§ About · the team

Built by an ESG banker and an engineer.

Two founders. Complementary by design. One spent his banking career rating ESG performance of real assets. The other built production AI for safety-critical environments. Both on every design-partner call.

§ Charles Beck

Co-founder · CEO & sales

Charles spent his banking career on the ESG side of project finance at ODDO BHF and SaarLB. In his last role he rated the ESG performance of roughly 200 wind farms for financing decisions — which means he's read more ESG due-diligence documents than most CSRD consultants will see in a decade. He knows what auditors actually look for, and where consultants typically cut corners.

That credibility is why Attera leads with a grounding verifier rather than a "trust the model" pitch. Every figure on a Charles-signed report has to survive the same scrutiny he applied to 200 wind farms.

ex-ODDO BHF & SaarLB · ESG specialist, project finance
Master's Data & Finance — Albert School · Mines Paris
Languages: FR · DE · EN
LinkedIn →

§ Zeff Cresens

Co-founder · CTO & product

Zeff built production AI for high-stakes environments before Attera. At Xenon54 he shipped AI-safety tooling deployed on construction sites — software where false positives cost real money and false negatives can cost lives. That experience set the bar for how Attera handles uncertainty: the verifier refuses to fabricate rather than guess, because in compliance, fabricating a number is the same class of failure as missing a hazard on a job site.

The technical wedge — local LLM, page-level grounding, append-only audit log with hash chain — is built on the same principles.

Built AI safety tooling for construction sites (Xenon54)
Master's Data & Finance — Albert School · Mines Paris
Languages: EN · NL · FR
LinkedIn →

§ Why we're building Attera

An ESG analyst spends hours hunting figures across PDFs. That work should not be the bottleneck of European compliance.

Charles watched companies pay €30–80k for CSRD engagements where the most expensive line item was a senior consultant typing numbers out of PDFs into Word documents. Zeff watched generative AI launch into compliance use cases where the model would happily invent a plausible figure if it didn't find one. The first problem is a waste. The second is a liability.

Attera solves both by reversing the AI's default: we make it refuse rather than fabricate. Every figure in a report has to be traceable to the document and page it came from, or it doesn't ship. That single discipline — combined with running the model on hardware we own, in the EU — is what makes the output auditor-defensible instead of suspicious.

§ The Attera Label · vision

One label. One methodology. Across every client we serve.

The Attera Label is our north star: a single ESG measurement standard, applied identically across every customer. Same scopes. Same emission factors. Same approval chain. Same audit trail.

Today, a pension fund holding 80 portfolio companies receives 80 different ESG calculations — each consultant uses their own methodology, their own factors, their own boundaries. The numbers can't be aggregated. Comparing this year to last year inside one company is hard. Comparing across the portfolio is impossible.

We're building the Label so that PE funds, pension funds, and large groups get apples-to-apples ESG figures across their entire portfolio out of the box. Not because each portfolio company changes how they work — because Attera quietly enforces the same calculation behind the scenes.

unified scopes unified factors unified workflow unified audit trail roadmap · Label as certified mark

The Attera Label — Bronze, Silver, Gold tiers

The Attera Label · roadmap

portfolio · illustrative fund · 80 companies

without unified methodology80 inconsistent reports

cross-portfolio comparabilityimpossible

year-on-year drift inside one companyhard to trace

with the Attera Labelsame methodology · n=80

aggregate Scope 1 + 2 + 3one query

year-on-year trajectoryapples-to-apples

label tier per clientbronze · silver · gold

§ Where we are

Belgium-built. EU-hosted. Designed for the post-Omnibus wave.

Built in Leuven. Hosted exclusively in the European Union (Belgium primary + Hetzner Falkenstein/Helsinki). Inference runs on hardware we physically operate — no third-party LLM API ever sees your documents. The full sub-processor list is on the security page, two entries long.

Want to meet us?

Both founders are on every design-partner call. Bring a sample document — we'll generate a draft section live.

Book a demo →