What this platform will not do
Most software is sold on what it can tell you. This page is about what MeridianCogent deliberately refuses to tell you, and why those refusals are the most useful thing about it.
Every piece of M&A software eventually faces the same commercial pressure: a customer asks a question the system cannot honestly answer, and answering it anyway would be easy, impressive, and wrong.
Produce a risk score. Estimate what is left to spend. Tell them what the notice period is in Germany. Each of these is a small step, each is individually defensible, and each moves the platform from recording what people decided to asserting things it has no basis to assert.
The constraints below are how that pressure is refused. They are design positions, not gaps waiting to be filled.
It will not tell you what the law requires
MeridianCogent does not maintain a corpus of jurisdictional rules. It does not tell you what a consultation requires, what notice period applies, what a regulator will want, or what your obligations are in any country.
Where jurisdiction-specific content exists in a programme, it was entered by the customer or their advisor, and formally adopted by a named person who takes responsibility for it. The platform records that decision. It does not make it, suggest it, or validate it.
This applies to the public tools as well. The readiness assessment on this site makes no jurisdictional claim of any kind — not even a hedged one, not even with a note to confirm with counsel. A hedge changes the wording, not who made the claim.
Two capabilities are held back for exactly this reason: legal entity step plans and regulatory approval tracking. Both are commercially attractive. Both would require the platform to assert what a jurisdiction requires. Neither will be built on that basis.
It will not calculate what it has no basis to calculate
Some numbers look like arithmetic but are not.
A risk register that multiplies likelihood by impact produces a score. High multiplied by medium is not a quantity. The platform records both dimensions and shows them together; it never derives a score from them, and never ranks risks by one.
A cost-to-complete figure derived from burn rate is a forecast, not a measurement. Where a cost-to-complete exists here, a person entered it, and it is recorded as their estimate with their name and the date attached.
An annual run rate and a one-time cost are different quantities. They are never summed. A synergy figure and an integration cost are independently reported and shown side by side — never netted into a single number that implies a precision nobody has.
A difference between a budget and an actual is only a variance if a phased budget exists. Otherwise it is remaining budget, and it is labelled that way.
These distinctions sound pedantic until a board pack built on them turns out to be wrong.
It will not infer what a person should decide
When a perimeter item is entered, the platform does not suggest whether it transfers or is retained. When a contract is entered, it does not infer whether it needs novation, assignment or nothing at all.
These look like helpful defaults. They are positions on contested commercial questions, and the platform has no standing to take them. Disposition is entered by the customer or their counsel. Nothing is pre-filled, nothing is flagged as wrong, and no suggested default appears anywhere.
The same applies to dates. Where the platform can model when something will finish, it shows that. Where it cannot, it says the date is unavailable — rather than quietly falling back to the date you originally hoped for. A target date presented as a modelled date is the most dangerous number in a programme.
It will not quietly overwrite the past
Corrections are made by superseding a record, never by editing one. The original remains visible, with who entered it and when.
When a figure is formally adopted, the platform snapshots exactly what was adopted at that moment. If the underlying figure later changes, the adoption is marked stale — it does not silently update to match. The decision that was actually made, on the information actually available, survives.
This matters most in the situation nobody plans for: eighteen months later, when someone asks why a commitment was made, and the honest answer depends on what was known at the time.
It will not show you a number it does not have
Benchmarks require comparable programmes. Until there are enough of them, there is no benchmark, and the platform will not manufacture one from a thin sample and present it as market context.
The same logic governs everything on this page. An empty state that says “not enough data” is more useful than a confident number assembled from too little.
What this costs you
These constraints are not free, and it would be dishonest to present them as pure benefit.
MeridianCogent asks more of you than a platform that guesses. Someone has to enter the jurisdictional content and put their name to it. Someone has to decide what a perimeter item's disposition is. Someone has to state what a figure represents before it can be stored, and someone has to formally adopt it before it affects a plan.
That is more work than accepting a default. It is the work that makes the output defensible, and it is the work a platform cannot do on your behalf without pretending to a knowledge it does not have.
If what you want is a system that fills in the blanks for you, this is the wrong product, and that is a reasonable thing to want.
One deliberate exception
There is a difference between a forecast the platform volunteers and a counterfactual you explicitly ask for.
MeridianCogent is being built to answer a direct question — move this date, and show me what it does to exit timing and cost. You supply the change, and the answer is computed through the same pricing and scheduling logic that governs the live programme. Nothing is assumed on your behalf; the input is yours, and the output is traceable to it.
That is not a prediction. It is arithmetic you asked for, on inputs you chose.
Early access
MeridianCogent is in development. Join the list for updates as we open access.