assess AI potential, before you develop let.
Which use case is really worth it, what does it require and what will it cost? Honest, technical assessment from a team that also builds AI itself.
Fixed price €2,500 · deductible upon implementation · honest assessment
Potential analysis for Fixed price
No months-long consultation process. A compact use case check that provides you with a reliable, technically sound basis for making decisions.
Can be fully credited if the implementation is commissioned. You can't lose anything: the resulting document is yours.
You will receive this in writing
- Use case evaluation with prioritization - Which of your AI ideas are worthwhile and in what order.
- Technical target architecture - How the AI solution specifically fits into your processes and systems.
- Data protection and operating model - Who runs what, where does the data run, who has access.
- Effort estimate for pilot & rollout - Realistic assessment of time and costs, separately for each phase.
- Concrete 90-day roadmap - Traceable steps from start to first productive MVP.
- Decision template for management / IT - Understandable for both sides, technically and economically.
A technically sound basis for your decision
We check your AI use case along the points at which projects really fail or succeed in practice.
Business goal
What effect should the use case create? Less manual effort, faster processes, better data quality?
Process reference
Where does the effort come from today? Which steps are repetitive, error-prone or time-consuming?
Data situation
What data and documents are available? Are they structured, complete and of sufficient quality?
Technical feasibility
Which architecture, interfaces and AI models are possible? What technical risks are there?
Risks & Limitations
What data protection, quality or operational risks are there? What can’t the AI solution do?
MVP scope
What would be a useful first version? Which defined pilot delivers measurable results with reasonable effort?
From fixed-price entry to scalable AI system.
We accompany you from the initial analysis to company-wide scaling.
AI consulting vs. pure strategy consulting
The difference lies in the development expertise in the consulting team.
| Criterion | Pure strategy consulting | CodeGuides AI advice |
|---|---|---|
| Fokus | Concepts and scenarios | ✓ Implementable use cases |
| Technical depth | General recommendations | ✓ Feasibility, APIs, architecture |
| System landscape | Often not rated | ✓ Data, ERP, interfaces checked |
| Result | Strategy paper / slides | ✓ Clear MVP or implementation plan |
| Cost estimation | Often too optimistic | ✓ Realistic from development experience |
| Implementation | New provider required | ✓ Direct transition possible |
How much does AI really cost in companies?
Concrete budget frameworks instead of marketing so that you can understand what to expect.
| Project phase | What happens | Typical budget framework |
|---|---|---|
| Potential analysis | Use case evaluation, data check, target architecture, roadmap | 2,500 € fixed price |
| Pilot / MVP | A delimited use case productive, including integration | 15,000 to 40,000 € |
| Rollout | Several use cases, data connection, roles/rights, operation | 40,000 to 150,000 € |
| Operation & Scaling | Maintenance, monitoring, new use cases, model updates | ongoing, project dependent |
Reference values from CodeGuides projects in Germany. The actual framework depends on the data situation, integrations and operating model (cloud or on-premise) and is specified in the potential analysis.
Cloud AI vs. local AI (on-premise)
One of the most important decisions before you have it developed.
| Criterion | Cloud AI (e.g. OpenAI, Azure) | Local / On-Premise AI |
|---|---|---|
| Data sovereignty | Data leaves the company | Data remains completely in-house |
| Entry costs | Low, ready to go quickly | Higher, own infrastructure required |
| Data protection / compliance | Depending on the provider & location | Full control, ideal for regulated industries |
| Operating expenses | Low (Managed) | Higher (own hosting & operation) |
| Suits for | Uncritical data, fast pilots | Sensitive data, trade secrets, regulation |
This is how AI consulting works
Structured, transparent, without a long lead time. From the first idea to the clear implementation plan.
Initial consultation (online)
You tell us the goal, starting point and ideas. If necessary, we will then take a look at the conditions on site.
Collect and evaluate use cases
Collect possible AI use cases in a structured manner and classify them according to benefits, data, effort and risks.
MVP scope & implementation plan
Determine a realistic first step, with architecture, effort and concrete next steps.
Example: from analysis to productive AI pilot
A representative, anonymized example of how an AI project typically works for us.
AI assistant based on your own documents (RAG)
Representative example of a medium-sized B2B service provider in Germany.
Industry
B2B service provider, medium-sized companies (Germany)
Initial problem
Quotation and document creation is very manual, takes a lot of time, inconsistent quality
Solution
AI assistant based on your own documents (RAG), connected to the existing DMS
Tech stack
LLM + vector database, API connection, role-based rights, GDPR-compliant hosting in DE
Procedure
Potential analysis → defined pilot → gradual rollout
Result
Significantly faster document creation, more consistent quality, employees relieved
Duration until pilot
approx. 8 weeks after analysis
Budget framework
Analysis €2,500 · Pilot in the €15,000 to €40,000 range
CodeGuides is an app and AI agency from Germany for Flutter apps, custom software, AI automation and local AI infrastructure. Consulting and implementation come from a single source: We assess AI use cases technically, build the first pilot and scale it up to company-wide operation, GDPR-compliant and with hosting in Germany.
Who is AI consulting useful for?
Makes sense if you know that AI could help, but not yet how.
Useful if...
- You have several AI ideas but don't know where to start
- You want to assess the costs and benefits before making an investment decision
- Existing systems, data or processes must be included
- You are aiming for an MVP instead of a large AI project
- The department and IT need a basis for decision-making together
Not particularly suitable if...
- You are just looking for a general AI lecture or tool training
- No concrete question or process idea is available
- No internal contact person from the department or IT is available
- No budget is planned for implementation or MVP
Rate first.
Then develop.
We would rather show you an inconvenient restriction now than have you realize later that the database is incorrect. Honest, technically sound, so that your project really works in the end.
What companies ask about AI projects in forums and on Reddit
Real questions that arise in discussions about AI costs, cloud vs. on-premise and agency selection, answered here honestly and without sales phrases.
Reddit question“Is an AI agency even worth it, or do I do it myself with ChatGPT?”
A ChatGPT subscription is sufficient for simple word tasks. As soon as it's over Own data, integration into existing systems, data protection and reliable results in operation It will be a software project, not a chat window. This is exactly where an agency that honestly evaluates the use case beforehand is worthwhile instead of immediately selling you a project.
Forum question“How much does AI realistically cost a company, without marketing blah?”
Honest ranges from practice: Analysis ~2,500 €, first pilot ~15,000 to 40,000 €, Company-wide rollout ~40,000 to 150,000 €. Anyone who gives you a fixed number without looking at the data and systems is guessing. The biggest cost driver is almost never the AI model, but rather data preparation and integration.
Reddit question“Cloud AI vs. local AI, which one should you choose in medium-sized companies?”
Rule of thumb: Non-critical data → Cloud (fast, cheap). Sensitive or regulated data → local/on-premise (data sovereignty). Many people drive hybrids. It's not the ideology that's important, but what data the use case actually touches on, which you clarify before the first line of code.
Forum question“How do I know if an AI project will fail?”
Typical warning signs: No clearly defined use case, no or bad data, No internal responsible person and no integration into existing processes. If three of these apply, a large project is risky; a small, measurable pilot is almost always a smarter start.
Frequently asked questions about AI consulting
AI consulting includes the analysis and evaluation of AI use cases in your company: identification of sensitive use cases, examination of the data situation, assessment of technical feasibility, analysis of integration needs and data protection as well as a prioritized recommendation for next steps.
If you know that your company could benefit from AI, but it is not yet clear: which use case? What data do we need? What does this cost? How do we integrate this into existing systems? Then a structured AI consultation is the right first step.
The entry point is a potential analysis for a fixed price of 2,500 euros, which can be fully credited if the implementation is commissioned. We examine the use case, data situation and system landscape and provide a reliable recommendation. This is followed by the concept, structure of the LLM for an initial use case and further scaling.
Strategy consultants evaluate market potential and business scenarios. CodeGuides evaluates technical feasibility, data availability, system integration and development effort. We are a software development agency, not a management consultancy.
Yes. If the analysis shows that an AI project makes sense, CodeGuides can go straight into development, no new vendor selection process.
Then let’s say it clearly. We are not interested in a project that doesn't work afterwards. The consultation provides an honest assessment, even if the result is a clear no.
Depending on the scope: A compact use case workshop with subsequent analysis typically takes 2 to 4 weeks.
No. All you need is a specific process or problem. Identifying the right AI approach is part of the consultation.
As a rough guide: The potential analysis is a fixed price of 2,500 euros. In our projects, a first productive AI pilot (a clearly defined use case) usually costs between 15,000 and 40,000 euros. A company-wide rollout with several use cases, data connection and operation typically costs 40,000 to 150,000 euros. The exact framework depends on the data situation, integrations and operating model (cloud or on-premise).
Cloud AI (e.g. OpenAI, Azure OpenAI) is quicker to get started and cheaper to get started, but sensitive data leaves your house. Local or on-premise AI (own LLM infrastructure) keeps all data in the company and is often the better choice for data protection, trade secrets and regulated industries, but requires more infrastructure. For many medium-sized companies, a hybrid approach makes sense: cloud for non-critical tasks, local AI for sensitive data. It is precisely this consideration that is part of the potential analysis.
The potential analysis typically takes 2 to 4 weeks. A first productive pilot can then usually be implemented in 6 to 12 weeks, depending on the data situation and depth of integration.
Yes. Especially in medium-sized companies, AI projects rarely fail because of the technology, but rather because of unclear use cases, poor data or a lack of integration into existing systems (ERP, DMS, CRM). A compact potential analysis prevents a six-figure budget from flowing into a project that is not operationally viable.
Potential analysis with real experts
Answer three short questions about your project. We will contact you personally within one working day, not an automated offer.
- Fixed price €2,500, fully creditable upon implementation
- Development team, no consulting sales
- Honest assessment, even if it's a no
Thank you very much!
We have received your request. We will contact you within one working day with a personal assessment.
An AI idea, but not one yet
Clarity about feasibility?
We check technical feasibility, data situation and system integration and give an honest recommendation for the next step.
Fixed price €2,500 · deductible upon implementation · response in 1 working day