Prismatic Labs

The sensing layer for planet-aware AI

You can see your API bill. You can’t see which call, which team, or which feature is driving it – and you have no data on the energy, carbon, or water behind it. Vetch gives you all four, per call – two lines of code, never reads your prompts.

pip install vetch

The problem

AI consumes planetary resources at scale – and we have zero granular visibility into its impact.

See Vetch in action

Two lines of code. Four dimensions of AI impact.

Wrap your existing LLM calls with Vetch. Your code doesn’t change – you get energy, carbon, water, and cost tracking on every inference.

from vetch import wrap

with wrap(region="eu-west-1") as ctx:
    # your existing code, unchanged
    response = client.chat.completions.create(
        model="gpt-4o", messages=[...]
    )

ctx.event["estimated_energy_wh"]  # 0.509 Wh
ctx.event["estimated_carbon_g"]   # 0.178 gCO2e
ctx.event["estimated_water_l"]    # 0.0009 litres
ctx.event["estimated_cost_usd"]   # $0.0055
ctx.event["signal_quality"]       # "live" | "stale" | "estimated"

Open-source AI monitoring

Measure AI energy, carbon, water and cost – per call

Vetch roadmap

Start measuring today. Automate compliance next.

Beyond AI

Making the invisible measurable

AI is reshaping energy grids. Supply chains are rewriting grocery bills.
Both are happening inside systems most people never see – until they get the bill.
We build the meters.