We promise clients two things: software delivered faster and cheaper than in the classic model, and quality guarded by automated tests at every step. Below is the whole process, from the first conversation to rollout and the care that follows. No marketing fog. What we do, what you see at each stage, and how billing works.
There is one overriding principle: the production work is done by a team of specialised AI agents, and a human, an experienced engineer, supervises and approves every stage. AI provides the speed and the cost, the human is responsible for decisions and quality.
Step 1: a conversation about the business problem
We don't start with technology. We start with the question of what should change in your company: which process is too slow, where people are rekeying data by hand, what is blocking growth. This conversation is free and carries no obligation.
What you see: a written summary of the conversation, meaning our understanding of the problem, initial directions for a solution and the questions that need settling. If we conclude that software is not the best answer to the problem, we say so plainly.
Step 2: analysis and a quote
If you decide to go further, we carry out an analysis: we refine the scope, get to know the existing systems and data, identify the risks. When working on an existing system we often start by reconstructing its documentation, because that shortens everything that comes after.
What you see: a document with the scope of work, a schedule and a quote. We prepare the quote after the analysis. We don't publish a price list, because every case differs in complexity; in return, the quote you receive is concrete and binding for the agreed scope. Up to this point you can walk away with no obligations.
Step 3: a prototype in days, not months
This is where the AI process shows its biggest difference from the classic model. Instead of weeks of waiting for a "first version to look at", we deliver a working prototype within days of the analysis being approved.
The prototype is not a mock-up or a slide deck. It's working software covering the main path, something you can click through and judge. The goal is a quick confrontation with reality: it's easier to evaluate something that works than a document describing it.
What you see: a working prototype in a test environment, available to your team, along with brief instructions on what to check and how.
Step 4: iterations with a full battery of automated tests
Based on your feedback, we develop the solution in short iterations. Each iteration ends with a version you can look at and comment on. There are no months-long periods of silence.
The heart of this stage is automated quality control. Every change to the software goes through a full battery of automated tests:
- unit: does each element work correctly in isolation,
- integration: do the elements cooperate correctly with each other and with your systems,
- E2E (end-to-end): do entire business processes run through from start to finish,
- user interface tests: does the application behave correctly in the browser, the way the user will see it,
- security: does the change open no vulnerabilities and is the data protected,
- performance: will the system stand up to real load,
- regression: has the new change broken anything that worked before.
In the classic model, testing this broad is often the first casualty of budget cuts, because every tester hour costs money. With us the tests run automatically on every change, so speed does not come at the expense of quality. The supervising engineer reviews the results and approves the iteration before showing it to you.
What you see: successive versions in the test environment after every iteration, plus the status of the work. What has been done, what is in progress, what comes next.
Step 5: rollout
Deployment to the production environment follows a previously agreed plan: with a backup, a rollback plan in case of problems and post-launch verification. We deploy only after you have accepted the test version. Nothing reaches production without the client's consent.
What you see: a deployment plan before, confirmation and a report after. Plus documentation of the solution, because we know better than anyone what its absence costs.
Step 6: care after rollout
We don't disappear after the rollout. We agree a care model with you: monitoring, responding to reports, small fixes and further development. Because the solution comes with a complete set of automated tests and documentation, subsequent changes are fast and safe. Each goes through the same quality-control process as the first version.
How billing works
Transparently and in stages:
- Initial conversation: free.
- Analysis and quote: agreed individually; afterwards you receive a concrete quote for the further work.
- Build and rollout: per the quote from the analysis, with payments tied to stages you can see and evaluate.
- Care after rollout: a model matched to the scale of the solution, agreed before deployment.
We don't bill "by the hour", where the client pays for time rather than results. You pay for the agreed scope, and progress is visible in working software, not in timesheets.
FAQ
Does AI mean nobody is accountable for quality?
Quite the opposite. A supervising engineer is accountable for every stage and approves the results of the AI agents' work, and the contract with ESKOM AI defines liability just as in the classic model. AI changes how the work is done, not the rules of accountability.
What about the confidentiality of our data and code?
We agree the scope of access, the confidentiality rules and where data is processed before work begins, and we write them into the contract. We work in controlled environments, and access is limited to the necessary minimum.
Is this process suitable for small orders, or only for big projects?
Both. It's precisely with smaller changes that the advantage of the AI process is most tangible: low organisational overhead means a small order doesn't cost disproportionately much, and the complete set of regression tests makes sure a small change doesn't break the rest of the system.
Let's see how this process would work for you
The simplest way to find out is on your own example. We invite you to a free consultation. You describe your business problem, and we'll tell you what the process, the schedule and the next steps would look like for your case.