All field notes

Least privilege for AI agents means granting only the authority a specific task requires, and agents accumulate permissions faster than people because nothing reviews them. This explainer covers scoping by task rather than role, time-bounding agent access, why read and write need separating, and how to review what agents hold.

Open-weight AI means the model's parameters are published for download, which enables self-hosting, fine-tuning, and inspection. This explainer covers what that permits and what it does not, why it is distinct from open source, the operational cost of running your own inference, and when it is the right decision.

When an AI agent is scoped as a Digital FTE, it can be budgeted, measured, and reviewed like a unit of workforce. This whitepaper gives finance and engineering leaders a shared economic model.

Enterprise AI adoption stalls when it is run as a series of experiments. This whitepaper lays out a phased roadmap from readiness through first workflow, platform, portfolio, and AI-native operation, with the decisions each phase requires.

Most RAG systems are built as a vector search bolted to a prompt. Enterprise RAG is a pipeline with ingestion, permissions, hybrid retrieval, reranking, grounded generation, and evaluation. This whitepaper describes the reference architecture.

Forward deployed engineering puts a senior engineer inside the business to own an outcome. This whitepaper is the operating playbook: mission scoping, embedding, discovery, specification, delivery, adoption, and handoff.

AI changes the economics of legacy modernization: code understanding, migration, test generation, and documentation are faster. This whitepaper explains where AI helps, where it does not, and how to run a modernization program that uses it safely.

A demo proves an LLM can do something once. Production requires it to do the right thing reliably, safely, and affordably under real load. This whitepaper defines the readiness criteria and how to verify them.

Aging mobile apps carry technical debt, poor ratings, and platforms that block new features. This whitepaper gives a decision framework and migration strategy for modernizing enterprise mobile apps, including AI-powered experiences.

Enterprise web platforms now carry commerce, customer portals, and AI features on the same foundation. This whitepaper sets out a modern architecture: rendering, APIs, data, edge, security, and operations, and how to migrate toward it.

How to outsource mobile app development the right way—native vs cross-platform, what to verify, cost drivers, and shipping production with an offshore partner.

A guardrail policy is the set of enforced rules constraining what an AI system may accept, produce, and do. This explainer covers input, output, and action guardrails, why enforcement must live outside the model, how to avoid over-blocking, and what belongs in a policy versus a prompt.

A retrieval index is the searchable structure behind grounded AI answers, and it holds considerably more than vectors. This explainer covers what each entry must carry, why update strategy determines freshness, how indexes decay silently, and what separates an index that stays useful from one that quietly stops working.

An agent registry records every autonomous agent an organisation runs, with its identity, permissions, owner, and authority limits. This explainer covers why agents need their own register rather than appearing in an AI inventory, what each entry must capture, and how agent identities escape conventional access governance.

Late interaction retrieval scores documents by matching query tokens against document tokens individually rather than comparing two summary vectors. This explainer covers why that recovers detail single-vector similarity loses, what it costs in storage and computation, and why it usually belongs as a reranker.

Query expansion rewrites or augments a user's query to improve what retrieval finds. This explainer covers the main techniques including hypothetical answers and multi-query generation, when expansion genuinely improves recall, how it introduces drift away from the original intent, and how to measure it.

Semantic chunking splits documents where the meaning changes rather than at fixed character counts. This explainer covers the methods, why structure-aware splitting often achieves most of the benefit more cheaply, what the extra ingestion cost buys, and how to tell whether your corpus needs it.

Tool poisoning manipulates an agent through the tool descriptions it reads or the results tools return, rather than through user input. This explainer covers how the attack works, why shared and third-party tool servers make it practical, what changes when a tool definition is updated silently, and what defends against it.

Procuring AI with software-era processes produces demos, lock-in, and unmet expectations. This whitepaper gives CIOs a procurement approach built for probabilistic systems: requirements, evaluation, RFPs, terms, and post-signature governance.

AI value claims fail when nobody measured the baseline or defined attribution. This whitepaper gives a measurement framework that stands up to finance: baselines, value categories, attribution, indicators, and reporting.

Healthcare organizations want AI for documentation, scheduling, billing, and triage support, and cannot afford a safety failure. This whitepaper maps where AI fits safely in healthcare operations and the controls each use requires.

Inference is the visible cost of AI and rarely the largest. This whitepaper sets out a complete total cost of ownership model for enterprise AI, from discovery and data to oversight, compliance, and the platform that amortizes across use cases.

Buying AI is buying a probabilistic system and the people who will make it work. This whitepaper gives buyers a due diligence framework for AI vendors and partners: capability evidence, method, security, data, IP, commercial terms, and references.

Most enterprise problems do not need a blockchain. Some do: multi-party records where no single party should be able to rewrite history. This whitepaper explains when tamper-evident ledgers earn their place and how to architect them.