Sovereign by architecture,
not by promise.
Seven commitments that govern how we build, what we sell, and what we refuse. Each one is stated with what it costs us — because a principle with no cost attached is only marketing.
Most enterprise software asks you to trust a contract. A clause about where your data sits. A promise about who can read it. An assurance that the model is fair. Every one of those is a promise made by a company whose interests are not yet aligned with yours, enforceable only after something has already gone wrong.
We think the contract is the wrong place to put the guarantee. If your data never leaves your device, no clause is needed to keep it there. If every score shows its inputs, no assurance about fairness is required. If you can export everything and walk away in a minute, lock-in cannot be exercised against you.
These are not features. They are the consequences of decisions we made early and will not reverse — even where reversing them would be commercially convenient. Especially there.
Your data does not move, because there is nowhere for it to go.
Our tools run in your browser or on your premises. GVS operates no server that holds customer data. Not encrypted-at-rest on our infrastructure. Not processed transiently. None. The jurisdictional question — which legal system can compel access — does not arise, because there is no GVS-held copy for any authority to reach.
Any architecture where GVS becomes a custodian of client data, however well secured or however profitable the telemetry would be.
No usage analytics, no telemetry-driven roadmap, no recurring hosting revenue, and no ability to recover data a client loses. We accept all four.
If a score cannot show its inputs, it does not ship.
Every number our tools produce is computed deterministically and traceable to the inputs that produced it. Where a language model is involved, it explains figures already calculated — it does not invent them, and it cannot change them. An AI that cannot be audited has no place in a hiring decision, a launch gate, or a regulatory filing.
Black-box scoring, opaque ranking, and any feature whose output we could not defend line by line to an auditor.
Slower feature delivery, and demos that look less magical than competitors who let the model decide.
Unfinished work is labelled unfinished.
What is live says live. What is roadmap says roadmap, on the page, next to the thing itself. We do not publish a capability because a prospect wants to hear it. We would rather lose a deal to a competitor's brochure than win one on a claim we cannot open in a browser right now.
Roadmap presented as product. Pilots described as deployments. Concepts illustrated as screenshots.
A visibly shorter capability list than firms who count intentions. Some buyers read the honesty as a smaller company. It is.
We publish no number we cannot evidence.
Every figure we state carries its source. Delivery figures from prior roles are labelled as prior roles, with the employer named — never blended into GVS corporate results. We do not publish maturity scores of our own, satisfaction percentages we did not measure, or ROI multiples we cannot trace to a client's books.
Invented benchmarks, unsourced percentages, decimal places used as a substitute for evidence.
Our materials are quieter than the category norm. We cannot answer "what ROI do you deliver?" with a number, and we will not start.
Leaving must be as easy as arriving.
Every tool exports its full state in a format you can read without us. There is no proprietary container, no migration fee, no wind-down period, no data held pending final invoice. Retainers cancel at the end of any cycle with no notice penalty. If we are worth keeping, we should have to earn it every cycle.
Lock-in of any kind — contractual, technical or commercial. Multi-year minimums. Exit fees. Data hostage.
No forward revenue certainty. Every client is permanently at risk of leaving, which is the point.
We claim depth only where the credentials sit.
Telecom OSS, autonomous networks and programme governance are where twenty-five years of Tier-1 delivery actually happened. That is where we compete. Elsewhere we are a capable generalist and will say so. We would rather decline work than be discovered mid-programme to have been learning on a client's budget.
Bidding into domains we do not know. Positioning a generalist as a specialist. Winning on a claim of coverage.
A far smaller addressable market than a firm willing to say yes to everything.
We say what we are not.
We are founder-led, with no delivery bench, no 24×7 desk and no operations centre. Key-person risk is real and total. We hold no infrastructure certification, because we hold no infrastructure. Every buyer learns this from us on the first page, not from a reference call on the last.
Selling capacity we cannot staff. Response-time SLAs a single practice cannot honour. Implying a bench that does not exist.
Disqualification from procurement processes that require scale. We would fail those anyway — better at the start than after signature.
Judge us on what you can open, not on what we assert.
Every tool is one click away and runs without an account. Every claim on this site can be checked against something that exists. If you find a statement we cannot evidence, tell us and we will remove it — that is the whole basis on which we ask to be trusted with anything larger.
See the portfolio →