Purpose-built agents for climate-focused venture firms.
Claude and ChatGPT become climate-competent when Climatalist AI runs behind them: a climate knowledge graph, your firm’s own graph, and agents that traverse both.
Runs inside Claude & GPT/Live in weeks, not quarters/Every claim cited/Discovery at our cost, no obligation
Confidently wrong on the facts that decide deals.
Screening, diligence and market reads still run on general-purpose AI. Right on the easy things, wrong on the ones that matter: the incentive that just changed, the comparable round that actually closed, whether the technology is even real.
Good enough to draft. Not good enough to act on.
Your process. Your knowledge. Your tools.
We translate how your firm works into agentic workflows, give them the context including your firms knowledge and our knowledge corpus, and wire in the tools they need to finish the job.
Your workflows, encoded.
Your thesis, your rubric, your process, captured as agent workflows rather than prompts someone retypes each time.
Two context graphs behind every answer.
The Climatalist graph for the climate economy, plus a second graph of your firm's own knowledge: deals, memos, notes and data rooms, all built and kept current by the agents.
Connectors that do the work.
Drive, Slack, Notion, calendars, data rooms, your CRM. The agents read, write, and land finished output where your team already works.
How it fits together

One connection. Everything behind it.
Your firm hooks up the Climatalist MCP once. Behind it we keep both graphs current, the agent workflows improving, and the connectors live, so the stack gets better without your team touching anything.
Six agents for the work you do every day.
A day-one stack, not the final list. Each one tuned to your firm — and we build more as your workflows surface them.
Deal Screener
A one-page screen of a single deal against your rubric: a composite score, a take-the-call or pass recommendation, the questions for the founder call, and the red flags.
Inbound Monitor
Batch triage of every deck that landed overnight, on the same rubric. You open Slack and know which three of last night's twenty are worth twenty minutes.
Memo Drafter
The IC memo for a deal past screening: full depth, with precedents and a comps table, written in the deal lead's own voice from your call transcripts and diligence.
Comparables & Benchmarks
Peer rounds, valuations, terms, and precedents for a target: the climate-specific comp set the generic databases don't cover well.
Sector Brief
An accurate, current state-of-market for a sector and geography, on demand, including who raised what, recently.
Portfolio Monitor
Parses founder updates into structured KPIs (ARR, runway, burn, headcount, logos, asks) onto each portfolio page, and flags what's material.
Together: fewer missed deals, less overhead per deal, and a risk-return view you can defend in front of your IC.
Custom agents extend the stack to your firm’s workflows.
If you can't cite it, you can't act on it.
The success of sector specific agents requires context that can be trusted, which is exactly what generalized AI models are missing.
Which legislation could move global feedstock supply for SAF?
“The IRA is probably the single most powerful lever right now — the 40B SAF tax credit (up to $1.75/gallon) and 45Z clean fuels credit are pulling enormous investment into feedstock aggregation infrastructure, particularly for corn ethanol-to-SAF, waste fats, and lignocellulosic pathways…”
Identifies the right levers. Misreads several of the settings.
Which legislation could move global feedstock supply for SAF?
“From 1 January 2026, 45Z feedstocks must be grown or produced in the US, Mexico, or Canada[1] — a rule that reshuffles global trade flows and advantages domestic ATJ pathways. The UK’s 35% HEFA cap[2] and the EU’s e-fuel sub-mandate (1.2% by 2030, 35% by 2050)[3] signal the same shift: waste-residue and Power-to-Liquid are the only structurally uncapped pathways.”
An agent you’d stake a capital decision on.
Every claim traces back to a cited entity, connected to the others through climate-specific taxonomies.
From a no-obligation discovery to a stack that keeps improving.
We agree what “working” means before you're on the hook for anything. Then we build it, connect it, and keep it sharp.
- 00
Discovery · two calls
At our costA 2-hour deep dive on your firm, then a 1–2 hour workflow review. You leave with a recommended stack and a proposed tier.
- 01
Roadmap
Within daysWe agree, up front, on what “working” looks like: the measurable bar the stack has to clear before you're on the hook for anything ongoing.
- 02
Your firm's Graph
We stand up your second graph: prior deals, memos, thesis notes, and data rooms, structured so agents can traverse it.
- 03
Configuration
Agent workflows tuned to your thesis and rubric; connectors, scoping, secrets, security, and model routing.
- 04
Live
In weeksCore is running inside the assistant your team already opens.
- 05
Refinement
Months of real use. We analyse and tune the agents, and add workflows as they surface under your subscription.
Proceed and the discovery cost comes off your subscription. No commitment beyond the discovery until you've seen what “working” looks like for your firm.
A subscription that grows with your firm.
Everything runs on a 12-month subscription — both graphs maintained, agents improving, and new frontier models as they ship. No build-and-walk-away.
Core
The standard agent stack, your firm's Graph, and the Climatalist MCP — maintained and current.
Core + Custom
Core, plus 2–3 custom agents built to your specific deal workflows.
Fully Managed
Everything in Core + Custom, run and improved by us end-to-end.
Proceed and the discovery cost comes off your subscription. No commitment beyond the discovery until you've seen what “working” looks like for your firm.
See it on your own deal flow.
Start with discovery: two calls, run at our cost with no obligation. A working session on your thesis, your rubric and the decisions you’re trying to support, then a walk-through of the workflows we’d map to agents. You leave with a recommended stack, a proposed tier, and an agreed definition of “working”.
Proceed and the discovery cost comes off your subscription.