LightSource vs. Procure Ai

Where Procure Ai is strong: automating requests and tail sourcing over the stack you already own

Procure Ai was founded in London in 2020 by co-CEOs Konstantin von Bueren and Yves Bauer, and it has been agent-first from the start: the product is a set of more than 50 specialized AI agents, grouped into autonomous, collaborative, and ambient categories, running across the procurement lifecycle. The defining architectural choice, in Yves Bauer's words to tech.eu, is that the platform "sits on top of fragmented data landscapes and makes them intelligible, enriching what is there rather than replacing it." It connects to SAP ECC and S/4HANA, Ariba, Coupa, Ivalua, Jaggaer, Oracle Fusion, and Microsoft Dynamics, and it is listed in SAP's Spend Management partner directory.

The evidenced results sit in intake and tail sourcing. Procure Ai reports that roughly 60 percent of quote-to-order intake requests are handled autonomously, average savings of 4.7 to 4.9 percent on tactical and tail-spend negotiations, and more than 50 billion euros in annual spend managed. Its November 2025 seed round ($13 million led by Headline, with C4 Ventures and Futury Capital) came with a reported 4x year-over-year revenue growth and an expansion plan from its DACH core into the UK, Nordics, Benelux, and France. The customer list includes real manufacturers: DMG MORI, Kärcher, and GMH Gruppe, alongside EnBW, Dr. Oetker, and Diebold Nixdorf. The flagship manufacturing testimonial, from Kärcher, comes from a sourcing specialist for non-production material, which is consistent with where the product is aimed.

Where LightSource wins: parts with drawings, revisions, and a BOM behind every award

LightSource starts where Procure Ai's public materials stop: the part. AI ingestion turns Excel files, PDFs, and drawings into a revision-managed item master; BOMs nest, snapshot, diff, and roll up cost; RFQs go out at the part level with drawings attached. Cost breakdowns are the sharpest contrast: LightSource is the only sourcing solution with true flexible cost breakdowns, category-specific bid sheets configured per event, and its AI normalizes any quote format against them so line items like tooling, packaging, and logistics surface before award. AI-drafted award scenarios become tracked contracts, and every award adds to per-item, per-supplier price history that survives buyer turnover.

The supplier side is live rather than marketed: 20,000 suppliers on the network, free for suppliers, onboarding in under 30 minutes, quoting through Excel round-trips. Implementation runs in days on flat files, sourcing is typically underway in about 30 days, and implementation is included. Manufacturers from Infinite Machine to BRP, Canada Goose, HelloFresh, Shure, Serta Simmons Bedding, Yum! Brands, and Arxada run direct-materials sourcing on it; LightSource reports 45 percent faster sourcing cycles at six months and 9 percent average savings per cycle, and Infinite Machine reports $10 million saved in nine months. LightSource was named a Gartner Cool Vendor in 2025 and carries venture backing co-led by Lightspeed and Bain Capital Ventures ($33M).

The structural difference: an agent layer runs on the data underneath it

Procure Ai's own description is the right lens: it enriches what is already there. Its agents read the data that flows through an ERP and S2P estate, meaning requisitions, catalog items, supplier records, tenders, invoices, and the spend cube built from them. That is why it is strongest on intake and tail sourcing: the purchases where several interchangeable suppliers compete mainly on price, and where an agent can run the event end to end because nothing outside the transaction data decides the award.

Direct materials awards work differently. The decision hinges on a drawing at a specific revision, tooling amortization, material and process cost structure, and supplier qualification, and none of that lives in a spend cube. It lives in PLM and in the sourcing record. As of September 2026, procure.ai's public materials do not mention BOMs, engineering drawings, revisions, part-level RFQ lines, or cost breakdowns, so in our evaluation there is no public direct-materials story to assess. That reads less as a flaw than as a statement of what the product was designed for: an agent layer over spend data has nothing to orchestrate for a part that never becomes a record in it.

LightSource is built as the system of record those decisions need: item and BOM master data in from PLM, part-level RFQs and cost breakdowns in the middle, awarded pricing out to the ERP through a GraphQL API that mirrors the whole platform.

Running both: Procure Ai on requests and tail spend, LightSource on the product

For a manufacturer, the coexistence case is unusually clean because the two products barely touch. Procure Ai automates the request-and-buy layer for general, indirect, and tail spend inside the SAP, Ariba, or Coupa estate you already run. LightSource runs the direct-materials layer underneath: BOM ingestion, part-level RFQs, cost-breakdown analysis, award scenarios, and awarded pricing exported back to the same ERP. There is no system-of-record fight, because Procure Ai does not aim to be a system of record and LightSource is one for parts.

The overlap appears only at generic, price-competitive sourcing events where no drawing or cost structure drives the award; those could plausibly run in either product. Geography is the other practical filter: Procure Ai's go-to-market is European, DACH-core and expanding into the UK, Nordics, Benelux, and France per its own funding announcement, while LightSource serves North American and global manufacturers. A European enterprise with heavy indirect volume and an engineered product line is the profile where running both makes straightforward sense.

What public reviewers say

On G2, Procure Ai holds 4.7/5 across 9 reviews and LightSource holds 4.9/5 across a small early base of about 11 reviews, both as of September 2026. Both bases are small, so individual reviews carry more information than the star averages, and both links go to the full, unfiltered review pages.

A transparency note on Procure Ai's reviews: G2 restricts full review text, and the Internet Archive holds no snapshot of Procure Ai's review page, so we could not verify verbatim quotes and summarize the themes from G2's indexed snippets instead. The positives are specific and consistent with the product's design: reviewers credit the unified data store for putting all procurement data in one place and giving a complete picture, say analyses and decisions got much faster, and describe the support team as exceptional. The recurring criticism is that some processes are not integrated yet and still require manual work, a fair picture of an agent layer mid-rollout, where coverage arrives system by system.

LightSource's review page is publicly archived, and the quotes below are verbatim, verified against that archive. The most specific reviews describe the direct-materials workflow this page is about:

"The platform allows for the seamless addition of RFP items/SKUs and BOM details through a simple drop/drag feature... The platform also has a robust set of analysis tools to quickly compare the different supplier responses, calculate forecasted savings and identify best of the best pricing for the RFP... We typically see a company deploy LightSource within a day or two."

-- G2 reviewer (Mid-Market, 2024)

"Having the cost and quoting information in a single location helps with rapid product launches and better visibility for our leadership group."

-- G2 reviewer (VP of Supply Chain, Mid-Market, 2024)

On the critical side, the recurring LightSource theme is supplier adoption rather than the product itself: "The biggest challenge of LightSource is just getting the suppliers to try it. Once they try it --- they're also hooked," per a G2 reviewer (Mid-Market, 2024).