Matti Kakkori

As head of AI product at Agilitas, I led the strategy, development, and launch of a novel agentic AI platform designed to transform slow, manual research and development processes. The core challenge was integrating cutting-edge large language models (LLMs) and retrieval-augmented generation (RAG) techniques into a robust, enterprise-grade solution. I directed a cross-functional team of engineers and data scientists, defining the platform architecture, which featured advanced semantic search capabilities powered by vector databases. We established a data-centric approach, building proprietary evaluation frameworks and a "data moat" to ensure high-quality results and competitive differentiation. I aligned the technical roadmap with user needs and business strategy, integrating AI capabilities seamlessly into the product UX/design. The platform accelerated key user workflows by 50x, providing significant value and strengthening the company's market position.

As chief product and technology officer at Rehinged AI, I architected and led the development of the Carenet.AI platform, the company's flagship product. The platform was designed to ingest, process, and analyze complex healthcare data using advanced AI and natural language processing (NLP) techniques to provide actionable insights for enterprise customers. I defined the technical vision and scalable architecture, building and leading the product and engineering teams (growing from 4 to 20+). Key responsibilities included designing the data pipelines, ensuring scalable infrastructure (likely cloud-based), incorporating relevant NLP models, and establishing product metrics for business outcomes. This platform rapidly gained market traction, scaling to over 40 enterprise customers and $1 million in ARR within 18 months, ultimately becoming responsible for 80% of the company's revenue. I also implemented a technical sales methodology, contributing to a 90% close rate.

I led the conceptualization and development of an innovative MVP for a medical AI startup, addressing the challenge of unlocking knowledge trapped within vast libraries of healthcare video content. The goal was to move beyond simple keyword search. I architected an agentic AI solution featuring an automated transcription pipeline to convert speech to text. This text was then processed for semantic understanding, likely leveraging vector embeddings stored in a vector database. We integrated large language models (LLMs) to rank search results based on relevance and context, providing significantly more accurate and valuable results than traditional methods. This MVP successfully demonstrated the feasibility of applying advanced AI techniques to make video content searchable at a semantic level, driving successful pilot programs and informing future product direction.

The project centered around building a complex AI application utilizing GIS and mapping technologies for a startup. A complete end-to-end strategy and development project that included:
• Developing a well-researched customer-needs-driven product strategy.
• Coding a quick and dirty mockup to test assumptions with lead customers.
• Specifying and leading the development of an MVP version of the software.
• Creating product marketing materials and launching the product at a large industry trade show.
• Leading the technical sales for the product.
• Running engineering and building a structured agile development process for further improvements.
• Through multiple architecture and technology improvements, reaching v4 of the software and 40+ B2B customers.

Some basic technologies used in the project are AWS, Laravel, Django, Mapbox, ESRI, JIRA, Python, and Jupyter.

Led the product definition for an AI/NLP-driven software product that can ingest a considerable amount of textual data, analyze and enrich it, and then provide detailed insights into people's and markets' perceptions about brands and products.

The software product is delivered as a SaaS product.

The product portfolio comprised 20+ devices that were very successful overall and generated $5+ billion in revenue annually. However, the marketing message and effectiveness of the product development suffered from a too-complicated offering.

I created a thorough market and consumer needs-driven analysis on how to restructure the product portfolio to be more balanced and more accurately capture the market volume and demand.

The strategy was adopted and implemented as part of a larger business unit restructuring.

As part of my advisory work at Galaxy & Sons, I led a confidential project to design and prototype a sophisticated chatbot specifically for the Japanese market. The key differentiator was its emotional intelligence and persuasive capabilities requirement, moving beyond standard transactional interactions. I architected the system to effectively store and leverage structured user data, enabling the chatbot to maintain context and personalize conversations dynamically. This required careful consideration of both the AI technology and the back-end data architecture needed to support truly dynamic and context-aware interactions.

Foldpass is a simple turn-based game in which users can write short text passages and invite friends to complete a three-line haiku poem.

The result can then be made into a well-designed poster through an embedded poster editor with downloadable poster packs.

I led the design, development, and launch of the application.

The architecture was a Phonegap-based iOS application connecting to a Django back end.

Led product strategy, design, and development for a mobile screen insurance solution tailored to the Mexican market. The cross-platform app, built with Apache Cordova for iOS and Android, was supported by a Django/Bootstrap backend and customer-facing web interface. I created detailed wireframes and UX flows, and directed the engineering effort in collaboration with an external development team.

Dsider was an idea of a decision-making platform with a social component. Users could enter criteria that are important in their decision, weigh them, and also include friends' opinions. The weighted algorithm then recommended the choice.

I ran the UX and product development for the application and managed the off-shored development of the back-end server and the mobile application.

Earlier the company development was more ad-hoc and focused on heads-down coding.

I established an agile development process when the team started growing. The development team adopted the system of 2-week sprints with well defined epics, user stories and tasks in JIRA.

The implementation of the agile process sped up the development of new features and products. (Of course, hring a larger team didn't hurt either. ;-) )

The project scope was to design a long-term plan (2-5 years) for a leading smartphone experience for the US market.

The project team was huge with participants ranging from executive office to sales and marketing to HW engineering.

The project considered various inputs from consumer and technology trends to competitive M&A transactions and developed HW and SW prototypes and mockups.