Five ways in for AI applications and automations, from due diligence to running them, each with an agreed scope, a written acceptance test and a named person accountable for the work. For consulting firms and integrators delivering under their own name, and for teams with AI work to take further.
Start from where your AI work is today, or check the code before you commit. Each package is priced, scoped and accepted the same way, and you can stop after any of them.
IdeaLive
One specific workflow delivered end to end. AI agents take on tasks one at a time, under set authority, with a named person accountable for every action.
Accepted when The workflow runs on agreed cases with agreed results, and every agent action is logged.
PilotLive
A pilot that demoed well, rebuilt to run for real: a locked scope, tests and evaluations, a security review, and deployment to the environment you choose.
Accepted when It passes the acceptance test written with you before the start: what must run, on which data, with which result.
StrugglingStable
An application struggling in production, reviewed first, with what is wrong listed in order of risk, then fixed in prioritised work packages.
Accepted when The review is a fixed-price first package. Each fix after it has its own agreed test.
LiveOwned
After go-live we keep it running and improving under an agreed support arrangement, or hand over the code, the documentation and people ready to own it.
Accepted when Service levels, or the handover checklist, are agreed at the start, not at the end.
Any stageDecision
Before you buy, fund, take over or fix it: the code, the architecture, security, open-source licences and who really holds the knowledge, evidenced and costed. An Express Audit of a live AI application takes about a week; full technical due diligence takes ten business days from access.
Accepted when The agreed reports and the readout are delivered. Never a minimum number of findings, and never a promise that nothing is wrong.
What is in and what is out, in writing, before anyone builds.
What must run, on which data, with which result, and who signs it off.
Agents work under set authority; a person reviews what they produce.
After each package: continue, change course, or stop.
We answer for the agreed scope and its acceptance test. A prototype shows an idea works on a few cases; production needs tests, evaluations, security, monitoring and an owner. We say which one you are getting before we start.
Buying a software company, funding one, taking over an application or about to fix one? Find out what the code really is first: evidenced, confirmed by a qualified reviewer and costed to fix.
For an AI application that is already live. It is also the usual first step of Stabilise.
For a platform you are buying, funding or taking over. From access to readout it runs alongside the wider deal work, so both land together.
System design and boundaries, how modules and services fit together, data and integration flows, and drift from the intended design.
Complexity, duplication, coupling and test coverage, with technical debt costed in person-days rather than described.
Application security, secrets, infrastructure, containers and pipelines, reviewed against OWASP ASVS and mapped to CWE and OWASP references.
Every direct and transitive dependency, known vulnerabilities, licence classification and copyleft exposure in the product.
Who holds which parts of the code, the bus factor per module, code written by people who have left, and the contractor and in-house split.
Test strategy and gaps, build reproducibility, pipelines and releases, and readiness to run in production.
Every finding is confirmed by a qualified reviewer before it reaches the report; AI does not set the final severity. An assessment is not a penetration test or a legal opinion on licences, and both can be scoped separately.
Senior engineering leaders at the Vietnam delivery partners we work with. They own architecture, agentic engineering practice and the review gates on AI-generated code.
Delivery Lead
18+ years as CTO, Director and Founder. Leads a Vietnam-based delivery firm trusted by Fortune Global 500 companies, shipping enterprise-grade systems across 15+ countries.
Chief Technology Officer
Runs engineering around a structured 9-stage delivery process (planning, four-layer testing and a signed closure report on every project), built over years of shipping software for international clients.
LinkedIn
Senior Agentic Engineering Lead and Architect
Senior architect leading agentic engineering delivery: system design, spec-driven build, and review gates on AI-generated code.
CTO and Co-Founder
RWTH Aachen graduate based in Germany. Co-founded and leads engineering at a Vietnam-based delivery firm, bridging European client relationships with Vietnam delivery for Fortune Global 500 companies.
299 projects in our library match this kind of work. Three examples, anonymised:
Generative & agentic AI
Generative & agentic AI
Generative & agentic AI
Every package is written down the same way before anyone starts. An illustration, not a client project:
Example: a document-processing pilot taken to production
Straight answers. Ask us anything else on a call.
We commit to the agreed scope and its acceptance test, in writing. What the application does for the business afterwards also depends on how it is adopted and used, so we do not promise outcomes beyond the scope.
Yes. A stabilise engagement starts with a fixed-price review of the code, the infrastructure and any evaluations, before a single fix is priced.
Task by task, under set authority. Each agent has a defined job and a named person accountable for what it does, and its output is reviewed before it is accepted. The level of autonomy is agreed per task, not assumed.
No, unless the owner approves it in writing. The assessment can run inside their own VPN, virtual desktop or private cloud, access is read-only and can be withdrawn at any time, and copies are deleted at close-out with written confirmation.
Neither. It reviews the source code, architecture, dependencies and how the team works. Penetration testing, a cloud configuration audit and a legal opinion on licences or IP can be scoped separately.
You do, when you are the prime. We contract with your firm, and your client sees your firm throughout.
With Nessoris when you want an agreed scope delivered, and run if you choose. If you need named engineers inside your own team, directed by you, that is our sister company vitX.
An idea, a pilot, or something live that needs fixing. Tell us where it is, and we will tell you which package fits.
