01Service · Products
Vibe Coding and AI-Based ProductsAI products the team can maintain.
We map how you build and what the product needs to solve. From there, we define a team standard and, where it fits, build a reusable AI foundation with human review, quality criteria, costs and owners defined.
AI-assisted development follows rules, tests and human review before release.
02Who it is for
Does any of this sound familiar?
- Each person builds with AI their own way, without a standard the team can review and maintain.
- The application is ready, but fixing errors or making changes depends on whoever created it.
- The first product works in the demo, but the next repeats the same retrieval, connections and effort.
- The team cannot check the source of answers, identify failures before use or track cost per task.
When the product becomes part of the routine, the team needs to know how to maintain, evaluate and take responsibility for what it delivers.
03Deliverables
From an isolated project to a product the team maintains.
One track defines the standard for building with AI. The other builds a foundation that products can reuse. Deliverables start with the applications, data and work already in place in each area’s operations.
You can engage either track or combine both, depending on the assessment.
To define how the team will build
The assessment shows what exists and what needs to change. We define a standard that the technical owner approves before implementation.
- Map of applications, repositories and owners
- Assessment of the team's capacity to review, test and maintain
- Business rules and application objectives
- Standard for development, testing and release to production
- Instructions that guide AI in the team's work
- Responsibilities for building, reviewing and approving
- Plan for implementing the standard, with priorities and limits
To build without starting from scratch
We bring the standard into the team's routine, with checks, a template for new projects and the first application built together.
- Reusable template for starting new applications
- Rules and instructions applied in development tools
- AI assistance for building, reviewing and testing by role
- Checks that block or flag errors before release
- First application built with the team within the approved scope
- Application catalogue with owners and documentation
- Support as the team reviews and maintains the standard
To define what the product needs to solve
Before building, we define the problem, permitted sources and how response quality will be assessed with the people who understand it.
- Product use problem and success criteria
- Information sources and authorized access
- Real questions, expected answers and references
- Design of the foundation the product can reuse
- Required integrations and verified dependencies
- Limits on product use, costs and actions
- Plan approved before building
To run the product with limits and quality checks
We build the foundation and put the first product into use, with controlled access, response evaluation and operational owners.
- Reusable foundation for knowledge, workflows and conversations
- Connections to the systems defined for the project
- Rules for data, actions and human review
- Answers supported by permitted sources, where applicable
- Quality tests before product changes
- Records of use, failures and cost per task
- First product in use, with documentation and an owner
- AI builds
- AI checks
- Human reviews
- Human approves
Checks support review. The product owner approves changes and defines limits on AI actions according to the risk of each use.
04How the work happens
First we define the problem and the standard. Then we build.
The assessment defines what needs to change in development and what the product must solve. You approve the plan and choices before implementation. We build with the team that will maintain the result in the company’s daily operations.
Discuss the scope05Work tracks
A standard for building. A foundation for reuse.
You can engage either track or combine both, depending on the assessment.
AI development to a shared standard
ProjectFor teams already building with AI but finding applications hard to review, maintain or hand over.
Assessment
We map applications and how the team builds, reviews, tests and releases today, identifying recurring difficulties.
Plan and standard
We define development rules and who reviews and approves. The technical owner validates the plan.
Instructions for AI
We apply instructions and limits in the tools the team uses to build, following approved rules.
Checks
We implement checks to identify or block problems before an application is released.
First application
We build the first application with the team and validate use of the template in subsequent projects within scope.
Team handover
We document the routine and support the team as it reviews applications and maintains the standard.
AI foundation for products
ProjectFor products that repeat the same information retrieval, conversation logic and connections without a reusable foundation.
Problem and data
We define product use, permitted information and the questions that guide quality tests.
Design and approval
We define the foundation, connections and limits on actions and costs. You approve before we build.
Building
We build the foundation with the team's engineers, applying access rules, quality checks and human review.
First product
We put the first product into use and validate with the team what subsequent products can reuse.
Delivery and operations
We document operations, monitor quality and costs and hand over each part to a named owner.
06Working together
We build with the team that will maintain it.
Who takes part on your side
The people who know the business define the problem and evaluate responses. The technical team takes part in choices, building and operational handover.
- A leader to approve priorities, scope and investment
- A technical owner and the engineers who will maintain the work
- A product owner and subject specialists
- Owners of the data sources and required access
After delivery
The standard and the foundation stay with you.
Code, instructions, tests and documentation stay with the team. We support the start of operations and leave named owners for maintenance, response quality and product changes.
07Results
What changes in the next product.
Results show when the team can review and maintain what it builds, reuse what works and monitor quality and costs within the approved scope, with owners for each part.
Applications within scope follow a standard the team can review and maintain.
Subsequent products can reuse the parts of the foundation that fit their use.
Responses, actions and changes follow the defined quality and approval criteria.
How we track progress
We record starting measurements when data is available and compare the same indicators after implementation, within the approved scope.
Possible indicators, selected for each project
- Time from idea to an application in use
- Rework and problems found in review
- Applications following the defined standard
- Time to maintain or change an application
- Reuse of the foundation in new products
- Response quality in business tests
- Failures identified before reaching users
- Cost per completed task and response time
08Governance
AI produces. The team is accountable for what goes into use.
We define who can access data, which actions the product can perform and which changes need review. Quality is checked against criteria approved with the team before releasing new versions.
Data and access
The product accesses permitted sources for each use. Personal and restricted information receives the protections defined for the project.
Actions and limits
Each agent or workflow has scope, permitted actions and limits on use and cost. Decisions and approvals stay with the people responsible.
Quality and review
We test real questions and situations before releasing changes. When information is insufficient, the product signals the limitation or routes it for review.
Operations
Records of use, failures and costs help the team monitor operations. Each part has an owner and instructions for handling recurring problems.
09Scope and questions
What we agree before starting.
What is included
The approved plan defines the applications, products, sources and connections to be addressed. The foundation is built for the defined uses and can be reused where it fits.
What we need from you
We need the technical team, subject specialists and authorized access to the environments and data within scope. Model, license and infrastructure costs are presented during planning.
What depends on the assessment
We do not assume every application needs an agent or that any system can be connected. Sources, tools and dependencies are checked before defining implementation.
What happens afterwards
Delivery includes documentation and initial support for the team to take over. Ongoing support, new applications, connections or products can be engaged under a separate scope.
Questions that help you decide
Do we need both tracks?
No. The assessment shows whether to organize development, build a product foundation or combine both tracks. The plan defines what each track will deliver.
Do you build the product for us?
We build with your team in the environments defined for the project. Code, instructions and tests stay with you. Documentation and initial support prepare the team to maintain what was delivered.
Are we locked into a model or tool?
Choices and dependencies are discussed during planning. A change is assessed against use, cost and expected quality, and must pass the tests defined for the product.
How do we assess whether an answer is correct?
We write real questions with your specialists, with expected answers and references where applicable. We test the product against these criteria. Evaluation helps identify failures; higher-risk uses still require human review.
Do we need to replace systems or organize data first?
We check what exists and what the product needs to access. If data quality or access prevents use, that dependency is included in the plan. We do not promise connections before checking feasibility.
How long does it take and what is the investment?
The assessment defines the size of the work and its dependencies. You receive the scope, sequence, timeline and investment for approval before implementation.
10Contact
Talk to us about your next product.
Tell us what you have built with AI, what is hard to maintain and what the next product needs to solve. Giulliano Soares is who answers.
- Speak with the microphone on
- Type whenever you prefer
- The contact is recorded at the end
GiuSo
Ready to talk
Where would you like to start?
Say what is stuck, in your own words.