Galileo

Primary category: ai-red-teaming (with heavy llm-observability overlap). Acquired by cisco, folding into Splunk Observability.

One-liner — An AI/agent evaluation and observability platform that measures LLM/agent output quality, catches failures (hallucination, bias, unsafe output) before they reach users, and enforces real-time guardrails across the agent development lifecycle.

What it does — Galileo instruments the full agent development lifecycle (ADLC) — prompt optimization, model selection, offline evaluation, and production monitoring — and adds real-time guardrails for multi-agent systems. It scores responses for quality, hallucination, bias, safety and cost, so teams can both evaluate models pre-deployment and watch them in production. It straddles two jobs: evaluation/guardrails (red-teaming-adjacent) and observability (tracing/monitoring AI apps).

Where it sits in the stack — Primary ai-red-teaming/guardrails with a strong llm-observability leg at the model/prompt layer. Risk role: mainly screening untrusted input (guardrails catching jailbreaks/unsafe prompts and outputs); its bigger contribution is trust/quality assurance rather than blocking egress. Adjacent to langsmith, arize-phoenix, langfuse, braintrust (observability/eval) and patronus-ai, lakera, promptfoo (eval/red-teaming).

Deployment & architecture — SaaS with SDK/API instrumentation dropped into the AI app/agent pipeline; supports offline eval and live production monitoring. Post-acquisition, Cisco is integrating Galileo into Splunk Observability Cloud, extending Splunk’s AI Agent Monitoring across the full ADLC with guardrails.

Positioning & differentiators — Known for “evaluation intelligence” — quantitative eval metrics for GenAI teams — and for moving from pure eval into runtime guardrails. Backed pre-acquisition by strategic investors (Databricks, ServiceNow, SentinelOne, Citi/Amex Ventures). Differentiates from pure observability tools by bundling evaluation + guardrails; from pure red-teamers by the production-monitoring side. Note: distinct from the unrelated “Galileo AI” UI-design tool acquired by Google.

Ownership, funding & M&A

Acquired by cisco. Galileo (Galileo Technologies, Inc. / galileo.ai; formerly Rungalileo) was founded 2021 in San Francisco by Vikram Chatterji (CEO), Atindriyo Sanyal and Yash Sheth, and raised ~$68M, including a $45M Series B (2024-10) led by Scale Venture Partners. Cisco announced intent 2026-04-09; secondary press (Network World, Channel Insider) reports the deal completed 2026-05-22, while Cisco’s own note said expected close in Q4 FY2026. Intent is high confidence; the exact close date is medium (no Cisco closing PR located). Terms undisclosed.

CTO / hedge-fund lens

Day-2 for most funds: you need this once you’re building or seriously evaluating LLM/agent applications and want quantitative quality/safety signals plus production monitoring. The evaluation output is genuinely useful SR 11-7 / model-risk evidence — documented, repeatable testing of model behavior over time. For a Splunk shop the integration is the draw; standalone, weigh it against arize-phoenix, langsmith, braintrust (eval/observability) and patronus-ai (eval). If you’re not yet building AI apps, it’s premature.

Competitors / alternatives

langsmith, arize-phoenix, langfuse, braintrust, patronus-ai, maxim-ai, lakera, promptfoo.

Open questions / to verify

• Confirm the Cisco close date with a primary (intent + secondary close found).
• Standalone product/pricing continuity post-Splunk integration.

Sources

• Cisco Blogs — Cisco Announces Intent to Acquire Galileo — fetched 2026-06-28 — supports: intent 2026-04-09, AI observability/eval/guardrails, into Splunk Observability; confidence: high
• Galileo Raises $45M Series B (PRNewswire) — fetched 2026-06-28 — supports: Series B $45M (Oct 2024), founders, ~$68M total, SF/2021; confidence: high

History

• [2026-06-28] Stub created from seed registry.
• [2026-06-28] Researched; confirmed Cisco acquisition of Galileo (intent 2026-04-09; reported closed 2026-05-22). Galileo = AI eval/observability + guardrails (galileo.ai / formerly Rungalileo), founded 2021 SF, ~$68M raised. Distinguished from Google’s unrelated “Galileo AI”. ownership=acquired, confidence medium on close date.