Keep the assignment focused.
Define what needs to be delivered and how it will be checked. Agree on price and timing before paid work, and agree separately on additions.
How we work
Clear assignments. Work you can review.
A useful result starts with understanding the work. MGV combines hands-on engineering leadership with focused implementation, testing, and review so you can see what was delivered and make an informed next decision.
Using the budget well
Put the effort where it matters.
Cost management starts with how the work is assigned, checked, and carried forward—not just the price of an AI tool.
Keep useful work.
If a task is interrupted, inspect what was completed and recover the missing step. Do not automatically pay for the same work to be done again.
Match the checks to the change.
Test the parts affected by a change and use independent review to find faults or unnecessary complexity. Repeat checks when the work changes, not just to create another report.
These are MGV's working practices for directing effort and managing rework. Actual project costs depend on the agreed scope, tools, and client environment.
The delivery process
From the first question to handover.
-
Agree on the outcome
Discuss the problem, the people doing the work, and the systems involved. Define a useful result and who will accept it. Agree on deliverables, price, and timing before paid work begins.
-
Make the first step concrete
Choose a manageable first assignment. Where a design decision needs testing, use a prototype or limited implementation to give the discussion something people can see and try.
-
Build within the assignment
Break the work into defined tasks. When AI agents assist, give them specific assignments and the context and permissions appropriate to their part. Engineering decisions remain human-led.
-
Test and challenge the result
Run the checks relevant to the agreed work. Use independent review to look for faults, question unnecessary complexity, and support findings with evidence. Explain unresolved limitations alongside the result.
-
Accept, release, and hand over
Review the result against the agreed acceptance criteria and record what is delivered. Any release stays within the authority you have granted. The handover makes the current state and any proposed next work clear.
Human-led delivery
AI helps with the work. People own the decisions.
Adolph sets the technical direction and leads the build-and-review process. Your team retains the acceptance and release authority defined for the engagement.
The method applies whether an assignment needs custom software, an integration, a prototype, or AI assistance.
A working example: the Gateway
MGV uses the Gateway to define AI assignments, set permissions for reviewing or changing files, and keep results available for review. Its case study shows the specific work and checks demonstrated so far.
The Gateway supports this delivery method; it does not automatically enforce every step or replace human judgment.
The first engagement
Start small enough to make a clear decision.
Begin with a brief introductory conversation at no charge. Together, we agree on a useful first assignment.
Deliverables, price, and timing are agreed before paid work begins. Any later work is agreed separately.