Mosaic Conseil is a small, fully remote workshop dedicated to rebuilding one business deliverable at a time. Whether it’s a three-year plan, a pricing grid, or a diligence file, we disassemble the work at the bench and rebuild it, line by line, against what the invoice ledger proves. The approach is personal and precise. We don’t take ongoing retainers and we don’t deliver slide decks. Instead, we return the same deliverable you sent—hand-redrawn—and backed by a written, reasoned argument. Our files are prepared on paper and supported by a phone number available for ninety days. We intentionally keep the workshop small so that the people doing the work also answer their own calls. There are no unscheduled meetings and no corporate performance. We focus on four areas: financial services, industrial networks, growing software companies, and healthcare systems.

About Mosaic Conseil We’re a boutique consultancy that prizes intellectual honesty and the discipline of a small, focused team. By staying small and working entirely remotely, we remain independent and anchored to the evidence at the heart of the business. Clients come to us when they need a business plan or a financial model rebuilt with extreme accuracy. We provide a workspace where the main objective is to get the numbers and the logic right—not to make things look good in a presentation. Every file we send out carries two names on the cover: one person who builds it and one person who challenges what was built. This ensures that every assumption is tested before a file is finalized. The environment is designed for people who prefer written argument over status meetings.

What you would do As an AI Testing Assistant, you’ll work at the intersection of business logic and automated analysis. You’ll test the outputs of AI models used to process financial and industrial data. This is an entry-level role in which you support our consultants by stress-testing the deliverables we build. Day to day, you will:

• Check AI-generated summaries and data extracts for accuracy against original sources such as ledgers or contracts.
• Run standardized tests on our internal AI tools to find where they misunderstand complex financial logic or industrial constraints.
• Create simple test cases to confirm that pricing models or business plans stay consistent when variables change.
• Write clear reports documenting specific cases where a model fails or produces hallucinations.
• Support the challenger side of our builder-and-challenger process by providing evidence where automated calculations diverged from the truth.
• Compare outputs from different AI models to determine which one handles specific healthcare or industrial terminology more reliably.
• Verify that the data used for a deliverable was correctly mapped by our automated scripts.

Helpful background We’re looking for people who naturally want to understand how technology works—and especially how it breaks. The role is suitable for those starting their career or seeking part-time work while finishing their studies. Helpful skills and experience include:

• Basic familiarity with Large Language Models and the ability to prompt them for technical tasks.
• Competence using spreadsheets such as Excel or Google Sheets to verify calculations.
• Strong writing skills and the ability to describe a technical mistake in clear, plain English.
• A background in business, finance, or computer science is useful, but we value the ability to follow a logical argument more than any specific degree.
• Comfort working with remote tools and a disciplined approach to managing your own schedule.
• Experience with personal projects or academic work involving data analysis or AI testing.
• Interest in the sectors we serve, particularly financial services and industrial networks.

How we work Our workshop is completely remote, so you can work from wherever you’re most productive. We don’t have an office and we only hold meetings when they are strictly necessary. Communication is primarily through written documents and internal notes. With our builder-and-challenger approach, your role as a tester is central to our reputation. You’re not just support staff—you’re a key part of the verification process that lets us stand behind every deliverable we produce. We value deep work and give you the quiet you need to spot errors others miss. There’s no requirement to live in a specific country or city; we operate a global, decentralized workshop.

What the role offers This is a part-time, contract position fully remote. You’ll have the chance to learn how high-level business deliverables are constructed and how AI can be integrated into professional services in a dependable way. The work is intellectually demanding and avoids the usual overhead of corporate life. You’ll work in an environment focused on technical capability and analytical rigor. The position is paid, with compensation that’s attractive for the role and the market, reflecting the care and attention to detail we expect from our testers. You’ll collaborate directly with experienced professionals who answer their own phones and audit their own work, giving you direct insight into how boutique consulting works in practice.
Precision in our workshop isn’t optional—it’s required. We believe a business deliverable is only as strong as the evidence behind it, and as an AI Testing Assistant, you’re the one who ensures that evidence holds up. By finding weaknesses in the model before the client sees the work, you help us maintain a standard of intellectual honesty that’s rare in modern consulting. Our process is intentionally slow, so we can provide the depth and thoroughness that faster, larger firms can’t match.

Upgrade to Premium to see salary and compensation for all jobs.

Location
Stockholm, Stockholm, Stockholms län, Sweden