01Mission Control ]


Dayo Kelani operates at the seam between business intent and running software: business analyst, product and project lead, and AI systems builder. This site is a working map of how that thinking runs.

Business AnalysisProduct ManagementProject ManagementAI Systems
3AI productsLameda, Yourly, and agent systems in production
₦3B+Monthly revenue platformEasyBuy consumer finance, delivered end to end
50+Projects deliveredSuccessfully delivered and handed off
80%On-time deliveryAcross regulatory, legacy, and executive workstreams

02Operating Principles ]

How the system decides

The system is the product
001

The system is the product

A feature that ships without its operating model is a liability. Every delivery includes the process, the controls, and the people who keep it running after go-live.

002

Decisions beat documents

Requirements exist to force decisions: what is in, what is out, and who owns the trade-off. If a document does not close a decision, it is overhead.

003

Constraints are design inputs

Regulators, legacy platforms, and immovable deadlines are not obstacles to the design. They are the design. The best architectures in banking were shaped by what could not move.

004

AI is leverage, not magic

Agentic systems earn their place by removing real operational load, with deterministic guardrails, human oversight, and a measurable before-and-after.

03Career Journey ]

The road that built the operating system

04Systems Thinking ]

One operating loop, four phases

Frame

01 / Frame

Turn intent into decisions

Discovery workshops, stakeholder maps, and structured analysis reduce a business ambition to the decisions that actually matter: scope, risk appetite, and the definition of done. The output is alignment, not paperwork.

02 / Architect

Design inside the constraints

Regulatory mandates, legacy cores, and vendor realities shape the target architecture. Event-driven patterns decouple what must be fast from what must be careful, so the system absorbs load instead of dropping it.

03 / Deliver

Run the machine that ships

Governance cadences, risk registers, cutover runbooks, and vendor coordination move the design into production: big-bang cutovers with zero customer disruption, compliance treated as a gating condition rather than a checklist.

04 / Operate

Measure, learn, compound

KPI frameworks defined before go-live make the outcome legible: availability, session growth, cycle time. What the metrics teach feeds the next frame, and the loop compounds.

05Portfolio ]

Decisions under load

Knowledge Graph
01

Stanbic IBTC: USSD Channel Re-Platforming

The *909# channel ran on a legacy menu manager that had been retired by its vendor and was no longer supported, creating live security exposure. It caused frequent downtime, capped scalability, made shipping new menu options slow, locked the bank into a single vendor, and drove a high session-timeout and failure rate at growing cost.

200% Volume Growth Post Go-Live99% Availability60% Faster Releases

Rebuilt on event-driven containerised microservices with Kafka as the messaging backbone. 200% increase in monthly session volume post go-live. 99% availability restored. Platform-side session response time cut by up to 90%. Feature release velocity up 60%. Delivered as a big-bang cutover under enforced regulatory, security, and compliance controls.

08

Lameda: Commerce Infrastructure for Nigerian Merchants

Nigerian merchants run multi-million naira businesses through WhatsApp voice notes, manual bank transfers, and Excel sheets. Every Western SaaS tool was built for a different market.

Live at lameda.com.ng₦1.58 AI cost per message15+ features shipped

Live at www.lameda.com.ng. 13+ sprints shipped: AI conversation engine, direct merchant payments, human handoff, official WhatsApp and Instagram channels, consent-first marketing, smart lead follow-up, subscription billing, and a full merchant dashboard.

10

Trading Intelligence Agent: NGX Equity Research

NGX equity research is scattered across broker emails, PDF reports, analyst houses, and macro feeds. Pulling it into one coherent weekly view by hand is slow and inconsistent, and there is no single synthesised briefing that tracks how the picture changes week to week.

Accuracy-first: numbers reconciled in code20+ research sources7-section analyst report

An autonomous Sunday pipeline that gathers and cross-checks 20+ sources, backtests its picks against the ASI, compares week over week, and posts a 7-section analyst briefing to Slack. Accuracy-first by design: the model proposes, but every price, fundamental, and macro figure is reconciled in code against authoritative feeds, and anything unverifiable is flagged as an estimate. Cost engineered to roughly $1.72/month.

06AI Systems ]

Running processes

Open the AI lab

08Writing ]

Field notes on delivery, systems, and applied AI, written from inside the work rather than about it.

Get in Touch ]

Let's build something
worth building.

Whether you have a system to build, a problem to solve, or a role to fill, choose the fastest path.