Program Intelligence

Straight answers

The questions executives actually ask, answered without the hedging. What Program Intelligence is, where it parts company with the tools you already run, and what it changes about a decision you have to sign your name to.

What is Program Intelligence, in one line?
It is the discipline of separating what your organisation actually knows from what it only believes, before it commits, and being able to prove which is which. A dashboard tells you the work is moving. Program Intelligence tells you whether it is moving towards what you agreed, and hands you the evidence to defend the call.
AI lets our teams ship faster than ever. How do we know the speed is worth the spend?
AI changed the currency of execution. Code is committed at the speed of the token now, and the volume of activity has never been higher. But volume is not value. A programme can produce more, faster, and still be drifting from what you agreed, only now it drifts at machine speed. Program Intelligence tells you whether that fast output is actually advancing what you agreed, so you can judge whether the spend is still buying you the outcome. And when the work cannot even be shown to serve the commitment, it says exactly that, instead of guessing. The tokens got cheap. What you are building with them did not.
Is Program Intelligence like Palantir?
Palantir acts on your data and recommends the move. Program Intelligence does the opposite. It leaves the decision with you and makes sure you can defend it, because every reading is built on evidence only and shows the evidence behind it. You keep the call. It gives you the ground to stand on.
Palantir’s power is its ontology. What is yours?
Theirs is drawn by hand, and someone has to keep it drawn. An authored map of the enterprise ages the moment reality moves, and it has no way to tell you it has aged. Ours works the other way round, and it comes down to one word: derived, not drawn. Program Intelligence builds its map of your software execution estate from the estate’s own evidence, its code structure, its runtime, its delivery and programme state, its documentation, and regenerates it at every observation, with the date it was taken on its face. An authored map ages in silence. A derived map cannot hide its own staleness, and where the evidence runs out it shows you the edge instead of carrying last year’s answer forward. We do not model your enterprise and hope it stays true. We read it, and we tell you where we cannot yet see.
We don’t want a system that tells our teams what to do, or scores our people. What is your posture?
Then this is built for you. Program Intelligence frames what is happening and what it means for the decision in front of you. It does not rank your next steps, prioritise remediation, or tell you what you should do. It never scores a person, pins a slip to a name, or guesses at a team’s motives. That is not a setting we leave switched off. It is a line the system cannot cross, by constitution, whatever anyone asks of it. The safest system in the room is the one that cannot act.
Our observability tool already feeds our exec dashboard. Isn’t that the same thing?
No, and the difference is structural. On a dashboard, every panel is one tool speaking straight to leadership, and the loudest instrument wins. Program Intelligence lets no single source own the executive answer. Readings are composed, disagreement between them is shown rather than averaged away, and each is ranked by how it was obtained: measured stands above reported, and reported above what was merely declared. A dashboard is the un-composed case. Here, bypassing that is not discouraged. It is impossible by construction.
Why do our dashboards stay green while the programme slips?
A dashboard measures motion, not direction. It shows work is happening, not whether it is happening towards what you agreed. A programme can move fast for a year and never once be told it is moving away from the plan. That gap has a name: Execution Blindness.
What does Program Intelligence offer that tools like SonarQube, CAST, Dynatrace or Elastic do not?
Each of those measures one slice of the machine, and measures it well: code quality, architecture, runtime, logs. None can tell you whether the programme is building what the board agreed to fund. Program Intelligence works a level up. It reasons from the programme’s own evidence, its code structure, its delivery and planning systems, its runtime, its documentation, and answers the one question a point tool cannot: is the work still heading towards what was agreed, and can you prove it.
Every vendor shows us a health score. Why should we trust yours?
Because you can take it apart. Point at any reading on the question you asked, any movement, any band, and it opens into the evidence underneath, down to what was observed and where. A health score ends the conversation at a methodology paper. This one ends it at the evidence. We do not ask you to trust the number. We hand you the means to challenge it.
Our plan says green, our engineers say trouble. Which one is lying?
Probably neither, and that is the useful part. Program Intelligence reads the plan, the delivery activity, the code, the runtime and the documentation independently, then looks for where they disagree on the question you asked. Green plan, stalled work, a scope promised but never moved: the disagreement itself is the finding, with each side’s evidence attached. Most tools merge every source into one store first, which quietly destroys the independence that makes the contradiction visible. You cannot triangulate from a single merged voice. This does not merge. It compares.
What does it do when it doesn’t know?
It tells you, plainly. Where the evidence is not there yet, it says so and marks the reading open, not settled. Where something cannot be grounded, you get exactly that, with what was examined to reach it. A copilot fills the gap with fluent text. A dashboard shows a stale panel and no sign of what to trust. Being told what cannot yet be established, and what was looked at, is the rarest thing an intelligence system can give you. It is the difference between an answer and a guess.
Why Program Intelligence if we have open-source and AI?
Open-source models and AI are the raw material. They give you the power to produce a confident answer in an afternoon. What we built is the governed stack around them: the evidence under each claim, the composition that holds it together, and the seal that lets it be reassessed. It is model-invariant by design, so you can swap the engine and the governance still holds. The model curates the evidence. It never gets to decide. Raw intelligence keeps getting cheaper. Being able to stand behind it does not.
You use commercial AI models. When the model changes, does our intelligence change with it?
No. The reasoning models are replaceable instruments. What decides your answer is the governed spine around them, and that is ours and it is fixed: the same evidence and the same rules produce the same conclusions, whatever the model. The words a model chooses may vary. The finding underneath does not. A model is swapped as a certified act, on the record, never a silent bump. In a copilot, the product is the model, so every quiet upgrade changes every answer and nothing exists to notice. Here the model curates the evidence. It never gets to decide, and the sealed record is the tripwire if anything tries to.
Do we still need enterprise architecture now the framework era is ending?
Enterprise architecture answered what the enterprise looks like. What decides outcomes now is what it actually knows, and whether that is true, at the moment it commits. A blueprint could answer that only if it were redrawn at the speed of the token. Program Intelligence stays current because it works from evidence, not a drawing.
Six months ago your product told us the platform was stable. It wasn’t. What do we do with that?
You put it back on the record, and that is the point. An answer Program Intelligence puts on the record is sealed with the evidence it stood on and the moment it was given, so it does not quietly recompute itself the way a dashboard does. The old answer is still there to be held to account. When the ground changes, you reassess that exact answer and get a verdict: whether it still stands, or whether it only held on what was known then. A tool that never seals an answer can never be shown to have been wrong. This one can, and that is what makes it worth trusting.
What happens when the person who understood everything walks out?
In most organisations the understanding leaves with them. The records stay, the reasoning behind them goes. Program Intelligence seals each answer with the evidence and the reasoning behind it at the moment it is made, so the part that usually walks out the door on someone’s last day stays on the record.
What does Execution Blindness actually cost?
The most expensive number in a company is the distance between what it agreed to build and what it is building. It is where budgets are spent that no one decided to spend, and it stays invisible until the deadline it beats you to.
We move fast. Why slow down to govern?
Program Intelligence does not slow you down. It keeps your speed honest. Move as fast as you like, as long as you can still see you are moving in the right direction and prove it. That is the difference between fast, and fast towards what you agreed.