Research, in the open.
For universities, research groups and development partners working on how institutions build and govern software now that AI agents do part of the work. Working notes first, then papers and open-source releases.
Agentic-first architecture
What changes in an enterprise architecture when agents are first-class actors: reference architectures, protocol layers such as MCP and A2A, data strategy as agent strategy, and model selection as an architectural decision.
AI-led software delivery and DevSecOps
How a delivery team runs with agentic coding tools: the rules, hooks, commands and purpose-built agents, and the review gates and evidence trail an enterprise will accept.
Digital public goods for education in West Africa
Open, adoptable platforms for education and civic data in low-bandwidth, low-budget settings, drawing on the Sierra Leone Learning Passport and the Sierra Leone Open Election Data Platform.
Earlier research.
Work our founder co-authored before the lab, kept here for the record.
Research co-authored at Pace University, Seidenberg School, in 2015 and 2016 on big-data analytics for telemedicine, ensemble models in histological examination, and a multivariate biomarker for Parkinson's disease. Published as arXiv preprints.
Bring us the question.
A joint paper, a pilot, a dataset or a student project: tell us your institution and the question you want to work on, and you will have a reply within a day.
Get Agentic-First Architecture
Get Agentic-First Architecture, the working notes of our research, written by our founder for architects, CTOs and delivery leads: what changes when agents become first-class actors, how to run an AI-led delivery loop an enterprise will accept, and how to build teams that run without you. Free, a few times a month at most.
Questions people ask
What does TSL Intelligence Lab research?
TSL Intelligence Lab runs three research programmes: agentic-first architecture (how enterprise architecture changes when AI agents are first-class actors), AI-led software delivery and DevSecOps (the guardrails, review gates and evidence trail that let an enterprise accept agent-written code), and digital public goods for education in West Africa. Working notes are published in the Agentic-First Architecture newsletter.
What is agentic-first architecture?
Agentic-first architecture is an approach to enterprise architecture that treats AI agents as first-class actors alongside people. It covers reference architectures in which agents call systems, protocol layers such as MCP and A2A, data strategy as agent strategy, and model selection as an architectural decision. It is a research programme of TSL Intelligence Lab.
How can a university or research group collaborate with TSL Intelligence Lab?
Email tamba@tambalamin.me with the subject line "Research collaboration", your institution and the question you want to work on. Collaborations can take the form of a joint paper, a pilot, a shared dataset or a student project, in agentic-first architecture, AI-led software delivery, or digital public goods for education in West Africa.
Has Tamba Lamin worked with donors, governments and universities in West Africa?
Yes. He contributed to the Sierra Leone Learning Passport with the Directorate of Science, Technology and Innovation, UNICEF and the Ministry of Basic and Senior Secondary Education, reaching 50,000+ students; architected SLOEDP, Sierra Leone's first open election data platform, adopted by the Independent Media Commission and hosted at Njala University; delivered election-violence monitoring platforms for IFES in the Maldives and Senegal; and led digital transformation for Njala University, the University of Sierra Leone and the Government of Sierra Leone's Open Government Initiative.