Provenance Record

Well-structured funding story with careful sourcing discipline: attributes every figure and quote to its specific source (TechCrunch interview vs. PR Newswire release) rather than blending them, and explicitly calls out the customer-count discrepancy between the two rather than picking one.

Read both snapshots directly. source-0 (TechCrunch) confirms every figure and quote verbatim: the $550M valuation, $60M prior valuation, $10M Series A, $58.5M total funding, '5,000 customers... up from more than 700', the $10M ARR/10-employee/tens-of-millions-of-dollars/~30-employee figures, the $1M/year largest-customer detail, and the 'still a bottleneck' quote. source-1 (PR Newswire) confirms the 'diagnosing CI failures, autofixing them, and keeping pull requests green' quote, the founder names (Jayaprakash, Shah, Maru), the '6,000 companies... up from approximately 800' figures, the 5-10% week-over-week CI growth, the 'hundreds of thousands of cores' / 'order of magnitude' compute plans, and both Jayaprakash quotes ('Writing code has gotten dramatically easier...' and 'several times more pull requests...').

All specifics trace verbatim to their cited sources. The one genuine discrepancy in the underlying sources (customer count: TechCrunch's 5,000/700 vs. the press release's 6,000/800) is accurately preserved as a discrepancy in the article's 'What We Don't Know' section rather than silently reconciled or picked -- correct handling per the no-fabrication rule.

Approve. Every quote and figure verified verbatim against both sources; the cross-source discrepancy is handled correctly by disclosure rather than fabricated reconciliation. No corrections needed.