Why Sanna
Sanna is building a new kind of accounting firm, designed from the ground up for a world where AI handles much of the manual work that has defined the industry for decades.
We combine accounting expertise with software and AI to deliver better service at a lower cost. Founded in 2025, we already serve more than 2,000 companies. We’re starting with holding companies in Norway, expanding into operating companies, and building for international scale from day one.
We’re intentionally building a small team of exceptional builders in Oslo and Stockholm. Engineers work directly with accountants and clients, with substantial ownership over what we build and how it performs.
What we are building
Accounting involves thousands of workflows that connect financial data, documents, rules, and conversations. We’re building an intelligence platform that gives AI the context and tools to carry out this work, with accountants reviewing results and handling exceptions.
Some of the problems you could solve:
Reconciling transactions with supporting documents and resolving discrepancies.
Detecting when an agent is wrong or lacks enough evidence to proceed.
Identifying missing information, asking clients the right questions, and continuing the work when they respond.
The opportunity goes beyond automating individual steps. A 30-step process might become three, and some workflows may disappear entirely. We build alongside our own accounting team, using real cases and reviewer feedback to improve performance and automation over time.
The Role
We’re looking for an AI Engineer to build systems that carry out accounting work reliably, from reconciling transactions to gathering missing information from clients.
You’ll own AI workflows from prototype to production: designing context and tools, building evals, and improving performance on real accounting cases. Working directly with our accountants, you’ll turn messy problems into useful systems and decide where models, deterministic code, or human review belong.
We have a high bar for engineering judgment and ownership. You’ll have substantial freedom to shape both the systems and how we build them.
What You'll Do
Build AI workflows that connect financial data, documents, and conversations with tools that take accounting actions.
Develop evals from real accounting cases, investigate failures, and improve accuracy, latency, and cost.
Build approval- and human review flows so workflows are dependable and decisions can be inspected.
Work directly with accountants to simplify workflows and measure whether your systems increase automation degree.
What We're Looking For
Experience shipping LLM-based systems to production, with a practical understanding of context, tool use, evals, and failure modes.
Strong software engineering foundations across backend services, APIs, integrations, and data, with the ability to build interfaces when needed.
Judgment about where to use AI, where deterministic code works better, and when human review is necessary.
The ability to take an ambiguous problem through investigation, experimentation, and into production.
Active use of AI in your own engineering work, with responsibility for understanding and verifying the output.
You don’t need an accounting background, but you should be curious about the domain and care about getting the work right.
How we work
Engineers work closely with accounting, design, and clients. We prototype quickly, ship continuously, and use real cases to decide what to improve.
You’ll own problems rather than work from detailed specifications. We expect you to challenge assumptions, including whether a feature or workflow needs to exist at all, and follow through until the solution works in practice.
We’re scrappy when learning and deliberate about correctness, security, and reliability when people depend on what we build.
Our tech stack
Frontend
React
AI
Mastra, AI gateway
Workflows
Inngest
Data & Backend
Supabase
Cloud
Vercel
We're always exploring tools that help us build better products and move faster.
