01
Process map
We clarify where recurring work arises, which teams are affected and which processes are suitable for closer examination.
- relevant areas
- recurring tasks
- manual handovers
- typical examples
- first use case candidates
thirdmind · AI Compass
AI potential analysis for the right first use case.
Many companies have enough AI ideas. The difficult part is choosing: Which use case is worth it? What data is needed? What can be implemented internally? And where is professional help needed?
The AI Compass creates a reliable basis for decisions. We examine specific processes, evaluate practical AI levers, and assess each use case by benefit, feasibility, data situation, and risk.
After the AI Compass, you do not just know what AI could do. You know which process should be checked, built, or deliberately set aside first.
Your AI ideas
- Automate support
- Prepare offers
- Check documents
- Maintain CRM
- Make knowledge from internal sources usable
5 review areas
01 Process map
02 Value rating
03 Feasibility
04 Build or buy
05 Decision basis
Decision · impact × effort
QuickWin
Strategic
Learning case
Park
01 Risk
An AI potential analysis does not start with the tool. It starts with the use case.
Many companies already have enough AI ideas: automate support, maintain CRM data, prepare offers, check documents, make internal knowledge usable.
The real decision is which of these is mature enough, important enough, and controllable enough to start. This is where a good potential analysis separates useful AI use cases from nice ideas.
That's exactly what the AI Compass is for.
You evaluate AI use cases before the budget goes into implementation.
- 01 which processes have real AI levers
- 02 which benefits can be realistically verified
- 03 which data, systems and rights are relevant
- 04 where risks, approvals, or control limits arise
- 05 what can be implemented internally and where external help makes sense
- 06 how each use case should be ranked by business impact and effort
The result is not a list of ideas. It is a basis for decisions.
02 Flow
The AI potential analysis in five assessment areas.
02
Value rating
We evaluate which use cases have enough operational value to deserve more than a neat AI demo.
- frequency
- Time spent
- Quality risk
- Team relief
- Connection to the business goal
03
Feasibility
We check which data, systems, rights, and control points are necessary for an idea to become a resilient AI process.
- Data sources
- System access
- Roles and rights
- Approvals
- Human-in-the-Loop
- typical edge cases
04
Build or buy
We distinguish what your team can handle internally from the parts that need professional architecture, integration, or implementation.
- internal implementation options
- No-code or tool options
- Need for integration
- Security requirements
- external support needs
05
Decision basis
At the end, you do not get a loose collection of ideas. You get a clear recommendation for the next step.
- prioritized start process
- sensible MVP scope
- Risks and open questions
- internal vs external implementation
- Classification in the impact-effort matrix
03 Decision
Each use case ends up in one of four fields.
Impact high · Low effort
QuickWin
High benefit with little effort. Can be implemented quickly, with a partner or often even internally.
Start now
Impact high · Effort high
Strategic pilot
High benefit, but higher effort. Worth considering as a larger project with an implementation partner.
Build with partners
Impact low · Low effort
Learning case
Limited benefit, low effort. Good for building experience with digital employees internally.
Implement internally
Impact low · Effort high
Set aside deliberately
Limited benefit with high effort. Set aside deliberately instead of tying up budget.
Set aside
04 Result
A use case assessment that can be reused internally.
The result can be used directly by management, departments, and IT. It shows which AI use case makes sense first and what needs to be clarified.
- 01 Process map with relevant AI levers
- 02 prioritized use case matrix
- 03 Assessment by benefit, feasibility, risk, and data situation
- 04 Build-or-buy assessment
- 05 Recommendation for the first digital employee
- 06 useful scope for a pilot or internal start
- 07 risks, assumptions, and open questions
- 08 decision basis for management, departments, and IT
What a digital employee can specifically do.
DefinitionA digital employee is an AI system that carries out or prepares operational work steps. It works with a clear task, relevant context, defined system access, and control boundaries. In the AI Compass, we check whether such an employee makes sense for your process.
- Classify support tickets and prepare draft responses
- Research leads and prepare accounts for sales
- Pre-structure offers, emails or documents
- Merge internal information from different systems
- Run recurring checks
- Initiate follow-up tasks and keep track of open points
05 Who it fits
For companies that see AI potential but want to choose the first use case carefully.
Fits when ...
- you have many AI ideas but no clear prioritization
- your teams already use AI tools but still lack process architecture
- support, sales, finance, or operations lose time to recurring work
- you want to know what your team can implement internally
- you want a reliable decision basis before a larger AI project
Fits less well when ...
you only need general AI training, or you cannot yet provide access to processes, examples, and internal stakeholders.
Typical topics in the AI Compass.
01 / Einsatzfeld
Support
Recurring requests, ticket triage, response preparation, knowledge gaps, escalations.
Example: A digital employee reads new tickets, recognizes topic and urgency, suggests answers, and flags cases that need human expertise.
02 / Einsatzfeld
Sales
Lead research, account preparation, CRM maintenance, follow-ups, offer preparation.
Example: A digital employee prepares accounts before a first meeting, researches relevant information, and creates meeting notes for sales.
03 / Einsatzfeld
Operations
Document review, email triage, data comparison, internal coordination, recurring approvals.
Example: A digital employee checks incoming documents against defined criteria and only prepares the cases where a decision is necessary.
06 Warum thirdmind
We check with an eye on implementation.
thirdmind builds digital employees for companies: with process understanding, AI architecture, controlled tool use, human-in-the-loop, and monitoring.
That is why we do not evaluate use cases only by asking whether AI can theoretically do something. We check whether the process can be built, controlled, and later expanded in day-to-day operations.
Control questions we answer
- 01 What data is the system allowed to use?
- 02 What actions is it allowed to perform?
- 03 When does a person need to review?
- 04 How do errors become visible?
- 05 What can be implemented internally?
- 06 Where is professional help needed?
References
IMV
For IMV, several automation fields were assessed, including documents, knowledge access, communication, and administrative standard processes.
- Collected and prioritized automation fields
- Derived a recommendation for useful first implementations
Innotec
At Innotec, existing AI experiments, process ideas, and operational bottlenecks were translated into a prioritized implementation logic.
- Structured use cases from billing, field sales, IT, sales, and back office
- Separated a quick win, lighthouse topic, and MVP scope for the next step
LCI
For LCI, sales and meeting processes were structured around a possible AI coworker, with a focus on research, outreach, offer work, and follow-up.
- Prioritized sales research, outreach, and offer workflows
- Mapped meeting documentation and follow-ups as the next work area
With us, internally or with existing partners.
The AI Compass is designed to stand on its own. You can then implement with thirdmind, continue internally, involve your existing IT team, or work with another implementation partner.
Our aim is to leave you with a basis you can use without us. You know which use case makes sense, which data and systems are relevant, what scope is realistic, and where the risks are.
Some results lead to a pilot. Some lead to internal implementation. Some lead to a deliberate no. That is part of the value.
07 Rahmen
The framework depends on your processes.
We deliberately do not publish a flat rate. Before a potential analysis, we check the scope, data situation, and internal stakeholders. We then propose a sensible framework.
Scoped individually
- Fit check
- Process and use case analysis
- System and data testing at a conceptual level
- Evaluation of possible use cases
- Build or Buy Assessment
- Final document
- Recommendation for the next step
We will discuss the specific framework after the fit check. The AI Compass should be large enough to make reliable decisions and small enough not to become a major project itself.
08 FAQ
What teams usually ask before an AI Compass.
What is an AI potential analysis?
An AI potential analysis checks which processes are suitable for AI, which data and systems are affected, which risks arise, and which use case makes sense first. The AI Compass turns that into a concrete decision basis.
Is a digital employee already being built in the AI Compass?
No. The AI Compass evaluates use cases, clarifies a sensible scope, and shows whether implementation is worthwhile. The build happens later, either in a separate sprint or internally.
Do we have to continue working with thirdmind after that?
No. The AI Compass is structured so you can use the result independently: internally, with your existing IT, with another implementation partner, or with thirdmind.
How much time do we need internally?
We usually need a kickoff, one or two process discussions, relevant example materials, and a results call. The exact effort depends on how many areas the AI Compass covers.
Do we need perfect data?
No. The data situation is part of the analysis. The AI Compass often shows which data is already usable, what is missing, and which use case is clean enough to start.
Is this an AI workshop?
Not quite. A workshop often provides orientation. The AI Compass is more focused on assessment: use case, scope, benefit, risk, data situation, build-or-buy, and next step.
What size company is this suitable for?
The AI Compass is best suited for medium-sized companies and B2B teams with operational processes in which repetition, manual checking or information searches take up a lot of time.
Can we start implementing it straight away?
If the use case is clear enough after the AI Compass, yes. The result can also be that you should first clarify data, rights, or process responsibility internally.
Next step
The best way to get started with AI potential analysis is a short fit check.
In 20 minutes, we clarify whether the AI Compass makes sense for your company. If it does, we discuss the framework, process, and start. If it does not, we say so openly.
A clear assessment of which AI use case makes sense first, with a basis you can keep using later.