Strategy, Product, and Engineering: Building ShermansTravel's AI Search and Compare Engine

Case Overview

ShermansTravel built its business on two things: expert editorial advice and hand picked travel deals that beat what travelers find anywhere else. With 4 million newsletter subscribers and 1.5 million monthly visitors, the brand is one of the most trusted travel curation voices in the US market. But the product experience didn't deliver on what the brand promised. To use the site, a traveler had to know what they were looking for, fill in generic filters, then do the real work themselves: open 15 browser tabs across partner sites, paste details into a notes app, and try to figure out which $1,400 vacation package was actually the best.

ShermansTravel asked us to design and build the AI engine that would become their new core product. Not a feature on top of the existing site. The thing the homepage opens to. The mechanism that turns their editorial brand promise into a product experience. We worked as their product, AI, and engineering team alongside their team end to end: business consulting, product strategy, UX, AI design, back end, front end, and partner integrations.

The result is a single natural language search box that ingests over 10,000 deals from across 8+ booking partners, lets users refine progressively in plain English without losing their place, and generates written AI comparisons across any two or three deals the user selects. It's the homepage. It's the differentiation. It's the product that defines ShermansTravel today, and it's built as a platform foundation that ShermansTravel can keep extending for years.

Company Background

ShermansTravel is a leading US travel media company founded in 2002 by Jim Sherman, in original partnership with The New York Times. The company curates hand picked deals across vacation packages, cruises, hotels, and travel inspiration, reaching 4 million newsletter subscribers and 1.5 million monthly site visitors. The business operates across two brands: ShermansTravel.com for the mainstream market and SmartLuxury.com for upscale travelers, and works with hundreds of booking partners across the travel industry. Their entire model depends on converting editorial trust into qualified, high intent traffic to those partners, which means the discovery and comparison experience on site directly determines partner revenue, audience retention, and brand differentiation.

The Challenge

Pain Point	What It Actually Looked Like
The Brand Promised Curation, the Product Felt Generic	"Expert Advice & Hand Picked Deals" was what the brand stood for, but the search experience looked like every other travel aggregator. The differentiation lived in the editorial team's writing, not in the product itself.
Forms Don't Capture How People Actually Travel	Real travel intent sounds like "a summer week in Paris up to $2500 per person." Standard dropdowns force users to pick a city, a date range, a budget, and a trip type in isolation, and they don't capture how travelers actually think about a trip.
Deals Lived Across Partners, Not Across Trips	ShermansTravel works with 8+ booking partners, each with their own catalog, pricing, and presentation. A user looking for "Paris" was effectively being asked to reconcile 8 different sets of results across multiple browser tabs.
Comparison Was Manual and Painful	Once a user narrowed it down to 2 or 3 contenders, comparing them was the hardest part of the journey. Different durations, different inclusions, different "save 50%" framings. Most users gave up and picked the first one that felt close enough.
The Whole Business Depends on Conversion	As a media business, ShermansTravel monetizes through qualified clicks to booking partners. Every step of friction in the discovery and comparison flow directly costs partner revenue. The product wasn't a side project. It was the conversion engine the business runs on.
The Product Had to Scale Beyond Day One	Whatever got built needed to work for the next decade: more partners, new inventory types, the SmartLuxury brand surface, future AI capabilities. A point solution wouldn't do.

Our Strategic Approach

ShermansTravel didn't hire us to ship a feature. They hired us to design and build the core product that would define their next chapter. We operated as a full product partner across every layer of the work: business consulting, product strategy, UX, AI design, back end engineering, front end engineering, and integration with their existing partner network.

Phase 1: Consult, Understand, and Codesign the Product

Before any code, we embedded with the ShermansTravel team to deeply understand the business: how editorial works, how partner relationships are structured, where revenue comes from, what the brand stands for, and what a traveler's real decision journey looks like from inspiration to booking click. From there we ideated the product direction together, scoped the architecture, made the calls on what the experience needed to be, and ran the product management discipline end to end. The product strategy was as much of the deliverable as the code.

Phase 2: Unify the Partner Catalog Into One Searchable Layer

We built the back end that ingests, normalizes, and ranks deals across all 8+ booking partners. A "Paris" search now returns one merged result set of 10,000+ deals, with partner provenance preserved on each card and hand picked editorial selections tagged and surfaced first. For the first time, ShermansTravel's curation has a real product expression that maps to the brand promise.

Phase 3: Build the Full AI Search and Comparison Experience

We reframed the search experience around intent instead of fields. A single natural language input ("a summer week in Paris up to $2500 per person") parses into structured parameters under the hood. Every result page exposes the current search as visual filter chips at the top, plus a refinement input that accepts more natural language. Users can select any 2 or 3 deals and click Compare to get a side by side table covering price, duration, inclusions, departure flexibility, partner, and trip type, plus a written AI summary that calls out the tradeoffs in plain English: best value per night, broadest itinerary, lowest total cost, which option fits which traveler. The hardest moment in the booking journey now has expert opinion built into it.

Phase 4: Productionize as a Platform, Not a Project

The engine is architected as a platform foundation, not a feature. New booking partners can be added without rebuilding the integration layer. New inventory types (hotels, flights, future verticals) can be plugged into the same search flow. The SmartLuxury brand surface can run on the same engine with a different presentation layer. Latest AI capabilities can be slotted in as the underlying models evolve. ShermansTravel didn't buy a product. They bought a foundation they can keep building on.

Solution: The ShermansTravel Search and Compare Engine

What It Handles	How It Works
Natural Language Search	Users describe the trip they want in plain English. The engine extracts destination, dates, duration, budget, trip type, and traveler count, then runs the search. No multi field form to fill in.
Unified Deal Catalog	10,000+ deals from 8+ booking partners ingested and normalized into a single result set. One search, one ranked list, partner provenance preserved on every card.
Hand Picked Surfacing	Editorial selections are tagged and surfaced first, giving ShermansTravel's curated picks a clear product expression that maps to the brand promise.
Progressive Refinement	Visual filter chips at the top of every result page, plus a refinement search bar that accepts natural language. Each refinement narrows the result set without losing what the user has already built.
Cruises and Vacation Packages, One Engine	Both inventory types live in the same search flow, with simple toggles to show only one or the other. Users don't switch sections, they switch filters.
Side by Side Comparison	Select any 2 or 3 deals, click Compare, and the system produces a structured comparison covering price, duration, type of trip, partner, departure flexibility, and inclusions.
AI Written Comparison Summary	Above the comparison table, an AI generated paragraph calls out tradeoffs in plain English: best value per night, broadest itinerary, lowest total cost, which option fits which traveler.
Platform Architecture	Extensible to new partners, inventory types, and brand surfaces (including SmartLuxury) without rebuilding the experience.

Results After Launch

A core product, not a feature. The search and compare engine is now the homepage, the primary differentiation, and the experience that defines ShermansTravel in the market. The brand promise of expert curation and hand picked deals finally has a product surface that delivers on it.

One unified search across the entire partner network. Travelers searching for a destination now see every available deal across 8+ booking partners in one ranked list, with hand picked editorial selections surfaced first. The previous 15 tab research process is collapsed into one screen.

Natural language replaces form based search. Users describe the trip they actually want, in the words they actually use, and the engine handles the rest. The cognitive load of translating intent into checkboxes is gone.

AI written comparison closes the decision gap. Once a user picks 2 or 3 contenders, the system produces a written breakdown of tradeoffs alongside a structured comparison table. The hardest moment in the booking journey, deciding between options, now has expert opinion built into it.

A platform ShermansTravel can build on for years. The engine is structured so new partners, inventory types, and brand surfaces can be added without rebuilding the experience. The product is the foundation, not the destination.

An end to end product engagement, delivered as one team. ShermansTravel didn't manage a vendor. They worked with a partner who owned business consulting, product strategy, UX, AI design, back end engineering, front end engineering, and partner integrations as one continuous deliverable.

Client Perspective

We had the brand, the audience, and the deals. We didn't have a product that actually delivered on the promise. The new engine is exactly that, and it sits at the center of our business now.

Mark Bloom

CEO • ShermansTravel

Takeaway

Most consumer businesses have the same gap. The brand stands for something specific. The audience trusts it. The product looks like everyone else's. Curation lives in editorial. Differentiation lives in marketing. The actual user experience is generic.

Closing that gap is rarely an automation project, and it's never a "ship a chatbot" project. It's a product project. It requires someone who can sit with the business, understand what the brand actually promises, design the experience that delivers on it, build it on the latest AI without making it gimmicky, and architect it so the work compounds instead of becoming legacy in 18 months.

For ShermansTravel, that meant designing and building the core AI product that now defines their business: a natural language search across 10,000+ deals from 8+ partners, progressive refinement that stays out of the user's way, and AI written comparisons that finally make the hardest part of the trip planning easy. The brand always promised expert advice. Now the product delivers it.

If your brand promises something your product doesn't yet express, and you need a partner who can own that build from strategy through scalable production rather than just executing a spec, this is what that looks like.