Can AI solve everyday SME problems?
A common discovery in everyday SME work: Bexio does not offer an integrated batch receipt download, and Kontera's automatic receipt pull is only available on more expensive subscriptions. Instead of accepting higher costs or manual workarounds, the question becomes: can artificial intelligence help here?
The answer is yes — and this solution was built entirely with AI tools. Not a single line of code was written manually. Even this blog article was written with AI assistance.
Where to start with AI
Small tools often have a big impact. Two Python scripts solve a typical SME problem:
Tool 1: Automatic receipt downloader
Fetches all documents from Bexio via API — no more manual clicking. An existing open source application with Microsoft technology was converted into a platform-independent script using AI in seconds, then improved.
Tool 2: Intelligent document analyzer
Google Gemini reads every invoice and receipt, recognizes the date, supplier and document type, and names files automatically in a meaningful way.
The result:
Instead of Scan_2023_X.pdf, files are named 2023-10-12 - Swisscom - Invoice - Internet.pdf.
What areas of application are there for AI in an SME?
This example shows what artificial intelligence can do for Swiss SMEs today — no grand transformation required:
Document management
Automatic sorting and naming of receipts. Recognition of invoice contents and types. Preparation for audit-proof archiving.
Process automation
API integrations without programming knowledge. Automated data preparation. Intelligent file management.
Development aid
Code creation even without technical expertise. Problem solving through AI-powered analysis. Rapid prototype development.
Why this is relevant for Swiss SMEs
Many SMEs shy away from AI solutions because they seem too complex or too expensive. This example shows the opposite:
- Easy to implement: With the right tools, specific problems can be solved without in-depth technical knowledge.
- Cost-efficient: Instead of booking expensive subscriptions, you can develop your own solutions.
- Practical: It's not about groundbreaking innovations, but about small improvements that make everyday work easier.
The AI tool stack for SMEs
The solution was built with a combination of AI assistants:
- Google AI Studio, Windsurf, Visual Studio Code with Gemini Extension and Cursor for script creation
- Dia Browser for the blog post
- Ideogram for the blog image
- Gemini CLI for analyzing the files and images
These tools are accessible and affordable for every Swiss SME today.
From problem to solution in 5 steps
If you are faced with a challenge as an SME, it can usually be solved with AI in five steps.
Identify problem
Where does the team lose time in everyday work? Which tasks are repetitive and error-prone?
Check AI options
Can AI help here? Are there already solutions or APIs that can be used?
Select tools
Which AI assistants fit the problem? Which ones are affordable and accessible?
Develop a solution
Create prototypes with AI help - even without in-depth programming knowledge.
Test and optimize
Adapt and improve the solution in everyday life until it works reliably.
Experience it yourself
The full project is available at github.com/Noevu/bexio-tools. Download the tools, try them out, and see firsthand how artificial intelligence delivers real value in everyday SME operations.
Potential for AI in your company but no clear starting point? Noevu supports Swiss SMEs in unlocking the full value of AI — from training to potential analysis.
Frequently Asked Questions
What does AI bring beyond email?
AI connects existing knowledge, automates routines and makes processes measurably faster. Typical effects: fewer media breaks, consistent quality, shorter lead times. Examples: draft offers from templates and price data, support responses referencing guidelines, automatic meeting minutes. Result: teams work more focused and deliver more reliably.
How does an SME get started with AI in a meaningful way?
Starts with a small, clear use case, e.g. B. Quotation creation or support FAQ. Defines goal, data source and success criteria (time savings, error rate). Build an MVP in 2-4 weeks, test with a pilot team, iterate in short cycles. Important: clear rules of the game, feedback loop, and quick wins that are visible early on so that the team follows suit.
How much does AI implementation cost and is it worth it?
Costs depend on scope and integration. The smart approach: start small (workshops + MVP) and measure ROI against the process. Typical levers: measurably less time on routine tasks, less rework, faster response times. Calculate openly: saved hours × hourly rate, plus quality and revenue effects. That quickly shows whether to scale.
What data does AI need?
Uses existing content: offers, emails, guidelines, product information, project documents. Organizes them for easy access (folder/Tags), sets versions and clarifies access rights. Doesn't start perfectly - the main thing is that the most important sources are structured and up-to-date. Supplemented with short FAQs so that the system prioritizes clear answers.
How does an SME integrate AI into the existing system?
Starts with secondary workflows (documents, email drafts) without risk. Then connect your tools via API or automation: CRM/ERP, DMS, tickets, calendar. Introduces guardrails (templates, approvals, logs). Rollout in stages: pilot, team, area. This way you remain in control and the effect increases step by step.
How do SMEs involve employees in the topic of AI?
Shows the concrete benefits in everyday life, not the technology. Provide brief and practical training (Do/Don't, examples), provide templates and define those responsible. Select a pilot team, collect feedback, improve processes. Visibly recognizes successes. Important: AI supported - decisions remain yours.
What are the risks of AI?
Risks can be managed: wrong answers (hallucinations), data leakage, provider dependency. Countermeasures: verified sources, release processes, access controls, audit trails, CH/EU-Hosting. Defines no-go data, monitors usage and regularly assesses benefit vs. risk. So AI remains a tool, not a risk.
