- I have no specification. Is that a problem?
- No, that is the usual situation. We work the task out together first: you describe how everything runs today and what you want to end up with, I suggest how best to do it and in what form. Out of that conversation I write the specification and show it for approval before any work starts.
- I only have an idea, I do not know how best to build it
- That happens often and it is a fine way to arrive. On the call I look at what can realistically be built, by what means and in what form: sometimes the task turns out simpler than it looked, sometimes the opposite. The solution comes out of that conversation and goes into the brief.
- Do I need an agent or automation?
- The difference is not whether there is a neural network inside: there is one either way. Look at who picks the next step. If every step can be drawn in advance, that is automation, even with three models inside it. If the step depends on what a person writes and what turns up in your systems, that is an agent. Most of the time automation is what is needed, with an agent working inside the one step you cannot plan for.
- What will it cost?
- It depends on the task: an integration or data collection costs noticeably less than an agent or a finished product. I give the exact figure after the call, together with the timing. The work is split into stages and you pay per stage, not for the whole project. The figure is fixed before we start and does not grow along the way.
- How long does it take?
- From one or two weeks for an integration or data collection to a month or six weeks for an agent or a product. I give the exact timing together with the price, before we start.
- Do we work under a contract?
- If you want one. If that feels safer we draw up a contract and sign it electronically. Some projects run without one, on what we agreed in writing, and that is fine too.
- Who actually does the work?
- I do. There are no managers between you and the person writing the code. For larger volumes I bring in specialists I have worked with, but I do not delegate the architecture or the responsibility for the result.
- What do you need from me along the way?
- A description of how things run today, and access to the services we are working with. After that questions come up here and there, and a couple of messages a day covers them. The rest is on me.
- What if it does not work out?
- That is what the first call is for: if I can see the task will not pay for itself, or that it can be solved more simply and cheaply, I say so straight away. And the work goes in stages, so you see the result of each one before going further.
- Whose code and whose accounts?
- Whichever we agree on, and it affects the price. I can keep everything on my side and hand over the finished result, or build it on your keys and inside your services: then the code and the access stay with you, with no tie to me. Which one works out better we go through before we start.
- What about security?
- Protection is built into the product itself: locked down database access, authenticated sign in, rate limits so the service cannot be flooded from outside. Backups are always in place. I take the minimum access to your services that the work requires. What the system stores, and what it does not store at all, we go through before we start: the bigger the project, the earlier that is worth agreeing.
- What happens after launch?
- I stay reachable. Real customers always turn up something the tests did not, and that is a normal part of the work. How support works after that is agreed together with the price.