Why olondi — Tool adoption is not AI value. Workflow redesign is.

Every company optimises against its own assumptions. Hires from the same pool. Reads the same vendor decks. Benchmarks against direct competitors doing the same.

The AI work it ships inherits the same blind spots. An organisation cannot diagnose what it has decided to stop seeing.

The model is good enough. The workflow is not.

Most boards are counting Copilot seats, ChatGPT Enterprise licences, and pilots launched, and reporting it as AI progress. None of those numbers say whether a workflow changed, who now does the work, or what the business can do that it could not do before.

PwC's 2026 study finds 20% of companies are capturing 74% of AI value. The split is not budget. It is whether the spend bought tool access or workflow change. Adoption is an input. Redesign is the outcome. The scoreboard most organisations are using cannot tell the two apart.

Strategy firms diagnose and leave. Dev shops do not question the brief. Cloud vendors build and keep the keys.

Inside the few companies shipping AI at scale, the work is being redesigned by people who understand the workflow and the system at the same time. That is the work we exist to do — the outside view, brought in with the standing to question the brief and the engineering to ship the answer.

There is a strain of AI work that treats more tokens, more agents, and more pilots as progress. It is not. A chatbot speeds up typing. An agent dropped into a messy workflow scales the mess. Another pilot adds another item to the portfolio review.

The work that produces value is slower, quieter, and more boring:

• A named workflow, not a tool category.
• A named owner, not a sponsor.
• A measurable outcome, not a capability.
• A handoff plan, not a pilot.
• An evaluation loop, not a demo.

We design the redesign first, then build the system the redesign needs. Not the other way round.

One team, end to end — assess, redesign, build, hand off. Code and architecture handed over. No vendor lock-in. No follow-on proposal in place of a working system. Every engagement ends with something running in production against a metric the business already cares about.

If the workflow has not changed, we have not done the job. If a company wants more pilots, more tokens, and more frameworks, there are plenty of firms set up to sell that. We are not one of them.

Use AI to change what the business can do, not just how fast people type.