LightSource vs. Pactum

Where Pactum is genuinely strong

Pactum pioneered autonomous supplier negotiation and has run it in production at serious scale since 2021. Its AI agents negotiate payment terms, rebates, discounts, and price-list updates with thousands of incumbent suppliers in parallel, inside the Coupa and SAP Ariba flows a large enterprise already runs. The public results are real. Walmart's pilot closed 64 percent of agreements against a 20 percent target, and at scale Pactum reports roughly 3 percent average gains and an average 35-day payment-term extension at about a 68 percent supplier agreement rate, per its own pages. Maersk has used it for spot trucking rates since 2021, with Pactum reporting 96 percent lane conversion at prices 22 percent below human test groups. The company says its largest single autonomously negotiated deal was $140.5 million and its fastest took 87 seconds.

Two things stand out beyond the numbers. Suppliers appear to like the experience: Bloomberg reported that 83 percent of Walmart's suppliers found the bot easy to use, and Pactum claims 82 percent supplier satisfaction overall. And it creates value where competitive sourcing is not an option: sole-source renewals, incumbent price lists, and working-capital campaigns across a long tail no buying team has hours to touch. The backing matches the traction: a $54 million Series C led by Insight Partners in June 2025, more than $100 million raised in total, and more than 60 global brands as customers, including Honeywell, Novartis, and Tetra Pak.

Where LightSource is the system of record

LightSource does a job Pactum does not attempt: deciding who should make a part, and at what cost structure. That work starts from the part itself. Items arrive by AI ingestion from Excel, PDFs, and drawings into a revision-managed item master. Nested BOMs carry snapshots, diffs, and cost roll-ups. RFQs go out at part level with drawings attached, from three-bids-and-a-buy to multi-round strategic events, against a network of 20,000 live suppliers where onboarding is free and takes under 30 minutes.

Quotes come back in any format and AI normalizes them into comparable line items. LightSource is the only sourcing solution with true flexible cost breakdowns: category-specific bid sheets per event, so a casting quote and a PCB quote each get compared on the cost structure that matters for that category. AI drafts award scenarios, including split awards, a buyer approves the decision, and awards become tracked contracts with per-part, per-supplier price history that survives buyer turnover. Manufacturers from Infinite Machine to BRP, Canada Goose, Shure, and Yum! Brands run this loop today, live in days on flat files and sourcing in about 30 days, with implementation included.

None of this exists in Pactum, and that is by design. Its public documentation has no item master, BOM, revision, drawing, or award-scenario concept; engagements begin with CSV templates of supplier contacts, current terms, and spend history. When the question is a new part, a design revision, or a competitive award, there is nothing for a negotiation agent to start from.

Different units of work: the agreement versus the part

The structural difference between the two products is what each one models. Pactum models the agreement: who the supplier is, what the current terms are, what spend history and market indices say the terms should be, and how far an agent may move in a negotiation. Its direct-materials product, Price Lists, follows the same logic. By Pactum's own scope statement it targets the mid tier and tail of direct-material spend, meaning C-class parts, MRO items, and packaging, renegotiating existing price lists with incumbent suppliers when index data says prices have drifted from market. Material Intelligence, added in July 2026, strengthens that lens by mapping each material to commodity, energy, freight, and currency indices. It is a market-drift model, not a cleansheet built from part geometry or process steps.

LightSource models the part. A revision-managed item with drawings sits at the center, BOMs aggregate items into products, RFQs carry the engineering context to suppliers, and cost breakdowns decompose what comes back. That model is what lets a buyer answer questions Pactum does not pose: which of five capable suppliers should make this bracket, what happens to landed cost if the award splits two ways, what revision C did to assembly cost.

The overlap is narrow: tactical quote collection on simple buys, which either product can run, and commodity SKU price lists, where Pactum's index renegotiation brushes against LightSource's per-part price history. Pactum has genuine manufacturing customers, and five manufacturers adopted Price Lists at launch, so it does touch direct-material spend. It touches the price of things you already buy from suppliers you already have; the sourcing decision itself stays upstream.

Running both: award the part, then automate the terms

For a manufacturer with engineered A and B parts plus a long tail of C-class and packaging spend, the two products can sit side by side without contention. LightSource runs the sourcing decision: ingest the BOM, send part-level RFQs against the network, analyze cost breakdowns, model award scenarios, and record the award as a tracked contract with price history. Pactum then works the resulting supplier base between sourcing cycles: campaigning payment terms and rebates across hundreds of incumbents, holding tail price lists to market indices, and negotiating sole-source renewals where a competitive event is not worth running.

In our assessment the boundary is clean because each system stops where the other starts. Pactum does not create suppliers, parts, or awards. LightSource does not autonomously negotiate; its AI recommends and a buyer owns every decision. LightSource's GraphQL API mirrors the whole platform, so item, quote, and award data can feed whatever negotiation or P2P agents an enterprise runs downstream.

What public reviewers say

On G2, LightSource holds 4.9 out of 5 across roughly 11 reviews, while Pactum's G2 profile has no reviews as of September 2026. Both counts are small, so read them as anecdotes rather than statistics. The zero is not a quality signal: Pactum sells large enterprise programs to a base of more than 60 global brands, a segment that rarely writes G2 reviews, and there are no public G2 quotes to draw on. The small number of verified enterprise reviewers who have written about Pactum elsewhere praise it for automating supplier negotiations at a scale humans cannot match and for finding value in long-tail agreements, while cautioning that results depend on clear process design and good data. The strongest public endorsement may come from the other side of the table: Bloomberg reported in 2023 that 83 percent of Walmart suppliers described negotiating with the bot as easy, and that suppliers liked being able to make counteroffers.

LightSource's reviews are specific about why teams adopt it. A vice president of supply chain wrote: "Having the cost and quoting information in a single location helps with rapid product launches and better visibility for our leadership group. It's got its quirks, but the constant updates are promising." G2 reviewer (Mid-Market, 2024). Another reviewer's caveat doubles as a read on the supplier network: "The biggest challenge of LightSource is just getting the suppliers to try it. Once they try it --- they're also hooked." G2 reviewer (Mid-Market, 2024).