How we assess

Every engagement runs the same diagnostic and ends in a clear verdict.

A structured test of objectives, process and feasible AI intervention. We apply it to every engagement, so AI is the right solution, in the right context, with the right safeguards.

One practice, two markets

The same assessment method runs for both. What differs is the clock and the shape of the build.

Political parties

In opposition or in office.

Weeks
Timeline to a working deployment
  • Built on proven tools rather than started from scratch.
  • Suited to the pressure of an active campaign, or a party already in office.
Government institutions

Ministries, agencies and authorities.

Phased
Timeline set by the scope of the engagement
  • Built alongside an institution's own systems and constraints.
  • Sequenced so each phase is assured before the next begins.
The four-node diagnostic

Only when all four nodes are satisfied do we recommend intervention.

01

Policy objectives

What is the agency trying to achieve, and where is it falling short?

Anchor in real outcomes.
02

Process mapping

What process delivers that objective, and where is the constraint?

Map speed, cost and quality.
03

AI fit test

Is the constraint of a type that AI can address safely and effectively?

Cognitive vs structural constraints.
04

Feasibility check

Can it be implemented with appropriate safeguards in place?

Data, legal, oversight, capacity.
01

Political objectives

What is the party trying to achieve: engagement, clearer communication, mobilisation, coordination? Where is it falling short?

Anchor in measurable political outcomes.
02

Political process mapping

Which activities and workflows deliver that objective, and where are the constraints?

Map reach, speed, consistency, impact.
03

AI fit test

Can AI address the constraint without replacing political judgement or democratic accountability?

Analytical, communication, workflow.
04

Feasibility and trust check

Can it be implemented lawfully, transparently and without undermining public trust?

Data protection, electoral rules, oversight.
The verdict

Clear outcomes. Confident decisions.

Every assessment concludes with a clear, evidence-led verdict so leaders can act with confidence.

Proceed
High potential, low risk

Strong strategic fit and feasible implementation with appropriate safeguards.

→ Implement with defined controls
Pilot
Medium potential, moderate risk

Value exists but needs validation through a controlled pilot with clear hypotheses and success metrics.

→ Define the pilot
Reconsider
Low potential or high risk

Limited value, or material risks that outweigh the benefits. Revisit if circumstances change.

→ Do not proceed now
Not suitable
Structural barriers

The problem is not amenable to AI, or cannot be implemented responsibly.

→ Redirect resources

Every verdict is accompanied by transparent rationale, evidence and recommended next steps.

Three pillars, one standard

Reliable. Responsible. Valuable.

Reliability

Does it actually work?

Output consistency and accuracy. Failure modes and robustness. Performance under real conditions. Reproducibility and testing.

Responsibility

Is it safe and appropriate?

Bias, fairness and inclusivity. Privacy and data protection. Explainability and transparency. Legal and ethical compliance.

Value

Does it deliver impact?

Outcome-improvement potential. Cost, time and quality impact. Scalability and sustainability. Return on public investment.

Governance is not a gate at the end. Intake, evidence, independent review, safeguards, oversight and continuous learning run through the whole lifecycle.

See what the verdict leads to.

Eight reusable capabilities, built only when the assessment says so.

What we build