Blog · Grow Amplified GmbH
What AI can do in office work today, and what it cannot.
Practical knowledge for fiduciaries, property managers, SMEs and financial institutions: how agents read and reconcile documents, where approval stays with you, and how to start with a single process. Examples from projects are always anonymised.
15 · All articles, newest first
RSS feedWhat 15 years in banking and consulting teach about AI
Four stations in banking and consulting and what they teach about AI with approval: reconciliation, checklists, replay logs, measured hours. By Timon Wick.
Read the articleFour-eyes principle with AI agents: liability and approval
If AI works on a client engagement, you remain liable. How to build an approval step that proves due care: six building blocks, one test, three mistakes.
Read the articleHourly rate or flat fee when AI saves time
Billing by the hour, faster work means lower fees. The fee models fiduciaries can use, and why measuring your own hours comes before any price change.
Read the articleShoebox receipts: what one booking really costs
What a booking from the shoebox costs in your own hours, whether paper may go after scanning, and what an agent takes over when receipts come in.
Read the articleReplay log and GeBüV: AI work on record for ten years
What Art. 958f CO and the Swiss GeBüV mean for AI-assisted bookkeeping, what a replay log must record, and how to test whether yours holds up a year later.
Read the articleAI in Bexio and Abacus: the review layer matters more
Bexio and Abacus increasingly post entries with AI. What stays with people, how a review layer works without new software, and what to ask your vendor.
Read the articleFlat fee or extra work: how to find the leak
Why extra work disappears into flat fees, and how to check one client: contract beside the time-tracking export, every line marked F, E or B.
Read the articleCount the tenant inbox before you advertise a job
One working day, two numbers: find out whether the hours in your tenant inbox sit in the volume, in a queue before approval or in boomerangs.
Read the articleFlagged is not billable: what your contract allows
An analysis flags extra work. You may only bill what the property management contract provides for: a three-column grid and one decision per item.
Read the articleClient data and AI: professional secrecy and the FADP
Swiss data protection law applies to AI, and its duty of confidentiality reaches further than the Criminal Code. What that means and what to get in writing.
Read the articleMaking KYC files review-ready: read everything, not a sample
How to get many KYC files review-ready in a short time: what review-ready means, a six-step process with approval, and where the decision stays.
Read the articleAI policy for SMEs: spotting and managing shadow AI
Shadow AI in small Swiss firms: a ZHAW thesis survey of 72 people, three risks in plain words, a one-page AI policy template and a 30-day rollout plan.
Read the articleFINMA Guidance 08/2024 explained simply: seven checkpoints
What FINMA looks at when institutions use AI: the seven checkpoints of Guidance 08/2024 in plain language, what they mean in practice, and a checklist.
Read the articleEU AI Act after the Omnibus: impact on Swiss SMEs
When the EU AI Act reaches Swiss firms, which deadlines apply after the AI Omnibus, what Switzerland plans and four steps for SMEs. As of November 2026.
Read the articleAI in the fiduciary office, explained simply
What AI can already do in a fiduciary and accounting office, what it cannot, and where responsibility stays: five terms, six tasks and a safe first step.
Read the articleFirst call
Talk directly to the person who leads the engagement.
25 minutes by video, at a time you choose.