on port2port.wine that takes a sentence instead of a keyword. You type what the evening actually looks like, and it comes back with a short list and the reason each bottle is on it. Built with the Port2Port team.

Why keyword site search fails wine shoppers

Catalogue search is a filing system. It assumes you already know the name of the thing you want, or at least the category it lives in. That works for someone who has come to buy a 2021 Swartland Syrah, and it works badly for everyone else.

Wine makes the mismatch worse than most categories. The catalogue is indexed the way the trade thinks: producer, cultivar, region, vintage, critic score. Customers think in occasions: what is for dinner, who is coming, how much to spend. There is no field for any of that, so the customer has to translate their need into the shop’s vocabulary first. Most people do not bother.

The failure is not subtle. Typed into the old e-commerce site search, “a bold red to impress my dinner guests” returned zero results, while the bottle that answers it sat in stock the whole time.

What Pivot Partners built

One search box that accepts a sentence. It sits where the old search box sat, and the old behaviour still works: type a producer name and you get that producer.

Type a sentence and the suggestion panel opens with what the system understood, written out as labelled attributes, and a small number of specific bottles underneath, each with a one-line reason. Press through and the results page opens with the same short list on top of the full filtered catalogue, so you can carry on browsing the normal way.

Three real queries, and what the search understood

Run on the live site in August 2026.

The system does not force every sentence through the same fields. A braai query produces colour, style and food context; a budget query produces budget and intent. It reads what is there and invents nothing.

How hybrid search works: keyword and semantic search together

The build is called Hybrid Search because two different searches run at the same time and their results are blended.

1. Listen. It reads the whole request rather than picking keywords out of it.
2. Understand. It pulls out what actually matters in the sentence: colour, style, budget, the occasion behind the ask. Only what is there.
3. Match. Keyword search matches the words. Semantic search matches the meaning. Both run, and the results are blended.
4. Guide. A shortlist is chosen from what came back, each pick carrying a one-line reason drawn from that bottle’s own details.

Each search fails in a predictable way on its own. Keyword matching is literal: it finds the bottle whose name contains what you typed, and returns nothing for a sentence. Semantic matching finds bottles similar in character even when no words line up, but drifts away from the specific bottle you named. Run together, the shopper gets the bottle they asked for and the ones they would never have found by name. Type a producer and a vintage and the answer is as instant as it ever was.

The search shows what it understood

Whatever the search understood is written out on screen as tags. Not buried in a debug panel, not implied by the results, but stated plainly: red, bold and structured, to impress, dinner party, under R500.

This is the part that gets skipped most often, and it decides whether people trust the thing. A recommendation you cannot interrogate is a guess. With the reading on screen, a shopper can see whether they were understood and adjust.

The shortlist, and what it is not allowed to do

The rules around the shortlist matter more than the shortlist itself.

• Real bottles only. Picks are drawn from actual, in-stock listings, on price, rating and character. Nothing is generated.
• Reasons come from the bottle. Each line is grounded in that wine’s own details rather than written freely.
• No popularity arguments. The reason is never a version of “lots of people bought this”. That is a sales tactic, not advice.
• No commercial thumb on the scale. Nothing steers a shopper towards a pricier bottle to lift the basket.
• Fail quietly. If the intelligent path is slow or unavailable, the search falls back to keyword search instead of an error page.

What the numbers show so far

Port2Port’s analytics did not record zero-result searches before launch, so there is no before-and-after zero-result rate to report yet. What the search log does show is a change in how people type. Sentence-style queries, three or more words carrying an occasion, a food, a style or a budget, were about one search in a hundred in June and July 2026. In September and the first week of October they were about two in a hundred. Small numbers, and a doubling, on a search box that returned zero results for a sentence like these before August.

The queries themselves tell the story: “white wine for summer night”, “braai for 8 on Saturday, lamb and boerewors”, “best quality dessert wines”. The case is not that search got cleverer. It is that a customer who arrives with a sentence instead of a search term has stopped hitting a dead end.

How Pivot Partners worked with Port2Port

Pivot Ventures holds a stake in Port2Port, and Pivot Partners was the AI engineering and implementation partner on this build. In practice that meant sitting with the Port2Port team: working out how wine customers actually buy, agreeing what a good answer looks like, and shipping in pieces that could be used and judged along the way.

Most of the difficulty was never the model. It was the translation layer between how a customer talks and how a catalogue is indexed, and the judgement about what the system is allowed to say when it is not sure.

Read the AI order intake case study →

Try hybrid search on port2port.wine

Go to port2port.wine and type a real sentence into the search box. Something you would actually say. If it gets it wrong, we would like to know what you typed.