New on Orbit Insight

Build custom investment research agents in minutes.

Orbit Agent Builder is a fully automated feature inside Orbit Insight that lets institutional investment teams create their own production grade research agents through a guided process. No code & no wait. It's the perfect tool to test all the research theories you ever wanted.

Start your free trial → See how it works Contact us
What is the Orbit Agent Builder

From idea to automation,
all in one simple process.

Our Agent Builder gives you a way to describe an investment research workflow in plain language, and turn it into an agent that runs autonomously.

An agent is a piece of automation. You define what data it should look at, your universe, what metrics you are interested in, and what kind of report it should produce and how frequently. Orbit takes that definition, runs it continuously from the back end, and delivers the output on whatever cadence you set.

The work that used to take weeks of bespoke development, or stayed locked inside an analyst's head as a manual process, now takes minutes to configure and runs autonomously thereafter.

Read more →
Guided demo

Watch a build session, end to end.

Walk through the Agent Builder in a self paced tour.

How to build

Two ways to create an agent.

Describe the workflow in plain language, or hand Orbit an existing document and let the platform read it. Either path takes you to the same place: a configured agent ready to deploy.

Path 01

Build through conversation

Describe what you need in your own words. The system asks the right clarifying questions, proposes a structure, and walks you through review layer by layer before testing.

Good for
Teams starting from a blank page, exploratory builds, defining a new theme or screening criterion, prototyping a workflow.
Path 02

Or upload a document

Upload an existing methodology in any format. A fund research process, an ESG policy, a sector tracker template, a spreadsheet. Orbit reads it and proposes a baseline you can refine.

Good for
Codifying an existing framework, operationalizing a written policy, replicating an analyst's manual process at scale.
Architecture

Every agent has two layers.

Designed so the logic is transparent and the work is reusable. One agent produces one structured dataset and any number of reports drawn from it.

01

Structured data extraction

The first layer transforms your requirement into a configurable data extraction. Every field, source, and rule is visible. You can review it, edit it, and rerun it on demand.

The extraction runs continuously from the back end at whatever frequency you set. One off, scheduled, or always on until you stop it.

Configurable Transparent Continuous
02

Reports on top of the data

The second layer sits on top of the structured dataset and generates reports. One agent supports any number of reports, each defined to suit a different audience, format, or cadence.

This is where the data becomes the work product. Daily monitoring digests, quarterly thematic deep dives, ad hoc comparisons, all drawn from the same underlying dataset.

Flexible Reusable Reviewable
Who uses it

Built for the workflows that drive returns.

Three audiences, three different use patterns, one platform. Orbit Agent Builder adapts to where your team already does the work.

Asset managers

Standardize the analyst workflow

Track portfolio companies across earnings, regulatory filings, ESG metrics, themes or specific events at scale. Agents produce consistent, auditable research outputs your team can rely on.

  • Quarterly Portfolio Scorecard
  • ESG Sustainability Assessment
  • Sector Theme Digest
  • Custom Stock Screen
Hedge funds & quants

Operationalize the proprietary edge

Codify proprietary theses, extract structured signals from filings and transcripts, monitor themes across exclusive China A share coverage. Connect agents into internal models via MCP or API.

  • Signal Extractor
  • Healthcare Theme Monitor
  • Earnings Expectation Impact Analyzer
  • Portfolio Scorecard
Corporate strategy & IR

Monitor the competitive ground in real time

Track peers, regulatory developments, and market sentiment with agents that produce briefing material at the cadence your team needs. Plug competitive intelligence into existing strategy decks.

  • Peer Monitor
  • Regulatory Digest
  • Investor Sentiment Tracker
  • Ongoing Sector Analysis
MCP native

Every agent is an MCP endpoint.

No extra configuration. Every agent you build inside Orbit Insight automatically becomes an MCP endpoint, ready to connect from Claude, ChatGPT Enterprise, Copilot, or your own internal LLM tools.

01

Connect to a specific agent

Users inside their AI tool of choice can point directly at the agent they need, instead of connecting at the platform level.

02

Same data through API

The same outputs flow out through the Orbit API. Self service, same agent, two distribution channels.

03

Infrastructure for AI

Orbit becomes part of the AI tooling your team already uses. No new interface to learn, no tab to switch.

Endpoint preview
Agent esg_screen_em_oil_gas Live
Endpoint mcp.orbitfin.ai/agent/…
Frequency Daily, 06:00 UTC
Sources Orbit KB + Internal
Last run 2m ago OK
The bigger picture

Part of Orbit's multi agent platform.

Orbit Agent Builder is the first feature that turns Orbit's clients into builders on the platform. It sits inside a broader architecture designed to run agents autonomously and connect them together, so research workflows can be automated end to end.

Agents in Orbit work as building blocks across three categories. Each category plays a specific role in scaling research from a single question to an automated workflow.

Capability

Capability agents

Specialized agents that provide a single, repeatable function. The smallest unit of work on the platform.

Task

Task agents

Agents that perform a discrete research job. The broker research agent is one example. These are the building blocks teams assemble most often.

Role

Role agents

Agents that orchestrate workflows across capability and task agents to play a defined role in the research process.

Where the Agent Builder fits Our Agent Builder is the first user-facing way to define and configure new building blocks. As the platform evolves, agents will increasingly compose into longer workflows automatically.
Data sources

Pull from anywhere.

Each agent can pull from Orbit's Knowledge Bases, your connected internal data, or both. The user chooses the combination at build time, and changes it whenever the workflow evolves.

Available sources

Orbit KBCore Filings and Governance→ Orbit KBCorporate Events→ Orbit KBEarnings Intelligence→ Orbit KBSustainability and Policies→ Orbit KBRegulatory Intelligence→
Your dataInternal documents, notes, models
Your dataBroker research and third party feeds
Where it lives

Inside the Agent Marketplace.

Orbit Agent Builder lives inside Orbit's Agent Marketplace, the library of off the shelf agents already available to every user. With this launch, the Marketplace gains a new section for each user's personal library of agents they have built themselves.

Agent Marketplace
Inside Orbit Insight

Off the shelf agents

Earnings Analyzer Orbit
Risk Extractor Orbit
Peer Benchmarker Orbit
ESG Scorer Orbit
Supply Chain Analyzer Orbit

Your agents

ESG Screen, EM Oil and Gas Custom
Quarterly Guidance Tracker Custom
China A Share Theme Monitor Custom
Regulatory Change Digest Custom
+ Build a new agent
The foundation

Built on the Orbit Insight platform.

Every agent runs on the same data infrastructure Orbit has built over a decade. The award winning platform that institutional investors rely on for grounded, citation backed research.

70M
Documents
per year
75K+
Companies
covered
120
Countries
80+
Exchanges
50+
Tier 1 clients
Frequently asked

Questions teams ask before building.

What is Orbit Agent Builder?
The Agent Builder is a self service feature inside Orbit Insight that lets institutional investment teams create custom research agents through conversation. No code, no consulting engagement. Define a workflow, deploy in minutes, run continuously.
Do I need technical skills or coding knowledge?
No. The Agent Builder works through plain language conversation or by uploading an existing methodology document. The platform proposes a configuration and walks you through each layer. The work is in defining the research question, not the technology.
What kinds of investment research can I automate?
Anything that follows a repeatable structure. Earnings tracking, ESG screening, regulatory monitoring, peer benchmarking, thematic monitoring, broker research aggregation, factor signal extraction. If your team does it manually today, the Agent Builder can run it continuously.
What data sources can my agents use?
Each agent can pull from Orbit's Knowledge Bases (filings, transcripts, corporate events, ESG data, regulatory intelligence, China A share data), your connected internal data (broker research, internal notes, third party feeds), or both combined. The user selects sources at build time.
How does the Agent Builder compare to ChatGPT or generic AI tools?
Generic AI tools start from web data and broad training. Orbit agents start from an institutional grade data layer: 70 million documents per year, 75,000+ companies, 120 countries, exclusive China A share coverage, with paragraph level citations on every output. Built for production research workflows, not casual queries.
Can I connect my agent to Claude or other AI tools through MCP?
Yes. Every agent automatically becomes an MCP endpoint. Teams running Claude, ChatGPT Enterprise, or their own internal copilots can connect directly to a specific agent. The same data also flows through Orbit's API in the same self service way.
How long does it take to build and deploy a working agent?
Minutes. The build session is conversational and takes typically 10 to 30 minutes depending on workflow complexity. The agent then runs from the back end at whatever frequency you set: one off, scheduled, or always on until you stop it.
Is my agent's output reliable and auditable?
Yes. Every output is grounded in source documents with paragraph level citations. The structured data layer is configurable and transparent: every field, source, and rule is visible. You can review, edit, and rerun the logic on demand. Built for compliance friendly environments.

Stop waiting for the agent you need.

Start building today, or talk to us about getting your team onto Orbit Agent Builder. We will walk you through a live build session and show you how it slots into your existing research workflow.

Start your free trial → Talk to us