LightSource vs. Archlet

Where Archlet is strong: large category events and award optimization

Archlet earns its reputation on the sourcing event itself. Constraint-based award optimization is the company's founding specialty (it spun out of ETH Zurich in 2019), and it shows in the product: unlimited award scenarios with capacity limits, incumbency rules, regional splits, and discount structures, plus price heatmaps and cost-element breakdowns. PepsiCo has run sourcing on Archlet across 50 categories globally since starting with packaging in 2020, and Walmart, FedEx, TotalEnergies, and Estée Lauder appear on its public customer list. The platform added native eAuctions in February 2025 (English, Dutch, and Japanese formats) and Autonomous Sourcing in July 2026, which lets its Spark agents run tactical and tail-spend events end to end under policy guardrails. Its AI is genuinely embedded rather than added on, and reviewers back the usability claims: 4.7/5 across 46 G2 reviews, with Summer 2025 badges for fastest implementation and highest ROI. For a category team running large multi-supplier tenders in freight, packaging, or MRO, Archlet is a credible, modern choice, and in-platform market data from Xeneta and Expana strengthens it further in logistics and ingredient categories.

Where LightSource is strong: engineered parts, BOMs, and the award record

LightSource is built around the part. Items live in a revision-managed master: AI ingestion turns Excel files, PDFs, and drawings into item records (the feature processed 3,300+ items in its first month), and nested BOMs carry snapshots, revision diffs, and cost roll-ups. RFQs go out at the part level with drawings attached, AI normalizes quotes in any format for comparison, and awards become tracked contracts with per-item, per-supplier price history. Suppliers respond free after onboarding that takes under 30 minutes, and the network now has 20,000 suppliers live with AI capability-based discovery. The customer evidence comes from direct-materials manufacturers: Infinite Machine reports 3x faster time to market and 25% lower BOM cost, Arxada has awarded $18M through the platform, and BRP, Canada Goose, HelloFresh, and Shure run sourcing on it. LightSource was also named a Gartner Cool Vendor 2025 (Sourcing and Procurement Technology). None of this replaces Archlet's optimization math; it is a different center of gravity, the bill of materials rather than the bid sheet. See the product and customer results for detail.

Events end but parts persist: the structural difference

The overlap between these two products is real. Both run RFQs, both analyze quotes, both build award scenarios, and both describe themselves accurately as AI-native. The difference is what happens when the event closes. Archlet's unit of work is the event: bid sheets, scenarios, and auctions organized per tender. It does not keep an item or BOM master, and its own Direct Materials page (which covers resins, metals, ingredients, and components) does not mention BOMs, drawings, revisions, or part-level history. In LightSource, the event is one step in the life of a part. The quote a supplier gave on revision C, the award that followed, the cost walk since then, and the contract expiry all attach to the item, so the next buyer who touches that part inherits the full history. For commodity categories, event-by-event working may be all you need. For engineered components, where the same part is sourced, revised, and resourced over years, where that memory lives is a bigger decision than which tool runs the better single event.

Running both: a workable split with honest overlap

The coexistence story here is narrower than with a procure-to-pay suite, because both products are sourcing tools. There is still a coherent split for a manufacturer: keep Archlet where its optimization and auctions shine (indirect, logistics and freight, packaging, and tail-spend events), and run LightSource as the system of record for engineered direct materials, where sourcing has to attach to items, BOMs, drawings, and qualified suppliers. At many manufacturers, engineered-component sourcing still runs through Excel and email even with an e-sourcing tool in place; that gap is what LightSource fills, and neither tool depends on the other, so there is no rip-and-replace. Where the overlap is genuine, in commodity direct categories like resins or metals, most teams will consolidate into whichever platform holds the surrounding data. For a manufacturer, that is usually the platform holding the BOM. Either way, LightSource exports awarded pricing to your ERP through its GraphQL API.

What adding LightSource looks like: live in days, sourcing in about 30

Implementation speed is not the dividing line in this comparison. Archlet claims two-to-four-week setups and holds a G2 Fastest Implementation badge, and LightSource typically has customers live in days on flat files and running sourcing within about 30 days, with implementation included in the subscription. The difference is what you stand up. Starting LightSource means building the item and BOM master, and AI ingestion does that from the Excel files, PDFs, and drawings you already have, so clean data is not a precondition. Most teams start with a one-program proof of concept before a wider rollout. Suppliers onboard free in under 30 minutes through an RFx invite, which is why adoption has held up at customers like BRP and HelloFresh. Nearly half of LightSource's staff works in deployment and implementation, all ex-McKinsey, and your IT team can integrate at its own pace through a GraphQL API that mirrors the platform.

What public reviewers say

On G2, Archlet holds 4.7/5 across 46 reviews and LightSource holds 4.9/5 across a small early base of about 11 reviews, both as of September 2026. The review bases differ by an order of magnitude, so read them differently: a large base shows patterns at scale, a small base shows what early customers say. Both links go to the full, unfiltered review pages.

The most useful Archlet reviews come from the category buyers the product is built for, and the recurring praise squares with our read above that Spark's AI is genuinely embedded rather than added on:

"I've been using Archlet for the past year, and it has truly transformed how I work as a Category Manager... it leverages generative AI to automate many of the tactical tasks that typically take up so much of our time... It's a game-changer for any Buyer or Category Manager who wants to focus more on strategy and value creation and less on repetitive tasks."

-- G2 reviewer (Category Manager)

The critical reviews are worth reading closely because they concern the scenario engine itself, the product's flagship, rather than peripheral features:

"For more complex tenders with many variables, the best price scenario analysis can get complicated."

-- G2 reviewer, who adds that the team has sometimes finished the analysis offline

Beyond that, the recurring dislikes on the same page are about depth at the edges: reviewers describe the supplier management capabilities as limited and note missing features compared with established suites, the same limits our capability grid records.

LightSource's base is smaller, and the most specific reviews describe the two things this comparison keeps returning to: sourcing cadence and where cost and quote data live.

"Relaunching an existing RFP literally takes seconds and can be done with a click of a button."

-- G2 reviewer, on frequently re-quoting commodities whose prices fluctuate

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

-- G2 reviewer