What does Oqult actually do?
Two things: AI consultancy — strategy, use-case selection, building and deploying systems, governance and training — and quantum computing consultancy, which is earlier-stage and narrower. Most of our work is the first; we are honest that the second matters to fewer organizations today.
How do we know whether AI is worth it for us?
Usually an afternoon of conversation and a couple of weeks of assessment. The question is not whether AI works, it is whether any process you own is repetitive enough, high-volume enough and measurable enough to repay the build and the running cost. Most organizations have a handful of candidates and a lot of distractions.
Should we build our own models?
Almost never from scratch. The realistic choices are an off-the-shelf product, an API with your own data and evaluation around it, or fine-tuning where you have something genuinely proprietary. We will tell you which, and we do not take margin on the answer.
Do we need a quantum computer?
Almost certainly not, and not for some years. The exception is cryptography: data you encrypt today can be captured and decrypted later, so migration planning is present-tense work whatever the hardware does.
When will quantum computing matter for us?
It depends on the problem. Simulation of molecules and materials is the strongest case and needs error-corrected machines, which roadmaps put in the 2030s. Optimization claims are largely unproven against good classical solvers. Cryptographic migration matters now.
What does an engagement cost, and how long does it run?
An assessment is measured in weeks and priced as a fixed fee. A pilot typically runs one to three months. Embedded work runs three to twelve. We would rather scope a small piece that answers a real question than sell a long programme whose conclusion is uncertain.
Do you work with teams that have no AI or quantum background?
Usually. Most of our clients are capable engineering organizations without specialists in either field, which is the normal state of affairs. Part of every engagement is leaving your team able to continue the work.
Will you tell us not to do the project?
Regularly. It is the deliverable clients tend to value most in hindsight, and it is why we build the baseline before we build anything else.
Who owns the work?
You do. Code, models, documentation and data stay yours, in your repositories, under your licences. Client work is confidential by default.
Something else
If your question is not here, it is probably the interesting one. Send it to us — enquiries are answered by the people who do the work.