--- title: Recent projects | AI Cost & Quality Engineering | Luna Nova description: "Three recent projects: an AI Software Factory, a sales AI agent and an AI product strategy engagement, with a focus on cost, quality and verifiable results." --- [Skip to content](https://lunanova.ai/en/projects#main-content) [![Logo (28)](https://lunanova.ai/hubfs/Logo%20(28).svg "Logo (28)")](https://lunanova.ai/) - [English](https://lunanova.ai/en/projects) - [Français](https://lunanova.ai/fr/derniers-projets/ai-software-factory) English Show submenu for translations - [Offers](https://lunanova.ai/en/services) - [Method](https://lunanova.ai/en/method) - [Latest projects](https://lunanova.ai/en/projects) - [About](https://lunanova.ai/en/about) - [Contact](https://lunanova.ai/en/contact) Open main navigation Close main navigation - [Offers](https://lunanova.ai/en/services) - [Method](https://lunanova.ai/en/method) - [Latest projects](https://lunanova.ai/en/projects) - [About](https://lunanova.ai/en/about) - [Contact](https://lunanova.ai/en/contact) - - [English](https://lunanova.ai/en/projects) - [Français](https://lunanova.ai/fr/derniers-projets/ai-software-factory) English Show submenu for translations - [Discuss your AI project](https://calendly.com/luna-nova/30min) [Discuss your AI project](https://calendly.com/luna-nova/30min) Projects # Recent projects Three projects, one standard: link real business needs to efficient, reliable and measurable AI systems. [Discuss your AI project](https://calendly.com/luna-nova/30min) ## 01. AI Software Factory: Harness & Loop Engineering **An agentic software factory used every day to build and improve our own projects.** ### The challenge: speed up delivery without losing control Coding Agents speed up execution, but multiplying agents, tools and loops guarantees neither accepted code nor controlled cost. We designed an AI Software Factory that lets the Product Owner move a project forward to pre-production, with explicit decision and validation steps. ### Our approach: specify, build, observe and compare The Factory orchestrates **nine specialised agents** covering the discovery, specification, planning and development phases and reviews through to deployment checks. Before the plan is written, the Product Owner can review **a clickable prototype and the product flow maps** to check the user journeys and the acceptance criteria. ![Project illustration](https://lunanova.ai/hubfs/SF-02.jpg) Clickable prototype reviewed by the Product Owner before planning. Execution traces, the **TDD**, quality and security reviews and a **cost / quality / autonomy benchmark** feed continuous improvement. Combinations of models, skills, delegation and loops are re-evaluated when new generations of models arrive, so that only the genuinely useful steps are kept. ![Project illustration](https://lunanova.ai/hubfs/SF-03.jpg) Flow map and benchmark view. ![Project illustration](https://lunanova.ai/hubfs/SF-04-outil.jpg) The Software Factory at work on the Sales Agent project: the Product Owner validates the example map before the next phase. ### What this project demonstrates A Software Factory is not just about writing code faster. Its performance is measured by its ability to **deliver an accepted task, with the expected level of quality and security, at a relevant cost**. This experience directly feeds our audit and benchmarking work on Coding Agents for clients. [See the related offer](https://lunanova.ai/en/services#ai-software-factory) ## 02. Sales AI agent: optimise processing and secure responses **Less unnecessary processing, more control over decisions and the quality of responses.** ### The challenge: an agent too generalist for a specific need The first sales processing agent relied on extended reading of the mailbox and a generalist model in charge of a large number of tasks. This setup multiplied the tokens consumed and needlessly widened permissions and the risks linked to incoming content. ![Project illustration](https://lunanova.ai/hubfs/SA-02.jpg) Example of an incoming request, anonymised. ### Our approach: reserve AI for what brings it value After an audit, we redesigned the architecture around **deterministic response templates** for known situations. Specialised AI components step in to recognise a situation, choose the relevant elements and word the variable passages. Permissions are better circumscribed and new cases enrich the system **after human validation**. ![Project illustration](https://lunanova.ai/hubfs/SA-03.jpg) Quote request processing interface. ### What this project demonstrates Optimising an agent sometimes depends less on choosing a new model than on **simplifying its role and its interactions**. This approach reduces superfluous processing, makes quality more predictable, limits the exposure surface and allows autonomy to progress within a controlled framework. [See the related offer](https://lunanova.ai/en/services#ai-margin-engineering) ## 03. CSR startup: linking product strategy to AI agent evaluation **From business need to a prioritised MVP and concrete criteria to evaluate the agent.** ### The challenge: choose a use case that brings demonstrable value A CSR startup was developing an AI agent rich in possibilities, but needed to define the right target and the right MVP scope. The challenge was to weigh market needs, the data actually available and the product's constraints, rather than build more features without a clear priority. ### Our approach: a product decision backed by the field We structured a **Product Discovery with professionals**, analysed the needs expressed, compared the possible directions and recommended an MVP direction based on user value and feasibility. This approach led to reasoned priorities and a validation roadmap. ![Project illustration](https://lunanova.ai/hubfs/AS-02.jpg) MVP trade-off, anonymised diagram. The needs identified during the discovery were also **translated into an LLMOps evaluation framework handed to the CTO**. Business situations thus become test cases and quality criteria to guide the evolution of the agent, with a more rigorous reading of the efficiency of its answers. ![Project illustration](https://lunanova.ai/hubfs/AS-03.jpg) From business need to LLMOps tests. ### What this project demonstrates AI Product Leadership does not stop at choosing features. It can also give technical teams **a testable definition of the expected value**, in order to link product choices, quality of results and engineering decisions. This continuity is the link between our product leadership expertise and our AI Cost & Quality Engineering approach. [Understand our cost and quality benchmark method](https://lunanova.ai/en/method) [See the related offer](https://lunanova.ai/en/services#direction-ia-temps-partage)   Contact us Describe your context and challenges. We will get back to you to clarify your needs.   [linkedin-in icon](https://www.linkedin.com/in/rikdams/) [X Twitter icon](https://twitter.com/Rik_Dams) [instagram icon](https://www.instagram.com/agenceseo.green/) - [Offers](https://lunanova.ai/en/services) - [Method](https://lunanova.ai/en/method) - [Latest projects](https://lunanova.ai/en/projects) - [About](https://lunanova.ai/en/about) - [Contact](https://lunanova.ai/en/contact) --- ![green-seo-reforest-action](https://seotraveler.com/hubfs/green-seo-reforest-action.png "green-seo-reforest-action") ![agence-seo-1-pourcent-pour-la-planete](https://seotraveler.com/hubfs/agence-seo-1-pourcent-pour-la-planete.png "agence-seo-1-pourcent-pour-la-planete") ![agence-seo-eco-conception](https://seotraveler.com/hubfs/agence-seo-eco-index.png "agence-seo-eco-conception") ![green-seo-carbon-negative-logo](https://seotraveler.com/hubfs/green-seo-carbon-negative-logo.png "green-seo-carbon-negative-logo") ![green-seo-unesco](https://seotraveler.com/hubfs/green-seo-unesco.png "green-seo-unesco") [Legal notice](https://lunanova.ai/mentionslegales#english) - [Privacy policy](https://lunanova.ai/confidentialite#english) Expertise: [LLM cost optimization consulting](https://lunanova.ai/en/llm-cost-optimization/consulting) - [AI FinOps consulting](https://lunanova.ai/en/ai-finops/consulting) - [Coding agent cost optimization](https://lunanova.ai/en/coding-agents/cost-optimization) Made with ♥ in Bordeaux and Paris, France Copyright © 2026 Luna Nova