AI Consulting before you invest.
We help your business decide which tasks justify an investment in AI. We assess your processes and available data and present a recommendation with priorities, costs and next steps. When an idea needs proving, we build a pilot project using examples from your work.
AI consulting services.
A sound decision on AI rests on three answers: which tasks justify the investment, which data is ready to use and what results a test can demonstrate.
AI opportunity assessment
You receive a prioritised list of AI opportunities, with the expected benefit of each and the reasons for ruling out the rest. We assess tasks such as classifying enquiries, finding information or drafting customer replies. Each task is weighed on how often it occurs, the impact of a mistake and the time your team would spend reviewing the output.
- Task assessment
- Solution comparison
- Review effort
AI data readiness assessment
You find out whether your company's information is ready for an AI solution and what needs correcting before testing. We review manuals, catalogues, enquiries and customer data, and identify outdated versions, gaps and access restrictions. For data intended for analysis, we also check dates, units and indicator definitions.
- Data quality
- Missing information
- Access permissions
AI pilot projects
A pilot project shows, using examples from your work, whether AI can handle the task before your business invests in a full integration. We test cases such as classifying support requests or summarising quote enquiries, including the difficult ones. Response quality, errors and the time needed to correct them are measured against acceptance criteria agreed with you.
- Pilot objectives
- Test cases
- Results evaluation
Codebase preparation for AI agents
Your team gets an organised, tested and documented repository in which AI agents such as Claude Code, Cursor or Codex make fewer mistakes. We review the project's structure, patterns and dependencies and set up an environment that starts with a single command. Fast tests and checks confirm every change. The architecture and project rules are documented for the team and for the agents.
- Structure and patterns
- Tests and checks
- Environment and documentation
AI solution architecture and development
We define the solution's architecture: which model to use, the information sources, the systems to query and the points where a person's approval is required. The choice rests on a comparison of costs and risks across the available options. We then build the first version, with tests and a history that shows how the solution reached each answer.
- Models and costs
- Sources and tools
- First version
AI implementation planning
You receive a plan that organises your AI investment in stages and serves as the basis for requesting a development proposal. Drawing on the assessment and test findings, we set out where to start, which systems to connect, who needs to take part and what costs to expect. The plan separates the first version from improvements to follow later.
- Implementation priorities
- Team responsibilities
- Cost estimates
Does AI make sense for this task?
The choice depends on your data, your process and the consequences of a mistake. In these examples, the right answer is a simple rule, AI assistance or a person's decision.
When each postcode maps to one team, a lookup table handles routing predictably. AI would make an already defined decision unpredictable.
How we run AI consulting.
Every stage involves the people who know the task and rests on concrete examples from your company's work.
Observe the work
We follow the task alongside the person who does it and record the inputs, decisions, exceptions and review time. On that basis, we define the question the analysis must answer and agree with you on the examples, access and people to involve.
Test the hypothesis
We analyse the examples collected and, when the engagement includes a pilot, test the idea against defined criteria. The results are reviewed with you, including errors, the time needed to correct them and the source of each problem: the data, the model or the process itself.
Decide what happens next
You receive a recommendation that weighs implementation effort against the benefit observed. We separate what has been proven from what remains an estimate and propose proceeding, adjusting or stopping. Any integration is quoted as a project in its own right.
Hear it from the people we work with.
Read about the experience of businesses that trusted us with the design and development of their website, from the first sketch to support after delivery.
Meeting INDIGIT was a great move. Good end result, quality and speed. The website came out exactly as we wanted it.
André AbreuAndré Abreu SegurosINDIGIT turned out to be a great partner for optimising our website. Fast, always available, and a mentor when it came to setting strategy. Strongly recommended.
Daniel AzevedoTerzihan ®Pleasantly surprised. A very helpful team, always with good ideas and always willing to put them to work.
Rui MartinsNew Wave InterioresA young, sharp team that actually understands what a company needs to say. Always there for their partners. Thank you!
Cecília AlvesClin ®I am happy. A professional team, they hit the deadlines, and I am pleased with the final work. Thank you and good luck!
Carlos CruzCEO, DomrealceVery professional. Software done as I requested, finished within the timeline. Will definitely work with this company again.
Olav AlbertsViral MarketerMeeting INDIGIT was a great move. Good end result, quality and speed. The website came out exactly as we wanted it.
André AbreuAndré Abreu SegurosINDIGIT turned out to be a great partner for optimising our website. Fast, always available, and a mentor when it came to setting strategy. Strongly recommended.
Daniel AzevedoTerzihan ®Pleasantly surprised. A very helpful team, always with good ideas and always willing to put them to work.
Rui MartinsNew Wave InterioresA young, sharp team that actually understands what a company needs to say. Always there for their partners. Thank you!
Cecília AlvesClin ®I am happy. A professional team, they hit the deadlines, and I am pleased with the final work. Thank you and good luck!
Carlos CruzCEO, DomrealceVery professional. Software done as I requested, finished within the timeline. Will definitely work with this company again.
Olav AlbertsViral MarketerBefore you invest in AI.
The questions that separate a useful AI opportunity from a project that only adds work.
How do I know whether a task needs AI?
AI is worth testing when a task recurs often and involves interpreting text or finding information. Someone must also be able to check the result. If a fixed rule or a better form would solve the problem, that is the simpler option. When the time spent correcting mistakes cancels out the expected benefit, AI no longer pays off.
What should I bring to the first conversation?
One specific task and a few examples without sensitive data. The most useful are a simple request, an incomplete one and one that required an exception. Tell us too who checks the result and which mistake would cost the business most. Tools are chosen later.
Can't I just use ChatGPT or Claude?
In many cases, yes. If the tool handles the task and its access controls, data terms and costs suit your business, it may be the best option. An AI Integration makes sense when the solution must query company systems, be available to several people or automate steps that a single conversation cannot handle. The consulting compares these options before any model or SDK is chosen.
Do I need to organise all my data first?
No. We start with a sample that represents the task. If answers depend on outdated tables or contradictory documents, correcting those sources may be the first step. More files are no substitute for correct information. We agree with you the minimum needed for the test and who can confirm that the information is valid.
Is the pilot already a tool I can use daily?
No. A pilot answers a question with defined limits: for example, whether a summary keeps the details needed to prepare a quote. It may use prepared data and require review every time it runs. Connecting company systems, managing users and monitoring failures are additional work, quoted separately.
How do we measure whether the test was worthwhile?
We compare the same task in the current process and during the test. Beyond the time spent on the task, we count corrections, cases referred to a person and errors that make the output unusable. A fast summary loses its value if it omits the quantity requested. The criteria and sample size are agreed with you. The results apply to the tested sample and do not amount to a promise of savings across the business.
Who from my team needs to be involved?
The person who does the task and someone who can judge whether the result is acceptable. Whoever does the work knows exceptions that a formal process description tends to miss. The system owner clarifies access and technical limitations. Review sessions are planned around the team's routine.
Which data can go into an AI tool?
It depends on the type of information, the provider, the contracted service and its configuration. Before any test, we identify what would be sent, who could access it and how long it would be stored. A provider that does not use data for training may still store it. Where possible, we start with fictional or anonymised examples. Contractual and legal questions need their own review.
Which costs matter beyond the consulting fee?
Implementation and the ongoing use of the solution. Running costs can include the model, document search, storage, external tools and maintenance. The time your team spends reviewing results counts too. We estimate scenarios based on normal workloads and peak periods, rather than extrapolating from a single successful demonstration.
What decision will the deliverables help me make?
A list of the opportunities assessed, with the reasons to pursue or rule out each one, the data required and the next steps. If there was a pilot, we add the tested examples, results and limitations. The document separates what has been proven from what remains an estimate. On that basis, you can request an implementation proposal with concrete criteria or postpone the investment.
Which task deserves a test?
If a repetitive task takes up hours of your team's time, that is where we start. Get in touch and we will assess whether AI justifies a test or whether a simpler solution exists.