Solutions
Solutions for decisions that must be explained.
We work where decisions are frequent, consequential and reviewed: finance and healthcare first, other regulated work after. For each area this page names the decisions, shows where the two gaps appear and says what changes.
Finance
Reviews that arrive with their reasons.
Credit, payment and compliance work runs on policy and judgement. Reviewers handle many cases, and every case needs an owner and a record.
Illustrative field: learned patterns on the left, rules and judgement on the right, expert intent in amber.
Typical decisions
Credit and risk review
Assess an application, a limit or an exposure against policy and history.
Payment and invoice controls
Decide whether a payment, an invoice or a change of supplier details goes ahead, waits or needs a second look.
Compliance and case reviews
Work through alerts and cases, and document why each was closed or escalated.
- Where the explanation gap shows
- A flag or a score without reasons sends the reviewer back through the file. The result may be right, but a sign-off cannot rest on it, so the review is done twice.
- Where the intention gap shows
- Each team weighs the same policy in its own way: which exposures matter most, when an exception is acceptable, what a second reviewer needs to see. None of that is in the prompt, and none of it is in the model.
- What we deliver
- A reasoning trace for every case: the evidence, the policy clauses applied, comparable cases and a statement of confidence, in the order the reviewer works. Policy is held as explicit rules. Priorities and exceptions are captured as structures that the system reads back to the team, and overrides are recorded with their reasons.
- The difference
Faster sign-off
The reviewer checks the steps that matter instead of rebuilding the case.
Policy applied the same way
The same rule gives the same result for every reviewer and on every day of the week.
Audit-ready records
Inputs, steps, versions and approvals are kept for each decision.
Healthcare operations
Complete records and workable plans. People decide.
We work on the organisation around care: documentation, coordination and planning. The system prepares and checks. Clinicians and coordinators decide.
Illustrative field: learned patterns on the left, rules and judgement on the right, expert intent in amber.
Typical decisions
Clinical documentation review
Check that a record is complete, consistent and ready for the next person who relies on it.
Care coordination
Prepare hand-overs, follow-ups and referrals so that nothing waits on a missing document.
Operational planning
Plan staffing, rooms and schedules against the constraints a team has to respect.
- Where the explanation gap shows
- A check that only says “incomplete” leaves the reader to find the gap. A plan that arrives without its assumptions cannot be adjusted with confidence, so it is rebuilt by hand.
- Where the intention gap shows
- Every unit has its own priorities: what must be on record before a hand-over, which absences a roster can absorb, what comes first on a busy day. That knowledge lives with the team, not in the system.
- What we deliver
- Checks that name what is missing and where it belongs. Plans that show their assumptions and constraints, and that re-plan around a change a coordinator makes. Preferences and rules written down once and read back to the team to confirm.
- The difference
Less blind re-checking
People confirm the points the system flags instead of reading everything again.
Complete, consistent records
Every record is checked against the same list, and gaps are named.
Expertise that stays
How the best coordinators weigh a plan becomes part of the system, and helps the next person learn.
- The boundary
- Our work stays on the organisational side of care. We do not diagnose and we do not recommend treatment. Those decisions belong to clinicians.
Insurance and public sector
Wherever a decision must be explained to someone else.
Our first focus is finance and healthcare. The method carries to other work in which someone outside the team reviews a decision: an auditor, a regulator, a claimant, a citizen.
Illustrative field: learned patterns on the left, rules and judgement on the right, expert intent in amber.
Typical decisions
Claims
Assess a claim against the policy terms, the evidence supplied and comparable cases.
Regulatory reporting
Prepare and check reports so that every figure traces back to its source and its rule.
Public administration
Prepare case decisions for permits, benefits or procurement, with the rule and the facts that apply.
- Where the explanation gap shows
- When a claimant, an auditor or a citizen asks why, an answer that cannot be traced to a rule and a fact is not an answer. Staff reconstruct the reasons afterwards.
- Where the intention gap shows
- Caseworkers apply rules with judgement: when to ask for more evidence, when the purpose of a rule outweighs its wording. A system that cannot take this in forces them to override it or to work around it.
- What we deliver
- Reasoning that cites the rule, the fact and the step. Rules held explicitly, so that a change in policy is a visible change to the rule base. Caseworker judgement captured as stated priorities and recorded overrides.
- The difference
Decisions that stand up to a question
The reasons are there when someone asks, not rebuilt afterwards.
Rules that change visibly
A change in policy is a change to the rule base, and each decision shows which version it used.
Overrides that teach
Disagreement is recorded with its reason, and the next similar case starts closer to the caseworker’s view.
How an engagement works
From one decision to a measured pilot.
Engagements are small and concrete. Each step ends with something your team can inspect.
Pick one decision.
Choose a decision your experts make often, that has an owner and that is reviewed afterwards. One is enough to begin.
What your team getsOne decision, its owner and the policies behind it.
Model how your experts reason.
We work with the people who decide and capture what they look at, which rules apply, how they weigh evidence and when they ask for more.
What your team getsAn expert model and a rule base, written down for your team to correct.
Pilot inside the workflow.
The system prepares the decision where your team already works. Experts see the reasoning, decide and override as they always do, and each override is recorded with its reason.
What your team getsRecommendations with their reasoning, and a log of overrides.
Measure decision quality with your team.
Together we agree how to judge the result: whether reviewers sign off faster, re-check less and apply policy more consistently. Then we decide what to extend.
What your team getsA shared reading of what changed, and what to do next.
What we need from you
A decision owner, time with the people who decide, and examples of past cases with the policies behind them.
The platform
CogniFlow, the decision-support platform we are developing with partners in finance and healthcare, takes shape from these pilots.
Bring us a decision.
Tell us which decision your experts make every day and where it slows down. We will say whether the approach fits.