Traza’s $2.1M Bet: Autonomous AI Agents for Industrial Procurement

Procurement in large manufacturers and construction firms moves billions of dollars each year, yet much of the work still runs on email, spreadsheets, and phone calls. New York–based startup Traza is betting that autonomous AI agents can finally modernize this neglected function — not by adding smarter dashboards, but by taking over large parts of the work itself.

Backed by a $2.1 million pre-seed round led by Base10 Partners, Traza is targeting the operational void between signed contracts and day-to-day execution, where enterprises quietly lose millions through missed savings, unmanaged suppliers, and process shortcuts.

The post-signature value gap: where procurement loses 11%

Despite years of investment in sourcing tools and contract management platforms, the fundamental economics of procurement execution remain stubborn. Research from World Commerce & Contracting and Ironclad finds that organizations lose an average of 11% of total contract value after agreements are signed — a phenomenon labeled “post-signature value leakage.”

Crucially, this 11% is not attributed to poor negotiation. As WorldCC’s Tim Cummins notes, the loss stems from how contracts are managed after signature. In practice, this gap reflects the frictions and blind spots in ongoing execution: missed savings opportunities, unauthorized changes, weak enforcement of terms, and underplanned renewals.

For a large enterprise with $500 million in annual contracted spend, that 11% translates into roughly $55 million disappearing each year. The losses are typically diffuse rather than dramatic — a price break not claimed, a volume commitment not monitored, a renewal rolling over on legacy terms. Because these gaps sit in everyday operational work, they are hard to see and even harder to systematically close with manual processes.

Operational constraints compound the problem. Most enterprises meaningfully engage with only their top 20% of suppliers. The remaining 80% — the long tail of vendors generating lower individual spend but substantial collective volume — often goes largely unmanaged. This is where vendor outreach, order tracking, invoice reconciliation, and compliance monitoring tend to be handled ad hoc, if at all.

Traza positions itself squarely in this operational layer. CEO and co-founder Silvestre Jara Montes argues that the 11% value gap spans commercial, operational, and compliance leakage, but that the operational piece is both the least addressed and the most recoverable. The company’s early deployments, while still limited, report a 70% reduction in human hours spent on procurement tasks and cycle times running three times faster than manual baselines.

From AI copilots to autonomous procurement agents

AI has been present in procurement software for years, but mostly as a recommendation or analytics layer. Incumbents such as SAP Ariba and Coupa, alongside newer players including Zip, Fairmarkit, and Tonkean, have used AI to suggest suppliers, flag anomalies, or prioritize tasks — while keeping humans responsible for nearly every decision and action.

Industry data underscores this implementation gap. Roughly half of procurement teams are piloting AI, but only a small fraction have scaled deployments that materially change operations. The prevailing pattern has been “AI as copilot”: tools surface insights; people still push the work through fragmented workflows.

Traza’s thesis is that current-generation AI agents have crossed a threshold. With multi-step reasoning, tool use, and contextual memory, the company believes agents can now execute full procurement workflows autonomously — from vendor discovery and RFQ management through to order tracking and invoice handling.

This is framed not as an incremental enhancement to systems of record but as a new product category: an AI workforce for procurement operations. In Traza’s view, existing platforms remain primarily data organizers. Their AI features sit on top of legacy architectures, providing recommendations that still require human follow-through. By contrast, Traza aims to replace much of the repetitive operational work outright, while still deferring key approvals to humans.

The broader market appears receptive to more aggressive automation. EY’s 2025 Global CPO Survey reports that 80% of chief procurement officers plan to deploy generative AI in some capacity over the next three years, with two-thirds viewing it as a high priority within 12 months. ABI Research finds that more than three-quarters of supply chain professionals already consider autonomous AI agents ready to handle core tasks like reordering, supplier outreach, and shipment rerouting — and early projects are showing 20–35% reductions in supply chain operational costs.

How Traza’s AI agents actually work

Traza’s deployment model centers on the operational work that typically lives in inboxes, spreadsheets, and loosely coordinated task lists. A standard RFQ (request-for-quote) flow illustrates the approach.

Once the need is triggered, Traza’s agent identifies suitable suppliers, drafts and sends RFQs, monitors responses, and automatically follows up when vendors are slow to reply. It then parses incoming quotes — regardless of format — and assembles a structured comparison view for a human decision-maker.

This pattern extends across adjacent workflows: vendor outreach, order tracking, supplier communications, and invoice processing. The agent is designed to handle routine interactions end-to-end, while routing higher-stakes decisions back to humans. The design principle is deliberate: autonomy in execution, human oversight at control points.

In practice, that means human approval is required for actions with material financial or compliance exposure, such as authorizing purchase orders above a defined threshold or resolving flagged compliance issues. Below those thresholds, the agent operates autonomously but logs every action for auditability.

For procurement leaders, one side effect may be as important as the automation itself: visibility. Long-tail supplier operations are often a “black box,” with limited consolidated insight into who is doing what, when, and under which terms. By executing and logging large volumes of low-visibility work, Traza’s agents can surface a more coherent picture of tail spend and supplier behavior — enabling governance improvements that are difficult to achieve with manual tracking alone.

Integrating on top of legacy stacks, not replacing them

Any new procurement platform must confront entrenched ERP systems, email workflows, and supplier portals that have evolved over years or decades. Traza’s answer is to integrate over the top rather than attempt wholesale replacement.

The platform connects via APIs or direct integrations into customers’ existing systems — spanning ERPs, email, and supplier portals. According to the company, it can reach across more than 200 enterprise tools. The AI agent then orchestrates work by reading and writing to these systems, instead of asking organizations to abandon them.

Go-to-market execution follows a similarly incremental logic. Traza typically starts with a two- to three-month proof of value focused on a single workflow. Integrations are built only at the key points required for that use case. As the engagement expands, additional integrations are layered in both within the account and across customers, so that each new deployment benefits from previously built connectors.

This high-touch approach — working side by side with customer teams during rollout — is notable for a company selling automation. It reflects the practical reality that process change, not just software configuration, is required when shifting from manual operations to AI-driven execution.

Traza reports that it is already working with large manufacturers and construction companies as paying customers, though it is not yet naming them publicly. The focus, the company says, is on turning pilots into durable production deployments rather than collecting experimental proofs-of-concept.

Competing on vertical depth in physical industries

Traza enters a competitive field. Large platforms like Coupa, Ivalua, SAP Ariba, Zip, Zycus, and Fairmarkit dominate enterprise procurement technology, while players such as Keelvar and Tonkean offer AI-enabled sourcing bots and no-code orchestration platforms.

Traza’s differentiation narrative centers on specialization. The company is focused explicitly on “physical industry” — manufacturers, construction firms, and other asset-heavy sectors where supplier relationships, compliance demands, and workflow complexity differ materially from software and services procurement.

In this context, the company argues that generic automation tools struggle with the realities of industrial procurement: extensive exception handling, strict regulatory constraints, and complex multi-tier supplier networks. Traza’s bet is that a vertically focused architecture and data model will better accommodate the nuance of these workflows.

Underpinning this is a data and learning strategy with two layers. At the agent level, each deployment contributes to an aggregate understanding of supplier behavior, RFQ dynamics, pricing anomalies, and workflow edge cases. At the data level, customer information remains isolated, preserving confidentiality while still enabling generalized learning across patterns of work.

If successful, this approach could create a compounding advantage: deep operational knowledge of how procurement actually runs in industrial environments, with all the workarounds and exceptions that seldom appear in RFPs. For buyers, the question will be whether this vertical depth translates into measurably better outcomes than horizontal AI platforms — particularly given the strong incumbency and existing relationships of traditional providers.

Founding team, investors, and the “Real Economy” thesis

Traza was founded by three Spanish entrepreneurs — Silvestre Jara Montes, Santiago Martínez Bragado, and Sergio Ayala Miñano — who came to the U.S. via the Exponential Fellowship, a program designed to bring European technical talent to build frontier AI companies in the United States.

The founders’ backgrounds combine industrial operations and applied AI. Jara Montes previously worked at Amazon and shipping giant CMA CGM, focusing on operations strategy and supply chain optimization. Martínez Bragado developed agentic AI systems at Clarity AI before joining Concourse as founding AI engineer. Ayala Miñano was a founding engineer at StackAI, an enterprise AI platform.

None of the founders comes from a chief procurement officer role, something the company acknowledges can surface in buyer conversations. Traza’s response has been to lean on results from early deployments and to work with senior procurement leaders as advisors — individuals who have managed procurement at the scale the startup is targeting.

The investor group reflects a focus on automating what Base10 Partners calls the “Real Economy” — sectors like supply chain and procurement that underpin physical industries but have historically been underautomated. Base10’s portfolio includes companies such as Notion, Figma, Nubank, Stripe, and Aurora Solar. Other investors include Kfund, a16z Scouts, Clara Ventures, Masia Ventures, and angels like Pepe Agell, who helped build mobile ad platform Chartboost to hundreds of millions of users and significant revenue before its acquisition.

Capital efficiency and the road to scale

At $2.1 million, Traza’s pre-seed round is relatively modest for a company selling into large enterprises, particularly given long sales cycles and integration complexity in procurement. The startup is intentionally constructed to stretch that capital.

One lever is talent strategy. Traza relies heavily on engineering talent in Europe, where the founders have deep networks and where, they argue, top AI engineers have fewer frontier opportunities than in the U.S. This geographic arbitrage, combined with a lean team structure, is intended to improve capital efficiency relative to peers competing for the same talent in San Francisco.

Commercially, the company is focused on rapid conversion from proof of value to paying partnerships, rather than extended pre-revenue pilots. Milestones for the next financing round are clear: more paying customers, stronger annual recurring revenue, and a repeatable, proven sales motion.

The three-year ambition is explicit: 20–30 large industrial enterprises in the U.S. and Europe running Traza across their procurement operations, with more than $1 billion in procurement spend flowing through the platform. Whether that target is realistic will depend on several external factors — including the ongoing maturation of AI agent capabilities and the pace at which traditionally conservative industrial buyers adopt autonomous workflows — as well as Traza’s own execution against incumbents and adjacent startups.

For procurement and supply chain leaders, the core question is practical: can autonomous agents reliably and audibly handle a material share of procurement execution, without introducing new risk? If Traza’s early results at scale hold — faster cycles, reduced human hours, and improved visibility across the supplier tail — the answer may determine not just the fate of one startup, but how quickly procurement moves from AI-assisted to AI-operated in industrial contexts.