I help founder teams work through complex operating decisions and put them into practice.

A company can run with one person per function, but only if the function is designed for it. That means rethinking processes from first principles and redesigning them AI-natively, instead of plugging AI into already broken systems. That’s what I do: deciding what still needs human judgment, and building the rest so it runs on its own. Banking, Big4 audit and a capital markets law degree are why I know where the details matter. My edge is seeing how strategy, finance, legal and compliance, revenue, people and business operations fit together, and being able to work across all of them.

See what I’ve built Read my writing

Since 2019 VC- and PE-backed companies Pre-seed → Series B startups and scaleups EU + U.S. cross-border operating scope Chief of Staff → Director Ops COO scope · hands-on build

What I’ve built

These are five of the recurring problems I got tired of solving by hand. I built something the team could keep using instead – sometimes with an agent, sometimes with an API connection, a scheduled workflow or a small browser tool.

SourcesBanks · accounting · CRM KPI engineNormalize · calculate · refresh Operating viewKPIs · cohorts · decisions

The numbers were spread across too many systems.

90 min → 15 min month-end reporting

Seeing cash, burn, revenue and pipeline in one place meant exporting bank data, accounting reports and CRM records, then stitching spreadsheets by hand. I built a Python cockpit that brings those sources together through APIs and scheduled imports, normalizes data across two entities and currencies, and calculates the KPIs a founder or a Head of Finance actually uses: cash, burn, runway, ARR, retention, pipeline and unit economics. It also shows the data underneath a KPI, not just the headline number – for example, when overall retention looks fine and one customer cohort is not. It refreshes daily, and month-end reporting takes 15 minutes instead of 90.

Python · REST APIs · Streamlit

ContextNotion · Slack · finance Claude + editorSynthesize · lock · review Investor emailLint · render · export

The investor update kept starting from a blank page.

~3 h → 10 min of my input per update

The source material already existed: weekly updates from product, engineering, GTM and ops in Notion and Slack, plus current financial reporting. I built a browser editor that turns a structured intake into a Claude-generated draft, keeps the recurring sections and company tone consistent, and renders the result as email-safe, newsletter-style HTML. The editor supports inline review and blocks export if the anonymization check catches a customer name. The update now takes ten minutes of my input and a final read from the CEO, instead of an hour of gathering and two more writing.

TypeScript · Claude API · Netlify

Attio CRMEnd dates · status Scheduled jobQuery · route Slack actionsAE renews · FDE deprovisions

Renewal dates should not live in someone’s memory.

90 days of renewal lead time, from zero

Contract end dates and account status already lived in Attio, our CRM, but the next step still depended on someone remembering to check, so a renewal often surfaced on the date itself, or worse, after it. I built a scheduled workflow that queries the CRM, browses Slack and Notion for customer context, and posts upcoming renewals to Slack 90 days ahead, tagging the account executive to start the conversation and assess expansion potential. A separate alert routes churned or non-renewed accounts to the forward-deployed engineer, so the workspace gets deprovisioned and access cleanup runs from the same source of truth.

Attio API · Scheduled job · Slack routing

Hiring inputsRole · salary · options Offer logicCap-table snapshot · 409A Candidate PDFLive preview · print

Creating an offer letter with equity took too many manual steps.

30 min → 5 min per offer

Drafting an offer meant combining role details, several salary and equity packages, cap-table data and a branded document. I built a self-contained browser tool and placed it inside the Notion hiring and onboarding guide. A hiring manager can configure different comp packages. The tool uses the approved fully diluted share count and 409A values, embedded as read-only inputs, to calculate the ownership percentage and model equity scenarios, update the branded offer live, and print it to PDF. An offer takes five minutes instead of thirty.

Browser tool · Equity calculations · PDF export

Customer redlineNew terms Clause libraryStandard · fallback Decision pathAnswer · escalate · sign

Customer contracts kept reopening the same questions.

days → under 1 h to draft

The same customer legal questions came back in slightly different forms, especially across Germany and the U.S. I built one contracting playbook with standard positions and fallback language that survives customer legal on both sides of the Atlantic. Drafting went from days to under an hour, including the special clauses and the commercial setup, and the recurring questions are answered in a tenth of the time because almost all of them have been answered before.

Legal ops · Clause library · Escalation rules

Notes from the AI Operating Layer

I write about compliance, agentic operations, cross-border decisions and the financial questions behind them.

All published notes

02 Published · Jul 2026 Finance and business models

Why people reach for a cap on AI spend

Companies track what AI costs to serve customers. The spend from the team using AI to build the product usually disappears into a general expense line. When finance can see the number rising but can’t explain why, the next move is often a spend cap.

Read the piece

Why I see it this way

I started with a full banking apprenticeship at TARGOBANK, formerly Citibank’s German consumer-banking business, and graduated in the top 5%. I later helped build the bank’s first end-to-end online car-loan application. Instead of visiting a branch, applicants could complete the application online and receive a credit decision within minutes.

At Deloitte I supported audits of BaFin-supervised banks and insurers. At Productsup I ran finance projects as the company grew from 90 to 450 people, cut DSO by half and led the finance side of a roughly 50-person post-merger integration. I completed an LL.B in banking and capital-markets law while working at Deloitte and Productsup.

In 2022 I joined a VC-backed AI startup as its first finance hire, moved into Chief of Staff and now run operations across the European and U.S. entities. The job spans finance, legal and compliance, revenue, people and business operations. The agentic tooling came later. I taught myself by building – first rough prototypes, then internal tools I can prototype, test and ship myself. I’m curious by nature and love a steep learning curve; this one has been especially fun.

2011–17: Banking and digital lending
2017–19: Financial-services audit
2019–22: Scale-up finance
2022–now: AI-startup operations

The conversations I’m useful for

Founder teams, operators and investors

Entity setup, where to hire, operating architecture, compliance right-sizing, finance and AI-native internal systems. Most often at AI companies and in fintech. I’m often the person who keeps finance, legal, product, GTM and people connected. I speak enough of each function’s language to understand what each team needs, translate between them and turn strategy into work that gets done.

Editors and hosts

Background, commentary and contributed pieces on agentic operations, compliance architecture, cross-border company-building and AI economics.

Let’s talk

If you’re working through one of these questions, send me a note. For sparring with founder teams, publication and media inquiries, or speaking invitations: hi@lenaschaefer.me

LinkedIn Substack GitHub