NRF 2026: Innovations showcase

Startups bet on advanced AI commerce foundations, reports Miya Knights, Retail Technology Publisher, from the New York trade show

The Innovations Showcase tour at NRF 2026: Retail's Big Show last week offered a cross-section of technology developments that point to where retail is heading next.

But the innovations on show also highlighted why much of the "agentic" conversation is now converging on an unglamorous truth: AI-driven shopping and productivity will only scale if retailers fix foundational issues, including product data quality, integration, and execution loops first, which are critical for operational success.

Across a series of short pitches and Q&A moments during a guided NRF tour, startup founders repeatedly framed 2026 as a shift from experimenting with artificial intelligence (AI) interfaces to hardening the operational and information layers beneath them.

In this way, products can be discoverable across new AI surfaces, orders can be orchestrated across an expanding channel mix, and associates can act on recommendations with confidence.

AI-ready produce catalogues

Several pitches tackled a challenge that retailers increasingly recognise but rarely budget for properly: ensuring product data is complete, structured and consistent enough to be interpreted by AI agents and "product card" engines.

Scot Wingo, ReFiBuy CEO (who is also President and CEO of marketplace management software provider ChannelAdvisor— now known as Rithum), said the company's focus is on helping brands and retailers "prepare their products" for what he characterised as a shortened consumer journey driven by generative AI.

ReFiBuy's approach is to optimise and enrich catalogues, then test how products appear in gen AI engines and apply "nudges" to improve match accuracy and visibility, down to details such as model numbers and variant naming conventions.

Search shifts with AI attribution

A second, adjacent theme was proving commercial value from AI discovery — not simply assuming it.

A representative from Brandback described the Berlin-based company's proposition as centred on "AI discoverability" for Shopify brands. This includes "AI revenue attribution" to quantify how much money retailers are making from AI models such as ChatGPT and Gemini, and then optimising sites at both the category and product levels to lift performance.

Outcomes replace "glorified chatbots"

If data readiness was one cornerstone, measurable outcomes formed the other.

Aniket Deosthali, co-Founder of Envive AI, positioned his firm as deploying multiple AI agents "to drive the business outcomes that our customers care about". He described automated merchandising and product discovery designed to improve on-site search and performance in generative engines.

Deosthali claimed typical results of "two to three times" conversion uplift without its AI in place, arguing that many deployments still rely on manual tagging and rigid rules rather than automation that scales with long-tail consumer search queries, both on-site and via search or answer engines.

Trust signals: reviews and authenticity

Notably, the tour's strongest "trust" narrative came not from privacy statements but from the mechanics of customer content.

David Rapps, Founder and President of Wholescale, said his company is addressing a persistent ratings-and-reviews gap between brands' DTC sites and major retail marketplaces.

He pointed to examples where a product may have hundreds of reviews on a brand's own site but only a handful on a retailer's listing, arguing that existing syndication models have not modernised. Wholescale's pitch emphasised "democratising syndication" and highlighted authentication and moderation as core requirements.

First-party data beyond DTC

Another recurring motif was that omnichannel shoppers are more valuable. However, brands lose visibility when sales are made through marketplaces, retail partners, or increasingly within the browser or social media app itself.

A Brij spokesperson asked the tour group to consider how frequently consumers buy across Amazon, Target, and other channels, then argued that brands "know nothing about those customers" when purchases occur outside the direct-to-consumer (DTC) channel, despite the value of omnichannel behaviour.

Brij positioned its platform as a way to "acquire first-party data," analyse it, and "activate it" across channels, including via agent-based querying of customer data and modernised value-exchange mechanics (such as registration and rebates).

Turning AI into decisions

Several pitches focused on operationalising AI, aiming to empower retail professionals to use AI models as practical, decision-driving tools.

A HyperFinity representative cited an often-repeated industry problem: "about 85% to 95% of AI projects fail to make production". In response, they positioned the company as an "actionable intelligence platform" designed to industrialise retail decisioning across pricing, ranging and personalisation.

To explain the concept, the presenter described HyperFinity as "Spotify for retail," but one that recommends products and offers rather than music.

Store realities still dominate returns

For all the AI talk, two of the most grounded pitches were about physical-world problems retailers can cost-justify quickly.

A Bailiwick representative framed the company's proposition around "access and protection," arguing that stores need to protect associates and product while enabling customers to "just buy".

The pitch stressed deployment that remains "completely off [the] customer network," positioning the approach as reducing security risk while addressing the operational challenge of locked merchandise.

In grocery, Adam Lytle, Afresh Chief Revenue Officer, positioned his company as an "AI-native" fresh supply chain management solution, arguing perishables remain "the hardest thing to get right" in ordering and inventory management.
Lytle cited waste reduction ambitions, claiming Afresh can "reduce shrink by 25%" and "increase revenue by 3% on average," attributing the gains to fresher availability, fewer stock-outs, and more efficient turns.

Commerce operations "plumbing"

If there was one pitch that most directly connected "agentic" to integration, it was Pipe17.

Kelly Goetsch, Pipe17 President, said selling channels are proliferating — from Amazon and TikTok to Shopify and "agentic AI heads of various sorts" — while fulfilment is simultaneously fragmenting across multiple nodes, warehouse management systems (WMSs) and partners.

He positioned Pipe17 as a network connecting sales channels to systems of record, including enterprise resource planning (ERP) systems, data lakes and warehouses, with logistics providers, surfacing inventory and shipping information, and routing orders to the correct fulfilment location.

Goetsch also highlighted the practical complexities retailers routinely underestimate: limited customer data in marketplaces, stockkeeping unit (SKU) mapping across marketing-led variants, and constraints imposed by shipping regulations and service-level expectations.

Product intelligence deepens

The tour concluded with an introduction to startup Harmonya that framed the company as a data business generating "very granular product attributes" — reinforcing the Showcase's broader focus on product truth as the substrate for personalisation, merchandising and AI-led discovery.

Given its benefits, the Harmonya representative said his company was working with brands and retailers alike, including Nestlé, Mars, PepsiCo, and the Texan grocer H-E-B.

The Innovations Showcase's most apparent trend was not that retailers need "more AI," but that AI is now forcing retail to confront longstanding structural weaknesses: inconsistent product data, expensive integration work, fragile order operations and decision loops that fail to reach the frontline.

"Next now" innovations

The startups drawing attention at NRF 2026 are disproportionately those that (1) make catalogue and customer data more machine-readable, (2) reduce bespoke integration and orchestration burdens as channels multiply, and (3) convert AI into decisions retailers can execute: in pricing, ranging, ordering, loss prevention and fulfilment.

In 2026, the most credible "next now" innovation story is therefore less about the flashiest interface and more about the discipline to make agentic commerce safe, measurable and an operational reality.