Customer login 
Claude AI is currently being hyped everywhere – but what can it really do? At OIT GmbH, we use both Claude and ChatGPT productively in the
team. Not as a gimmick, but in real everyday life:
for texts, concepts, internal knowledge work, support and automation. In this article, we openly share our
Claude AI experiences, typical use cases, cost issues (tokens!), and risks you should be aware of.
Why we use both in the team: Claude and ChatGPT
Our experience: there is no “one best AI”. The strengths vary depending on the task.
That’s why we take a coworking approach: humans + AI together – and we choose the model that delivers the best results for the use case
.
Claude: our strengths in practice
- Very clean in structure, organization and “calm” writing style
- Strong for long texts and documents (e.g. protocols, concepts)
- Good summaries and clear chains of argumentation
ChatGPT: our strengths in practice
- Very flexible for brainstorming, variations, creative formulations
- Practical for quick iterations in workshops and coworking sessions
- Top if you “think along” and want to try out alternatives quickly
Concrete application examples from our everyday life (cowork)
To give you a feel for how we really use AI, here are typical examples from our day-to-day work:
1) Offers & concepts
Structure, service modules, benefit argumentation, clear outline – we save time, increase quality
and deliver more consistent results.
2) Support & Tickets
Summary of processes, suggestions for answers, prioritization according to urgency –
always with human control.
3) Knowledge work & documentation
Notes become comprehensible documentation, meetings become clear to-dos,
“islands of knowledge” become internal standards.
4) Marketing & SEO
Landing page structures, FAQ blocks, article drafts, meta variants – faster in the first version,
then fine-tuned by us.
Costs & tokens: the point that many underestimate
In practice, AI costs are not just “subscription per user”. As soon as AI moves into processes (e.g. support, CRM, automation),
consumption arises – and this is often billed via tokens or API usage. If you don’t have this under control,
it can become expensive: prompts that are too long, documents that are too large, too many runs or uncontrolled automation.
Our approach: transparency + control + efficiency
- We talk openly about typical cost traps and how to avoid them.
- We show which tasks really need AI – and which don’t.
- We optimize prompts/workflows so that fewer tokens are burned.
Risks & dangers: honestly named (so that it works in the company)
AI can be extremely productive – but only if it is introduced correctly. The most important points that we always address in projects:
- Hallucinations: AI can formulate convincingly wrong things – therefore: test steps & approvals.
- Data protection: Which data can go in? Which not? Clear rules and secure environments are a must.
- IP & confidentiality: internal content should be protected – policies and technical measures help.
- Cost explosion: uncontrolled automation can drive up token costs – monitoring & limits are important.
- Process chaos: AI without structure often leads to more work instead of less – we build it in cleanly.
Our own product: reduce token costs, operate AI efficiently
Because we have seen precisely these cost and scaling problems in practice, we at OIT have developed our own product,
, to reduce token costs and operate AI in companies in a more controlled and efficient way.
This is particularly exciting if AI is regularly integrated into support, sales or internal knowledge processes.
You can find out more about our AI offering and how we implement AI solutions in companies here:
Fancy seeing this live?
If you want to know whether Claude AI or ChatGPT (or both) is right for your company, we will be happy to show you –
including experience, cost reality and typical stumbling blocks. Relax and enjoy:
for a coffee or in an online meeting.
