How we build.
The standards every engagement runs on, whichever practice it sits in. Written for the person who will review our pull requests, own the system afterwards, or answer to a board about how AI was used. If that person is not you, this is the page to send them.
Four things written down before any project starts.
They are agreed against a working version you have used, and they do not change once you are in the work. The same four appear on our See it work page, because they are the same four.
S1
Scope, priorities and explicit exclusions
A clear build boundary. What is in, what is out, and what is deliberately later.
S2
Measures of success and acceptance evidence
A shared definition of done, stated as something observable rather than something argued.
S3
Milestones, fees and payment points
Known costs and decision points. Fees attach to milestones you define; payment follows completed milestones.
S4
Code ownership, access and handover
A system your team can operate. Source, environments, credentials and decisions in your hands throughout.
Engineering standards
01
Repository in your account
We work in your repository from the first commit. There is nothing to transfer at the end because nothing ever sat with us.
02
Review before merge
Every change is a pull request, reviewed by a second engineer, and by yours if you have them. Nothing reaches a shared branch unreviewed.
03
Automated tests from the first sprint
A regression suite grows with the build and runs in your CI. It is the same discipline our quality engineering practice delivers as a service.
04
Environments as code
Development, test and production are defined in the repository and can be rebuilt from it. The cost model is written down before anything is provisioned.
05
Decisions and documentation as we go
Architecture decisions, the current system map and the handover notes are written during delivery, so handover is a review rather than a scramble.
06
Security baseline
Least-privilege access, secrets outside the codebase, and dependency updates as part of the work.
AI in our delivery
Where AI is used, where it is not, and who reviews it.
AI runs through our delivery because it makes finding out affordable: the working version before commitment, the first-cut tests, the analysis of an export. It does not run unreviewed, and it does not see data it has not been cleared to see.
Used for
- Analysis of the real example — exports, logs, screenshots and process descriptions, to map what the current system does.
- First-cut screens and prototypes — the working version you use before commitment is built faster because of it.
- Test generation — drafting regression cases from observed behaviour, then reviewed and pruned by an engineer.
- Review assistance — a second reader on pull requests, never the only one.
- Documentation drafting — decision records and handover notes, corrected by the engineer who made the decision.
Not used for
- Unreviewed code to any shared branch. Generated code goes through the same pull request as hand-written code.
- Your data outside agreed environments. Which tools may see which data is settled at contract and does not change without your agreement.
- Training. Your code and data are not used to train any model, ours or anyone else's.
- Production access. No agent or assistant holds credentials to your production systems.
How we price
01
Fees attach to milestones you define
The client states what must be observably true for each milestone. We may negotiate the condition before signature; we do not reinterpret it afterwards.
02
Written down before signature
Scope, milestones, fees, payment points and exit terms are in the agreement before anything is signed. There is no model to discover after the first call.
03
A clean exit at any milestone boundary
Either side can end the engagement at a milestone boundary on notice, without exit fee, with everything transferred in a state another capable team can operate.
04
Retainers report their hours
Support and stabilisation retainers state the hours used each month, and are cancellable at the month end.
05
No reseller margin
We hold no platform loyalty and earn nothing from the technology we recommend. The simplest sufficient option is the one we propose, including when it is not us.
One delivery team across the UK and India.
Client engagement and delivery leadership sit close to the client. Engineering depth sits where the depth is. Same team, same standards, one repository.
United Kingdom and EU
Engagement, architecture and delivery leadership. The entity most UK and EU contracts are written against, with data handling arranged for EU clients at contract.
Vyrsa logic Ltd
Company registration number: 11383928
India
Engineering, quality and delivery capacity, and the entity for Indian clients. Not a back office: the same engineers, in the same repository, under the same review.
Vyrsa Logic Pvt Ltd
CIN: U74900MH2013PTC251380
Delivery leadership has previously held combined COO and CTO responsibility for a 40-person engineering and delivery team.
Next
Send this page to whoever will review our work. Then show us what needs to work.
Show us what needs to work →United Kingdom · European Union · India