23 October 2026 · Research brief
The memory wall in LLM reinforcement learning
LoGRA’s low-rank gradient sketches show why post-training memory — not another demo — decides who can improve a reasoning model.
Read full brief →Insights
We write for CIOs, CTOs, and programme leaders — on AI, GenAI, blockchain, and the core platforms those investments depend on. Perspective first. Implementation when it is earned.
23 October 2026 · Research brief
LoGRA’s low-rank gradient sketches show why post-training memory — not another demo — decides who can improve a reasoning model.
Read full brief →
30 July 2026 · Client story
An IT services organisation redesigned architecture and delivery around production outcomes — and restored trust in the work.
Read full story →
14 July 2026 · Industry brief
From use-case counsel to evaluation and MLOps, the work that puts generative AI into production is not interchangeable.
Read full brief →
21 June 2026 · Industry brief
AI and blockchain take the headlines. Cloud, cybersecurity, data, and platform engineering still decide whether programmes endure.
Read full brief →Outlook 2026
Leading estates now place AI, generative systems, and selected blockchain use cases on the same operating model as software, cloud, and security. Those who prevail will integrate those domains without confusing the architecture or diluting delivery culture.
Our outlook examines scarce capabilities, GenAI in production, distributed systems, and platforms operators can actually run — written for people accountable for the estate.
Request the briefing