Digital Technical Assistant
An assistant with project documents and the current scenario, one that always shows its working and never changes a number without your approval.
What Digital Technical Assistant does
The digital technical assistant answers queries by cross-referencing your documents and current scenario parameters. Unlike general AI, it identifies discrepancies, such as misalignment between availability agreements and technician staffing levels.
Every response cites its sources. When referencing a document, it opens instantly in place, allowing you to verify claims immediately rather than relying on trust.
Suggest mode turns the assistant into a productive agent which can generate editable drafts, suggesting changes for you to review, adjust, and apply. Every action is recorded in the audit trail.
Project threads centralise conversations with colleagues and the model. Reviewers can participate in discussions without having permission to alter the model, ensuring seamless collaboration with all project members.
Live · Demonstration project
7
Threads asked and answered on this project
Each one stays with the project, so the reasoning is still there six months later
What you get with Digital Technical Assistant
-
Grounded in your documents
Answers come from the knowledge base documents scoped to you and your project — and every document used is sourced with the response.
-
Utilises project scenario data
The assistant can see the parameters of the scenario you are working on, so "is this assumption consistent with the contract?" is a question it can actually answer.
-
Suggest mode with review before write
Requested changes arrive as an editable draft in the composer rather than being applied. You review the parameters, change what you disagree with, and apply — or not.
-
Focus on a parameter
Point the assistant at a scenario assumption and ask where the number comes from, what the knowledge base says about it, or what a sensible range is — to get instant suggestions.
-
Uses Velosco AI
Enterprise-grade AI that does not train on your data. Velosco is the end-to-end data processor using 'open-weight' AI models, hosted in the UK.
-
Project team threads
Conversations are kept per project and per thread. Read-only project viewers can be invited from within or outside your organisation.
Digital Technical Assistant in use
Drawn in the product's own interface, and every figure is real: they come from our live Demonstration project — a 288.75 MW floating wind farm in the central North Sea — and its Default Scenario, so anything shown here can be reproduced in front of you. No client or commercial data appears on this page.
An answer grounded in this project's own scenario
A thread from the Demonstration project. The assistant answers from the scenario's parameter snapshot and the documents scoped to the project, and names the inputs it read — so a claim can be checked against the model rather than taken on trust.
Digital Technical Assistant
Key drivers for floating foundationsFloating-foundation sizing is driven primarily by the load-carrying behaviour of the selected 11.55 MW direct-drive WTG. The parameters that feed straight into the sizing of the substructure are:
- Rated power — 11.55 MW — Sets the ultimate electrical torque the turbine must survive in extreme events, and so the tower-top bending moment the mooring system and platform have to resist.
- Rotor diameter — 236 m — Sets swept area and therefore rotor thrust, which is the horizontal load envelope the platform carries.
- Hub height — 150 m — Places the centre of aerodynamic loading higher above the platform, raising static stability demand and amplifying pitch and roll moments.
- Tower mass 1,500 t, RNA 900 t — Drives the ballast and displacement the hull must provide before any environmental load is applied.
21 Sep 26 08:23 am ⧉
Seven threads have been run on this project, from "Understanding PV in PC CAPEX" to "WTG Model Design Basis Differences". Each one stays with the project, so the reasoning behind a decision is still there when somebody asks about it six months later.
A question the model can answer and a spreadsheet cannot
The follow-up in the same thread. Answering it means combining the contract's defect notification period with the component failure model and the exchange strategy — three chapters of two different modules — which is only possible because they are one model rather than three files.
After the 2-year defect notification period and the agreement's cap of 4 exchanges, the owner carries the rest. On this scenario's failure model that is 28.6 expected exchanges over the life, concentrated in the bearings.
21 Sep 26 08:31 am ⧉
Components the owner carries after the warranty ends
Answered from the scenario| Component | Expected exchanges over the life | Exchange team | Share of the owner's exposure |
|---|---|---|---|
| Main bearing | 8.14 | 24 | 28.5% |
| Blade bearing | 7.66 | 24 | 26.8% |
| Blade | 5.91 | 24 | 20.7% |
| Generator | 4.43 | 24 | 15.5% |
| HV transformer | 2.47 | 12 | 8.6% |
| Total | 28.61 | — | 100.0% |
"Thoughts" is the model's reasoning, collapsed. It is there to be expanded, not hidden — an answer you cannot interrogate is an answer you have to take on faith, which is the opposite of what this product is for.
What the assistant writes, the audit log records
Nothing reaches a scenario without an explicit action, and when it does it lands in the same audit trail as a manual edit — with the log recording what wrote it. These are real rows from this scenario's log.
Input audit log
Every write, whatever made it| Parameter | Module | Was | Now | Written by |
|---|---|---|---|---|
| Monthly mean wind speed — Jan 18 Sep 2026, 08:56 | OPEX | 12.7593 | 12.7897 | cascade:monthly-wind-speeds |
| Monthly mean wind speed — Dec 18 Sep 2026, 08:56 | OPEX | 12.1789 | 12.2080 | cascade:monthly-wind-speeds |
| Hub height 18 Sep 2026, 08:56 | CAPEX | 90 m | 150 m | Hand edit |
| Rotor diameter 18 Sep 2026, 08:56 | CAPEX | 120 m | 236 m | Hand edit |
| WTG mass 18 Sep 2026, 08:56 | CAPEX | 435 t | 1,800 t | Hand edit |
| Custom persistence (wave) 17 Sep 2026, 15:51 | OPEX | Default | Site waves (CMEMS WAVERYS) | Hand edit |
| WTG model 17 Sep 2026, 12:46 | CAPEX | SGRE DD-220 | Generic Direct Drive 11.55MW | Hand edit |
The monthly wind speeds at the top of this log — twelve rows in the product, two of them here — were written by a cascade, the model updating derived inputs after a data source changed, not by a person. Recording that distinction is the difference between an audit trail and a list of timestamps.
See it on your own project
Book a demo and we will walk through the module with your numbers, not ours.
More of the toolkit
CAPEX
Build capital cost from the engineering — foundation, turbine, array, installation — so a design change shows up as a number the same day.
OPEX
Model the cost of running the farm from the things that actually drive it: agreements, vessels, weather windows, technicians and component failures.
AEP
Calculate the net energy yield from the site location, wind record and turbine choice. Make losses visible within each project scenario.