Pump

Pump's website describes a platform for cloud and inference cost management. It connects to cloud accounts (AWS, Google Cloud, Azure, and Oracle Cloud) via read-only access and shows spend, savings, and anomalies. The page says Pump buys pooled commitments and passes rates through, routes LLM requests to the cheapest model that meets a quality bar with fallbacks, allocates both bills by team and tag, provides budgets with Slack warnings, and checks spend hourly for anomalies. The page claims about 30% savings and about $10B+ spend managed, and says Pump is free to join. Customer quotes and a free-to-start onboarding flow are also presented. Specific savings figures and customer names are as shown on the page and are not independently verified.

Key features

• Pooled cloud commitments that Pump buys and passes through to customers
• Model routing to the cheapest model that passes a set quality bar, with fallbacks when a provider is down
• Tag allocation across cloud and AI bills, with suggested owners for untagged spend
• Budgets per team across cloud and AI spend, with Slack warnings before limits
• Hourly anomaly detection that traces unusual spend to a resource and person
• Rightsizing recommendations and savings tracking
• LLM cost auditing and invoice recap
• Read-only connection to AWS, Google Cloud, Azure, and Oracle Cloud accounts
• Reports and a natural-language assistant for cloud and inference spend questions

Best suited for

• Engineering and finance teams managing cloud spend
• Teams with growing AI or LLM inference costs
• Organizations seeking cloud cost savings without code changes
• Teams needing spend allocation by team or environment

Pricing

Pump is free to join according to the page; no paid tiers are detailed. Custom pricing is mentioned for enterprise bring-your-own-cloud arrangements.