What Just Happened

On August 12, 2026, Graas, a Singapore-based retail AI company, closed a $17 million Series B and acquired Trustana, an AI product data platform. (graas.ai) LemmaTree, the investment arm of Temasek (Singapore's sovereign wealth fund), led the round and became Graas's largest shareholder. The message behind the deal is unambiguous: AI agents cannot drive commerce without clean, structured product information.

For you, the DTC operator managing $500K to $5 million in revenue, this deal signals a shift in what your actual asset is. It's not your marketing. It's not your traffic. It's your product data. And for most of you, that data is a mess.

The Data's DNA: What Product Enrichment Actually Solves

I've reviewed hundreds of deals at the Angel Investors Network. The brands that sell for multiples on revenue share one characteristic: their product catalog is organized, complete, and consistent across every sales channel. The brands that struggle in diligence, or fail to get optionality, have the opposite problem.

Here's what product data enrichment means, stripped to specifics:

Structured product information. Your spreadsheet says "Water Bottle, 32 oz, Stainless Steel." A structured catalog says: [SKU=WB32SS, Category=Hydration, Material=Stainless Steel, Capacity=32 fl oz, Color=[Black, Blue, White], Price=USD 29.99, Compliance=[Food Grade, Leak-Proof]]. The difference is not cosmetic. The first blocks cross-sell recommendations, comparison shopping, and regulatory compliance. The second enables all three.

Consistency across channels. When you sell on your website, Amazon, TikTok Shop, and a wholesale network, each platform has different data requirements, character limits, and attribute taxonomies. Your website needs brand narrative. Amazon needs bullet-point specifications. Wholesale needs cost structure and bulk discounts. Without enrichment, you hand-edit descriptions for each channel, and errors compound. A word count violation on Amazon costs you rankings. A missing specification on TikTok Shop tanks the algorithm. A stale price in your wholesale system breaks distributor trust.

Cross-sell and recommendation signals. When product, customer, and inventory data live in a single ontology—Graas's term for a unified semantic database—AI agents can answer questions like "which products does customer cohort X buy together" and "which items are at risk of stockout before the demand spike in region Y." Without that unification, those questions remain unanswerable.

This is the core of the Graas-Trustana integration. Graas already managed customer history and inventory signals. Trustana adds the product layer. Together, they create a unified data structure that AI agents can reason across.

Why Standalone Trustana Failed (And What That Teaches You)

Trustana was founded in 2020 as an AI-native product enrichment platform. The technology was solid. It could take messy supplier catalogs:inconsistent attribute names, missing specifications, incomplete descriptions:and transform them into clean, structured, search-ready data.

Trustana's customer roster proved the value proposition: David Jones, Chemist Warehouse, Toys"R"Us. The problem, per Dealroom reporting, was that between 2023 and 2024, revenue fell 75 percent. Consecutive net losses followed.

Why? Because product enrichment as a standalone service depends on the AI commerce platforms that use enriched data growing fast enough to create demand. Trustana was brilliant at enrichment but had no anchor platform that deployed that enriched data continuously. CEO Rebecca Xing framed it carefully: the combination represented an "opportunity to accelerate growth." Translation: building an end-to-end stack alone is structurally difficult. You need demand-side integration.

Inside Graas, which processes more than $1 billion in live commerce transactions and serves enterprise clients like Unilever and Schneider Electric, that demand-side anchor exists. Enriched data is used immediately and continuously by active agents driving orders, recovering revenue, and pre-empting stockouts.

The lesson for you: If you're a $2 million DTC brand, your product data is not a standalone asset. It's only valuable when it's actively feeding sales operations:recommendations, inventory forecasting, channel optimization. If it sits in Airtable, it's a cost. If it drives agents, it's a moat.

The Commerce Knowledge Graph: What Unified Data Architecture Looks Like

Graas markets itself as a "System of Intelligence for Retail Commerce." That phrase describes a specific architectural choice. Instead of building separate AI tools for analytics, ordering, and customer service, Graas ingests all fragmented commerce data:orders, stock movements, customer interactions, invoices, support tickets:and consolidates it into a Commerce Knowledge Graph.

A knowledge graph in this context is not a database in the conventional sense. Where relational databases store data in tables and struggle with multi-hop queries, a graph architecture enables multi-hop reasoning. Example: "Which customers who bought product A in the past 90 days are also at risk of stockout on product B, and what's the optimal price point to recover their revenue in region C?"

Trustana provides the product layer of that graph. The acquisition completes the ontology. When product, customer, and inventory data live in unified structure, agents shift from answering questions to driving sales autonomously.

In Prem Bhatia's words: "When product, customer history and inventory data live in a single ontology, agents stop answering questions and start driving sales, both in ecommerce and in general trade."

That statement is worth reading twice. Agents stop answering. They start acting.

Why This Matters for DTC Operators: The Asia Precedent

Graas and Trustana operate primarily in Southeast Asia and India. Why does that matter for you?

Because the complexity of commerce in Asia outpaces the United States. A brand like Unilever operating across Indonesia, India, Malaysia, Thailand, Vietnam, and the Philippines manages six languages, six regulatory environments, and a split between modern e-commerce channels and "general trade":the informal network of small shops, distributors, and sales reps that still generates 75 to 80 percent of FMCG retail volume in many Asian markets.

Each Shopee market has different promotional mechanics and seller fee structures. The offline distribution layer generating no digital data historically. The regional complexity forces a choice: either build a fragmented stack of separately managed channels, or unify the data layer first.

Graas's bet is that unification first makes agentic commerce possible. That's a pattern that scales. If you're a DTC brand running Shopify, Amazon, TikTok Shop, and direct wholesale simultaneously, you face a parallel decision. Fragmented channels or unified data. The brands that choose unification will move faster and scale further.

Competitors and the Broader PIM Market

Product information management (PIM) is not new. Salsify, Akeneo, and Pimberly have established the category. The market is projected to reach $59 billion by 2034, per Acquia.

But traditional PIM solves the cataloging problem. Graas-Trustana solves the commerce problem. They're building not just a repository of product truth, but an active intelligence layer that uses product data to drive orders, capture revenue, and optimize inventory in real time. That's a different business model than traditional PIM licensing.

Salsify operates in the "system of record" space, focused on content syndication across channels. Akeneo focuses on PIM for manufacturers and multi-brand retailers. Pimberly is strong in food and beverage compliance. None of them are optimized to embed AI agents that *act* on enriched product data.

That's the whitespace Graas is occupying. Not "have clean product information." But "agents that use clean product information to drive sales."

What DTC Brands Typically Do (And Why It Breaks)

Most DTC brands under $5 million in revenue manage product data in spreadsheets or directly in Shopify. As you grow, this breaks in predictable ways.

Channel expansion: You add Amazon. You add TikTok Shop. Each platform has different requirements. You start hand-editing descriptions. Errors compound. Keyword optimization slips. Compliance flags accumulate.

Brand consistency: Your website says "Luxury waterproof jacket." Amazon says "Waterproof jacket." Wholesale says "Rain shell." A customer who visits your site then shops Amazon perceives fragmentation, not consistency.

Speed of launch: Seasonal drops and limited editions require you to go from new product to live on all channels in days, not weeks. Manual enrichment kills that timeline.

Small team burden: One person manages product data alongside five other jobs. Every manual step is friction. Every step is a place where errors hide.

Data completeness: You don't know which products lack images until a customer complains. You don't know which have incomplete specs until a channel flags compliance violations.

This is the exact problem Trustana was built to solve. Taking unstructured supplier catalogs:missing specifications, inconsistent names, incomplete descriptions:and transforming them automatically into clean, structured, search-ready data. Combined with Graas's agentic layer, that transformed data becomes a sales engine.

The Temasek Signal

LemmaTree becoming Graas's largest shareholder signals how sovereign wealth is consolidating around agentic AI infrastructure. LemmaTree was founded by Temasek specifically to incubate ventures around decentralized data and identity technology. Trustana was already a LemmaTree-backed company before this deal.

This is not a venture capital firm placing a diversified bet. This is a state investor consolidating a thesis: AI agents grounded in verifiable, structured product data will transform how enterprises buy and sell. Graas becomes the vehicle. Trustana becomes a component.

Glenn Gore, LemmaTree's CEO, spent seven years at Amazon Web Services as Chief Architect. His move to lead Temasek's AI portfolio, and now his endorsement of the Graas model, signals an architectural bet from the highest institutional level: the Commerce Knowledge Graph approach is the right foundation for agentic retail at enterprise scale.

For you: When sovereign wealth funds consolidate around a platform architecture, that architecture wins. It doesn't mean you have to use Graas immediately. But it does mean the direction is set. Clean product data feeding AI agents that drive orders. That's the architecture of commerce in the next 18 months.

What Gartner Says (And Why It Matters)

Gartner predicted in August 2025 that 40 percent of enterprise applications will feature AI agents by end-of-2026. For retail and DTC, that figure is already higher.

But here's the constraint: agents are only as good as the data they access. An agent that lacks clean product information cannot recommend, compare, or comply. An agent built on messy data makes messy decisions. It drives chargebacks, returns, and compliance violations.

For brands operating across modern and general trade simultaneously, data unification is the prerequisite. Graas is betting that brands that achieve it first will own the top of the funnel for the next two years.

Sources

FAQ

Q: Do I need Graas specifically to enrich my product data? No. You can use traditional PIM tools like Salsify or Akeneo. But those are cataloging systems. They organize what you already know. Trustana (now inside Graas) uses AI to *generate* structured information from incomplete raw data:a different problem. If your supplier catalogs are messy, you need enrichment, not just organization.

Q: What does integration take? Timeline? Graas has committed to integrating Trustana's capabilities into the Commerce Knowledge Graph by December 2026. For a DTC brand evaluating the platform, that timeline matters. You'll be on a migration path with clear milestones. Graas also processes more than $1 billion in live transactions daily, so it's battle-tested on scale.

Q: Is product data really my biggest sales bottleneck? If you're selling across more than one channel, yes. McKinsey's October 2025 analysis projected agentic commerce could orchestrate $3 to $5 trillion in global transaction volume by 2030. That's not just checkout:it's demand sensing, inventory pre-emption, and order capture. Agents that lack complete, consistent product information cannot execute any of those functions. Product data is not a back-office detail. It's the foundation of autonomous commerce.

Q: Does this mean I need to replace my PIM? Not necessarily. If you already run Salsify or Akeneo and they're working, you have a system of record. Graas is building the agent layer on top of unified data. Some brands will plug into their existing PIM. Others will migrate to a unified stack. The choice depends on your complexity. A $1 million DTC brand with 200 SKUs can probably get away with enriched Shopify + a spreadsheet. A $5 million brand with 2,000 SKUs across 10 channels needs infrastructure.

Q: When will DTC brands face real pressure to adopt this? Within 12 months. As agents become table stakes for enterprise commerce (Gartner says 40 percent of enterprise apps embed them by year-end 2026), DTC brands will discover they can't compete with enterprise scale without unified data. Shopify's AI features are improving. TikTok Shop is embedding agents. Amazon is deploying agents. The brands that have clean product data to feed those agents will move faster and convert higher. The rest will be left managing manual workflows in an increasingly automated market.

Capitalism Creates Value, Clean Data Compounds It

At the Angel Investors Network, I've seen brands with exceptional marketing, strong cohorts, and growth curves that looked unassailable fail in M&A diligence because their product catalog was a liability, not an asset. I've also seen unglamorous brands with boring marketing but pristine product data get acquired at multiples on revenue, because their operational infrastructure was clean enough to integrate at scale.

Data's DNA framework asks this: What signals does every customer decision leave behind? For DTC brands, the answer is product information: what you sell, how it's described, where it's available, how it's priced, what it does. When that signal is consistent, structured, and unified across all sales channels, it becomes a moat. When it's fragmented, it becomes a tax on every transaction.

The Graas-Trustana deal closes a gap. Graas had agents. Trustana had enrichment. Together, they have a platform where agents act on clean data. For you, the move is to start thinking about your product catalog not as a Shopify admin task, but as your core infrastructure asset. The brands that move first on unified product data will move first into agentic commerce. The rest will be slower, smaller, and harder to exit.

That's not a technology prediction. That's a capital allocation fact.

Sources:

  1. Graas.ai official announcement, August 12, 2026. "Graas Raises US$17M Series B and Acquires Trustana."
  2. TechTimes analysis, August 14, 2026. Ryan Cook, "Graas Raises $17M and Acquires Trustana: Temasek Bets on Unified AI Data for Asia Retail."
  3. Dealroom.co research note. "Graas raises $17M, buys Trustana in AI commerce push."
  4. Salsify PIM Software documentation. "Product Information Management: The Foundation for Great Product Experiences."
  5. Acquia Market Research. "Product Information Management Market to reach $59 billion by 2034."
  6. TidySKU Guide. "PIM for DTC Brands: Every Channel, One Source."
  7. McKinsey Quantum Black, October 2025. "The Automation Curve in Agentic Commerce."
  8. e-Conomy SEA 2025 Report, Google/Temasek/Bain, 2025. Southeast Asia digital economy and general trade analysis.