On May 5, 2026, Anthropic and Perplexity simultaneously launched their offerings for institutional finance. Ten configurable agents on one side, 35 ready-to-use workflows on the other: two opposing visions targeting the same rapidly evolving market.
Key Takeaways
- Anthropic launches 10 AI agents for institutional finance, powered by Claude Opus 4.7
- Perplexity deploys 35 financial workflows with full traceability back to SEC source documents
- Both companies directly target hedge funds, investment banks, and asset managers
Anthropic Rolls Out 10 Agents for Institutional Finance
Ten agents deployable within days. That is the offer Anthropic is extending to major finance players: JPMorgan, Goldman Sachs, Citi. These templates cover the most time-consuming tasks in the sector, from building pitchbooks to KYC screening, accounting reconciliation, and monthly closing.
Each agent comes with its own business instructions, data connectors, and specialized sub-agents. Teams can plug them directly into Claude Cowork or Claude Code, or run them in autonomous mode via Claude Managed Agents for overnight processing or entire portfolio workflows.
The data ecosystem behind it is substantial. Anthropic draws on FactSet, S&P Capital IQ, MSCI, PitchBook, Morningstar, and LSEG, now joined by Dun & Bradstreet, Verisk, Third Bridge, and Moody’s. The latter integrates its credit ratings covering 600 million companies directly into the Claude interface.
The entire stack runs on Claude Opus 4.7, which scores 64.37% on the Vals AI Finance Agent benchmark. This benchmark does not measure theoretical capabilities: it evaluates an agent’s ability to complete real financial tasks end to end, under near-field conditions. A meaningful score for a sector as demanding on output reliability as institutional finance.
Anthropic is not selling an improved financial assistant. It is selling a custom agentic infrastructure, built to absorb the volumes and constraints of institutional back-offices. The difference is structural: this is no longer a tool, it is an integrated automation system.
Perplexity Bets on Full Traceability
On the other side, Perplexity directly targets hedge funds, private equity, and asset managers with its “Computer for Professional Finance.” The product includes 35 dedicated workflows, designed around the specific needs of institutional finance.
Its main strength is transparent integration. The system connects directly to licensed data terminals via MCP connectors: Morningstar, PitchBook, Daloopa, and 14 additional integrated providers. Analysts work from their existing environment without any workflow disruption. The tool runs natively in Microsoft Teams and soon in Excel.
Perplexity’s central argument is full traceability. Every number the system produces is clickable and links back to the exact line of the original SEC document, with the intermediate calculation displayed. A promise calibrated for finance, where an approximation in an investment memo or a due diligence report can cost millions.
A single query is enough to generate a complete summary sheet or an interactive sourcing screen. Perplexity prioritizes fast adoption: no complex configuration, no infrastructure to deploy. The analyst submits a query from their usual environment and gets a usable deliverable immediately.
Where Anthropic builds a layer of configurable agents for deep back-office processing, Perplexity positions itself as the ready-to-use deliverable production engine. Two answers to two different problems within the same financial value chain.
Also on Horizon:
- Google Is Building Remy, a Personal AI Agent 24/7
- Anthropic x SpaceX: Claude Code doubles its limits
- Google I/O 2026: Gemini is getting a new look
Finance: The New Battleground for Agentic AI
The simultaneous announcements are not a coincidence. They reflect a shared perception across the industry: professional finance is the sector where AI can demonstrate immediate and measurable economic value. Jamie Dimon, CEO of JPMorgan, points to a profound transformation of work driven by these tools.
Anthropic and Perplexity are not occupying complementary niches. They are competing for exactly the same clients: the same investment funds, the same investment banks, the same asset management teams. As we recently covered, Anthropic’s revenue growth is being driven by the adoption of Claude in professional environments. Institutional finance represents the natural next frontier of that expansion.
The two technological philosophies diverge sharply. Anthropic sells an army of configurable agents built for mass processing and deep back-office work. Perplexity bets on immediate integration and data-level traceability. Two different answers to the same question: how do you convince ultra-conservative institutions to delegate decisions to an AI model?
Finance has officially become the real-world laboratory for agentic AI. The stakes are no longer purely technological. They are psychological and regulatory: how far will an analyst or a portfolio manager be willing to let an AI model engage their professional responsibility?
Follow the story on Horizon.
Pingback: ChatGPT Finance: Connect Your Bank Accounts to AI - Horizon IA