AI adoption that works in everyday business
I help owners, teams and specialists turn AI experimentation into real work: reliable prompts, repeatable workflows, local tools, automation and clear quality controls.
Theo Gottwald · software developer and AI consultant based in Dettenheim near Karlsruhe, Germany

When this format fits
- Owners, teams and specialist departments that want to turn AI experiments into real work.
- Small and mid-sized businesses with lots of email, proposals, documents or reports.
- Teams that want or need to work locally because data must not go online.
What you get
- Templates, prompt libraries and routines for your real tasks.
- A first workflow that we build together on your real process.
- Review checklists: what may AI write, what must a human check?
- Clear rules on which data stays local and what may be processed externally.
Concrete tasks where I guide customers directly
The format is deliberately practical: we sit down together in front of the real process, I explain how you can use AI, and we build templates, routines or a first workflow.

Amazon business coaching
Question: How can I put AI to work in my Amazon business?
Improve product listings, analyze reviews, research keywords and competitors, and create supplier or support templates.
- Improve product listings: titles, bullet points, description, A+ content.
- Evaluate customer questions and reviews: problems, wishes, new product ideas.
- Keyword and competitor research with a clear checklist.
- Templates for supplier emails, support replies and product comparisons.

Email assistance for business owners
Question: How do I use AI without my replies sounding artificial?
Create natural response drafts, tone and quality rules, follow-ups, proposal sections and human review checklists.
- Response drafts for recurring inquiries.
- Tone and quality rules for customer communication.
- Follow-up templates and proposal building blocks.
- Checklist: what may AI write, what must a human review?

PDFs, contracts and documents
Question: How can AI prepare long documents without me losing the overview?
Produce source-linked summaries, risk and action lists, document comparisons and clear handoff formats.
- Summaries with source references and open questions.
- Risk and to-do lists from contracts, reports or specifications.
- Comparison of several offers or document versions.
- Handoff format for staff or external partners.

Marketing, website and social media
Question: How can AI help me publish better content regularly?
Build content ideas, editorial plans, review controls and repeatable prompts for web, social and newsletters.
- Derive content ideas from products, customer questions and competitors.
- Prepare LinkedIn, blog, newsletter and website copy.
- Set up an editorial plan and quality control.
- Prompts and templates for repeatable content production.

Team onboarding in AI use
Question: How do we get our staff to use AI sensibly and safely?
Train with real examples and establish privacy, source-checking, approval and role-specific prompt libraries.
- Practical examples from your own daily work.
- Rules for privacy, source checking and approval.
- Prompt library for sales, support, administration or management.
- Before/after examples so the value becomes visible.

Local AI workstation
Question: How can I use AI with my own files when this data must not go onto the internet?
Set up local tools and data folders, explain WorkJinn and MCP, and separate local from approved external processing.
- Set up local tools, data folders and working rules.
- Explain WorkJinn, our MCP server and local models.
- Build the document and project workflow step by step.
- Clear separation: what stays local, what may be processed externally.
Specialized setups for Codex, LM Studio and local AI

Codex for developers and teams
Question: How do I use Codex productively in real projects instead of just as a chat?
Repository orientation, well-scoped tasks, bug fixes, tests, refactoring, reviews and team handoff practices.
- Explain the repo structure and phrase tasks cleanly for Codex.
- Run bug fixes, tests, refactorings and documentation with Codex.
- Working rules: never overwrite foreign changes, verify tests, write reviews.
- Team workflow for issues, pull requests, local verification and handoff.

LM Studio and local models
Question: How do I set up local AI that works with my data?
Installation, model choice, hardware limits, local prompt templates and controlled tests with real files.
- Install LM Studio, choose models and explain hardware limits.
- Build local prompt templates for documents, code, emails and research.
- Make the difference between local models and cloud AI understandable.
- First tests with real files and controlled result checking.

MCP, WorkJinn and tool workflows
Question: How can AI not just answer, but work locally with files, tools and processes?
Connect AI to defined local tools, files and workflows with explicit permissions and safety boundaries.
- Explain MCP basics: tools, servers, permissions and safety boundaries.
- Set up our MCP server and WorkJinn as a local workspace.
- Read files, process text, prepare tables and generate reports.
- Rules for controlled actions instead of messy automation.

Local-first AI for business
Question: How do we use AI when confidential data must not go onto the internet?
Define data classes, operating rules, approvals, controlled file workflows and staff guidance.
- Define data classes: local, internal, external, forbidden.
- Formulate working rules for staff and management.
- AI workflow with local files, controlled tools and an approval process.
- Training plus documentation for repeatable use.
MCP server and local AI in practical customer use

Local file workflow
We show how AI works with approved local folders: reading documents, generating summaries, preparing tables, creating text blocks and saving results in a controlled way.
With local processing, sensitive source data stays on the customer system or in the approved local network.

Our MCP server explained
We explain in plain terms what the MCP server does: it gives the AI defined tools instead of unrestricted access. The customer decides which tools and folders are allowed.
This is especially helpful for companies that need local control, traceability and clear boundaries.

Staff instruction
We train staff on real examples: Which data may go into which AI? When do we work locally? How do we check whether an AI result is technically correct?
The goal is safe daily use, not just a technical installation.

Simple working rules for AI
We create simple working rules: data classes, approved tools, approval paths, prompt rules, logging and examples of good and bad use.
This makes local AI easier to control for management, IT and specialist departments. I do not provide legal or tax advice; final approval stays with you.

Packages, pricing and process
Choose a clear entry point from a short working session to guided implementation. Packages and delivery steps are summarized on a focused page.
View freelancer packagesWhat goes well with this
More about the tools: SindByte MCP Server · WorkJinn · LM Chat
Talk first, free of charge – then decide
Describe the process that eats your time. In a free intro call we work out together whether coaching, a workshop, an audit or an automation sprint is the best entry point.
Free 20-minute intro call Send the topic by emailNo obligation, no documents needed – remote or on site near Karlsruhe, Germany. Or write directly: info@it-berater.org · All prices at a glance