Cast Insights raises $4.5M for real-time speech monitoring

Cast Insights has raised $4.5 million in pre-seed funding and launched a platform for tracking public speech across television, radio and podcasts in real time. The San Francisco startup is pitching the product as competitive and risk intelligence for institutions that want earlier signals from broadcast and spoken media, a category where the company says much of the data disappears before it is indexed.

Abstract Ventures led the round, with participation from HF0, Village Global, Max Ventures, Embassy Ventures and Stratus Ventures. Cast did not disclose its valuation, revenue, customer count or headcount.

The company emerged from HF0 Residency, a program for repeat founders. Co-founder and Chief Executive Otávio Costa Miranda argues that public speech is an underused data source for investors, policy teams and other institutional users because statements on local broadcasts, legislative proceedings, radio and podcasts are difficult to capture at scale. That claim is directionally plausible, though Cast has not said how it handles rights, retention or coverage gaps across jurisdictions.

What the platform does

Cast says its system continuously monitors thousands of global sources, ingests audio, transcribes speech and uses AI models to identify speakers, detect narrative formation and measure sentiment changes as they happen. The startup describes the target dataset as “ephemeral speech,” meaning public statements that may influence opinion or markets but are not typically available in structured feeds.

The pitch sits between media monitoring, alternative data and geopolitical risk tools. The AI framing is conventional for 2026, but the operational question is whether Cast can deliver timely, accurate speaker attribution and signal extraction at broadcast scale. The company says its platform can generate alerts within 30 seconds of broadcast and track speakers and narratives across thousands of sources, according to HF0’s Evan Stites-Clayton.

Cast says it has already processed more than 2.3 million hours of speech across 20 countries and 10 languages while working with early adopters. With the new capital, the company plans to expand to 20 million hours processed and monitor 500,000 hours of live content per day. It claims that would create the world’s largest dataset of this type, covering major U.S. television broadcasts, radio stations and podcasts. The company did not provide third-party verification for that ranking.

Use case: Strait of Hormuz

Cast pointed to a case study on public discussion around the Strait of Hormuz during the U.S.-Iran conflict. According to Miranda, mentions were close to zero before the conflict began, then rose to about 400 per day and later peaked at 2,946 after a recent ceasefire collapsed.

The company said its system showed positive sentiment falling from 7.7% to 3.3% after the ceasefire broke down, while also tracking related market effects such as higher war-risk insurance premiums for shipping companies. Cast also said it captured shifts in the discussion after U.S. officials proposed a 20% transit toll on ships passing through the Strait and then reversed the policy less than 24 hours later.

For founders and investors, the financing is small but notable because it points to continued demand for AI products built around proprietary data pipelines rather than generic copilots. Cast’s next test is whether institutions will pay for spoken-media intelligence as a durable workflow, rather than treating it as another dashboard in an already crowded risk stack.

This story draws on original reporting from SiliconANGLE.