No exposure to activities your exclusion policy rules out. Scope 1 and 2 intensity is above the sector median. One open controversy and two recent news items need a view before Thursday's IC.
Today, one ESG question
takes six tabs.
Next, it takes
one sentence.
Two holdings account for most of the change. Holding B's 2025 report replaced a 2023 estimate, and Holding D's position grew 40%. Same method as your periodic report, PAI 3.
Today, pre-deal due diligence
is a week of chasing.
Next, it's one request.
Today, quarter-end means
an inbox full of asks.
Next, one message
instead of an inbox.
Good morning. I've drafted the SFDR periodic report disclosures for your Article 8 cohort, all 12 funds, with multi-fund reporting. 10 are ready to sign off. Fund D and Fund J need you first.
ESG work is changing shape.
The same data and calculations, with a different way of working around them.
What makes an agent an ESG expert
Everything an agent needs to work on ESG, maintained by us and kept current as the regulation moves.
- 01 / 06 · foundation
Calculations that run in the platform
The agent calls our calculations rather than working figures out itself, so a PAI indicator is the same in a chat as it is in your report.
- 02 / 06
Our methodology and regulatory knowledge
How every metric is calculated, which sources it uses, and what the regulation asks for.
- 03 / 06
Your firm's context
Your exclusion criteria, sustainable investment methodology and report templates, so answers follow your policy rather than a generic one.
- 04 / 06
Skills written by specialists
Step-by-step instructions for ESG tasks, written and tested by our data and regulatory team. A skill is plain text, so over time your team will be able to add its own.
- 05 / 06
Tools it can call
An MCP connector and agent-ready APIs that let an agent look up funds, holdings, indicators and controversies in your Connect Earth data.
- 06 / 06 · the interface
Your agent, or the Connect Earth coworker
Claude, or any agent that supports MCP, sits on top. Models will keep changing, whilst the harness underneath keeps your answers consistent.
- How it fits together
A question goes in at the top and a cited answer comes back, with every layer in between maintained by us.
Bring your own agent, or use ours
Your own agent
Connect Claude, or another agent that supports MCP, using your own Connect Earth login. Your team keeps working where it already works.
- Works inside the agent your firm already uses
- Same data and permissions as your platform account
- Our skills and context, not a generic prompt
harness
Same data, calculations and permissions either way
The Connect Earth coworker
The coworker built into the platform, for teams that don't run their own agent. Ask a question next to the data it's about.
- Nothing to set up
- Answers alongside the screen you're on
- Uploads and documents handled in the platform
What Claude, ChatGPT or Copilot can do with your ESG data
Early access skills today, with more on the way. Each one written and tested by our data and regulatory team.
How the answers hold up
Fund A's GHG intensity rose 0.8 this quarter, calculated as PAI 3 in the platform.01 Two holdings account for most of it: Holding B's 2025 report replaced a 2023 estimate02, and Holding D's position grew 40%. Neither holding is flagged under your exclusion policy.03 Scope 3 data for Holding F isn't available yet, so it is left out rather than estimated.04 This describes what the data shows. It is not a rating.05
- 01Calculations stay in the platformThe agent calls our calculations and cites them. It never works out a PAI indicator from memory.
- 02Every answer cites its sourceEach figure links back to the company report or data point it came from.
- 03Your policy is the contextAnswers follow your exclusion criteria and methodology, not a generic definition of sustainable.
- 04It says when it doesn't knowIf the data isn't there, the agent tells you rather than filling the gap.
- 05Facts, not opinionsIt describes what the data shows. It doesn't issue ratings of its own.
- 06Tested before it changesSkills and knowledge are versioned and checked against example questions before any change goes live.
Your agent is one file away.
Connect Claude, ChatGPT, Microsoft 365 Copilot or any agent that supports the Model Context Protocol (MCP), with your own Connect Earth login. Same data and permissions as your platform account.
Models are fast becoming a commodity. The ESG knowledge and context around them is the hard part, and that's the part we build.
An LLM brings the reasoning and an enormous breadth of general knowledge. What it typically lacks is the specificity: your policies, the calculations behind each metric, and a view of regulation that stays current as SFDR, the EU Taxonomy and the rules around them change. We look after that part, so you can outsource the work of staying up to speed with regulation and rest assured that what your agent produces lines up with it.
Questions about using AI agents for ESG work
How do I get Claude, or Claude Cowork, to work with my ESG platform?
Can Claude or ChatGPT create my SFDR periodic report?
Can I use ChatGPT, Microsoft Copilot or Gemini with my ESG data?
What is an MCP connector for ESG data?
Can an AI agent run ESG due diligence before a deal?
Will the figures from my agent match my regulatory reports?
Is it live?
Can the agent change our data?
Does it give opinions on companies?
Make your agent an ESG expert.
Early access is opening to existing customers. Tell us which agent your team uses and we'll show you what it can do with your data.


