AlphaSense rolls out chain-of-thought, tool use and coding features in latest AI update

AlphaSense is layering chain-of-thought reasoning, autonomous tool use and coding capabilities into its market intelligence platform, according to commentary from Charlie Zvibleman shared on X by RagingVentures. The update points to a heavier test-time compute profile and a wider rollout of coding features beyond the platform's technical users.

The commentary, sourced from Zvibleman and surfaced by RagingVentures on X, frames the changes as a step toward more autonomous research workflows on the platform. Chain-of-thought and test-time compute approaches let a model spend more inference cycles on a single query, trading higher per-query token spend for stronger reasoning on harder questions. The auto Think Longer behaviour, as described, hands that decision to the system rather than the user.

Tool use sits alongside that reasoning layer. In the agentic pattern, the model calls external tools (search, retrieval, code execution, structured data lookups) inside a single session rather than returning a one-shot answer. Pairing tool use with auto Think Longer suggests AlphaSense is letting the system route between fast lookups and longer reasoning passes depending on query complexity.

The coding capabilities, per the same source, are being opened to a wider set of users on the platform. AlphaSense has not, in the cited material, specified which user segments or which coding interfaces are in scope.

What to watch over the next two quarters: pricing disclosures on any usage-based component tied to extended reasoning, latency benchmarks from buy-side users putting the auto Think Longer behaviour through real workflows, and any compliance documentation AlphaSense publishes around tool-use audit trails. The agentic direction is the right one for the category. Execution will show up in token bills and review cycles.