Moonshot AI Enters Real Financial Workflows | BlueRun Ventures

蓝驰创投蓝驰创投··0·8

Bringing Model Capabilities to Financial Scenarios

Kimi is entering complex, real-world scenarios. Today, Moonshot AI officially launched its financial industry AI solution, integrating 10+ authoritative data sources, 9 financial Skills, institutional-grade data modeling and report delivery capabilities, and security and compliance features built for real business operations — bringing AI into more complete financial workflows.

BlueRun Ventures was one of Kimi's earliest investors, and doubled down across four consecutive rounds. We're also an early-stage VC firm actively using AI and exploring how investment work itself is changing. We follow the evolution of model capabilities, use Kimi for project and industry research, and experiment with distilling investors' research methods and judgment frameworks into reusable Skills. Throughout our journey alongside Kimi, we've watched technical boundaries expand — and we've used and validated these capabilities in real work.

As technology unlocks model capabilities, AI begins solving a series of real problems: from basic information processing, to boosting research efficiency, to reorganizing and reactivating long-accumulated experience — opening up new ways of working.

Below is how Kimi brings model capabilities into real financial workflows. Enjoy reading:

Today, we're launching the Kimi Financial Industry Solution. This suite integrates Kimi's products and capabilities for the financial sector, covering 10+ authoritative data sources, 9 financial Skills, 5 compliance and security measures, plus institutional-grade data modeling and report delivery.

From morning holdings briefings and earnings commentary, to deal screening, deep research, and portfolio reviews, Kimi connects data retrieval, professional analysis, and deliverable production. In live institutional deployments, this solution has compressed document processing and first-draft production — work that used to take days — down to hours, freeing professionals to spend more time on analysis and judgment.

Dozens of financial institutions are already innovating on top of the Kimi model, Kimi Enterprise, and Kimi Work — or co-building financial solutions with Kimi — spanning leading brokerages, commercial banks, venture capital funds, and investment research data providers:

Swipe left or right to see all →

Executives from CICC, CSC Financial, and Deutsche Bank shared with Kimi where they see financial AI heading next:

Pain points and challenges in finance

Finance is one of the most data- and information-intensive industries. Its productivity has long been constrained by two factors: the speed of data processing and the breadth of information gathering.

The first challenge is the market's ever-rising demand for timeliness. Analysts draw on filings, news, and research reports. A research conclusion used to take a full day to assemble; now the market demands minute-level responses, and shifts in global market structure need to be absorbed instantly.

The second challenge is extreme information fragmentation. Key data is scattered across financial reports, legal documents, research notes, and announcements. Professionals spend enormous amounts of time querying, cross-checking, and organizing data before they can even begin analysis.

In 2026, as large models improve in reasoning and tool use, AI is moving from single-point assistance to end-to-end delivery across complete financial workflows: understanding the objective, searching professional data, running analysis and modeling, and producing institutional-grade deliverables. But finance demands exceptional accuracy and compliance. AI output must be professionally structured, clearly sourced, and easy to verify before it can truly enter business processes.

This is exactly the starting point for the design of Kimi's financial industry AI solution.

The Kimi Financial Industry Solution

❶ 10+ authoritative data sources, one-stop access to reliable real-time data

Data drives every part of finance, from investment research to risk management.

Kimi's full product line — the kimi.com web app, the Kimi Work desktop client, and Kimi Code — comes with built-in access to authoritative domestic and international data sources including Wind Information, East Money, S&P Global, Cailian Press, and Caixin Data.

This data flows directly into research tasks via MCP, and every key figure can be traced back to its original source, with provenance and definitions clearly documented. Researchers no longer need to repeatedly search, download, and compile information across market terminals, company filings, macro databases, and corporate information platforms — eliminating the work of hunting and hauling data across systems.

For company research, researchers can also tap Tianyancha and Yuandian Legal to supplement business registration, judicial, IP, and regulatory information — dimensions beyond the financial statements.

Click images to enlarge ↓

Beyond data sources, Kimi ships with a full set of research and production plugins: full-text academic paper search via Scholar, generation tools for Office formats like Word, Excel, and PDF, image and audio generation tools, and more.

In addition, the Kimi browser extension (formerly Kimi WebBridge) and Computer Use let an Agent, with proper authorization, operate browsers and local desktop applications — so systems without APIs can still join the workflow. The financial research suite covers secondary-market research, end-to-end primary-market deal workflows, and corporate financial close; the Cloudflare developer tools provide over 2,500 official API endpoints to support further customization and integration.

We're also building connections with Chinese office systems like Lark, Tencent Docs, and WPS, so research materials and analytical output flow more smoothly into team collaboration.

❷ Nine Skills that crystallize AI capabilities into professional financial workflows

For finance professionals, AI's value lies in grounding general model capabilities in specialized workflows.

Working with frontline institutional analysts, fund managers, and investment banking teams, Kimi has packaged key stages of financial work into 9 professional Skills: institutional financial modeling, institutional research reports, institutional PPT, dynamic financial charts, earnings commentary, consensus estimate mapping, portfolio reviews, morning holdings briefings, and the HK IPO Lens. Each Skill corresponds to a type of work with stable analytical steps and delivery standards:

These Skills have already been proven in real tasks across secondary markets, primary markets, banking, and insurance:

● Investment banking / private equity: financial modeling

Using the "Institutional Financial Modeling" Skill, a financial model was built for a commercial spaceflight company based on management accounts, audit reports, and due diligence materials — outputting an Excel model with adjustable assumptions and a comparable-company valuation table. The human effort required for financial modeling dropped from 5–7 person-days to 0.5–1 person-day.

● Industry research: deep research

Using the "Institutional Research Report" Skill, combined with expert interviews and earnings call materials, a deep-dive report on the quantum technology industry was completed — cutting the research cycle from 10–20 days to roughly 2 days.

● Commercial banking: credit reports

Credit investigation reports were generated from prospectuses, field research notes, and credit data, delivering clear admission conclusions, proposal recommendations, and risk flags — improving per-account preparation efficiency by more than 5x.

● Insurance: pet insurance analysis report

Using the "Institutional PPT" Skill, analysis of competing pet insurance products — market size, pricing design, and unit economics (UE) — was assembled into a 21-page investment committee deck. The human effort for PPT production fell from 7–10 person-days to about 2.

Beyond these financial Skills, Kimi offers several general-purpose capabilities:

Turn research output into interactive HTML reports: compared to static documents, HTML carries far higher information density, spreads via a single link, and lets readers drag parameters and test assumptions. The "Dynamic Report" Skill converts sell-side deep research into presentation-ready visual reports for research marketing, client communication, and public release.

Keep long-tracked information in one place: the "Dashboard" in Kimi Work is built from freely arrangeable "widgets" that can be pinned to your desktop. For example, you can build a financial terminal dashboard with the financial terminal plugin, linking professional sources, portfolio holdings, and covered companies — open the page and see the latest changes, no daily re-searching required.

Professional interactive financial charts, with direct annotation and data editing: on the institutional-grade charts produced in Kimi Work, the "circle and draw" feature enables quick markups and edits — including but not limited to CAGR and delta annotations, switching chart types (stacked bars, clustered bars, line charts, Mekko), changing color schemes, adjusting label formats, and editing data.

Custom plugins to connect any system to your workflow: with Kimi Work's Plugin Builder, describe the system you want to connect and the operating flow in natural language, and create a custom plugin via one of three paths — MCP, API, or browser skills. Your existing data subscriptions, internal materials, and business systems can all enter the Kimi Work workflow under controlled, secure permissions.

❸ Five security and compliance measures to bring Agents into real business

Finance is an industry with extreme demands for accuracy and compliance. Which information in a task can be sent to external models? Is access to data sources and tools properly authorized? Is the output verifiable? Can the execution process be audited? Only when these questions are answered can AI truly enter business processes.

Around these questions, we are working with CSC Financial to explore a path for compliant AI Agent access to financial business. Together we're building a "Risk Assessment Gateway" that translates financial institutions' security and compliance requirements into 5 concrete measures governing agent access and operation:

  1. Data classification and grading: define usage and external-transfer boundaries for data at different sensitivity levels per institutional policy; the gateway inspects tasks, attachments, and tool-returned content against the rules;
  2. Personal information protection: following the principle of minimum necessity, personal information is filtered, de-identified, or blocked per the rules, and only task-essential, authorized content is passed to the model;
  3. Authorized access to data sources and tools: authorization requirements are enforced on every call, unauthorized data queries or tool operations are intercepted, and returned content is inspected;
  4. Output verification and human review: outputs are labeled with sources, timestamps, and items pending verification, with traceable supporting evidence; preliminary materials are reviewed by business staff and do not directly constitute project decisions — business judgments still follow existing approval processes;
  5. Audit and accountability: tasks, data sources, processing decisions, and tool call records are stored in association, enabling task-level review of the entire execution process.

This "Risk Assessment Gateway" has completed pilot validation in real business. CSC Financial chose interim trustee report generation as its first validation scenario, building an "Interim Trustee Report Generation" agent on Kimi Hosted Agents and connecting it to internal business systems. The first round of validation focused on issuer announcement recognition, managed bond screening, and report draft generation — while preserving business staff's review and decision-making responsibilities.

The pilot covered the production of 60+ interim trustee management reports for over 50 issuers and more than 100 outstanding corporate bonds, delivering the following efficiency gains:

  • Integrating the "Interim Trustee Report Generation" agent with existing business systems was originally planned to take 2 months; it was actually completed in 3 working days;
  • Manual production time per report fell from about 30 minutes to 10 minutes — a 67% reduction in human effort.

More importantly, this gateway is reusable across scenarios: new use cases can reuse the same compliance-checking mechanism for integration and launch review, laying the efficiency foundation for scaling agents into more business scenarios.

Availability and deployment

The Kimi Financial Industry Solution is fully live:

  • All users can install and try the 9 financial Skills and 10+ data source plugins on kimi.com;
  • Knowledge workers and developers can call data source plugins and financial Skills in the Kimi Work desktop client and Kimi Code, and experience custom plugins and local computer Agent capabilities;
  • Kimi Hosted Agents for enterprises provide a configurable, auditable runtime environment for agents entering real business processes.

Try it now:

  • Kimi Enterprise: kimi.com/business
  • Kimi Work: kimi.com/products/kimi-work
  • Kimi Code: kimi.com/code
  • Kimi Hosted Agents: platform.kimi.com/docs/hosted-agents/quickstart
  • Kimi API Open Platform: platform.kimi.com

Kimi K3 Launch: A Sleepless Night, Witnessing Long-Termism Pay Off

Kimi K2.6 Released and Open-Sourced: Major Advances in Coding, Long-Horizon Tasks, and Agent Swarm Capabilities | BlueRun Ventures Headlines

No Titles, No Departments, No OKRs at Kimi: How 300 People Support a $120 Billion Valuation | BlueRun Ventures Headlines