Execution intelligence
SNext
From the strategic plan a minister signs to the invoice line that pays for it — one Arabic-native system, one audit trail.
Counted in the running platform, September 2026
The problem
Strategy lives in slides. Delivery lives in a scheduling tool. Money lives in a fourth system.
When a regulator asks what a pillar cost and what it bought, the answer gets assembled by hand from four exports and a phone call. SNext keeps the chain intact end to end, so the answer is a query instead of a project.
Direct the strategy
Pillars and themes, plans and sub-plans, strategy maps, cascading objectives and weighted KPIs. Balanced Scorecard, OKR or Hoshin Kanri — the framework is a configuration, not a different product.
Deliver the work
Portfolios, programmes and initiatives down to a work breakdown structure, dependencies, critical path and earned value. Risks, issues, changes, decisions and lessons all hang off the same spine.
Govern the decisions
Committees with notice-period compliance built in, a decision pipeline, and minutes sealed in a hash chain — so a signed minute cannot be quietly edited after the fact.
Close the commercial loop
Engagements, bills of quantities, measured progress, payment certificates with retention and advance recovery, variations, final accounts — plus procurement and client billing.
The module map
This is the actual navigation, grouped.
Not a roadmap and not a feature list — the sections a licensed user sees on the left of the screen. It is a long list because consolidating four systems produces a long list.
Strategy
Portfolio & delivery
Programmes & resources
Governance
Commercial
Analysis & reporting
One timeline
Every plan, where it lands and whether it is on track.
Decision Studies
An institutional second opinion, with every claim traceable to a page number.
Commission a study, ingest the corpus, and a panel of AI seats — each with a written mandate — produces claims anchored to specific document units. Every claim is independently challenged and marked holds, holds in part, does not hold, or unverifiable, with dissent recorded rather than resolved away. The readiness verdict is weakest-link, so one strong section cannot outvote a broken one.
Decision Studies is in production pilot with one government entity and is enabled per organisation rather than by default.
The AI question
Your strategy never leaves your boundary.
Most tenants run SNext's assistant on a model hosted inside their own network. Every request passes a classification gate — public and restricted material may reach a cloud model, classified material may not — with mandatory redaction before anything leaves, and a build that fails if any code path tries to go around it. An organisation with no configured provider gets AI switched off rather than quietly falling back to a shared key.
And you name it
The assistant carries your brand, not ours.
Name, icon and persona are per-organisation settings — one commission calls theirs Nadim. It reaches 67 governed tools across the platform under a default-deny permission model, so it can draft a KPI band or summarise a risk register without being able to do anything nobody granted it.
Forecasting, health scoring and simulation are statistical models — regression and Monte Carlo sampling — not trained machine learning. We would rather say so than call it AI.
Arabic is not a translation layer
The whole chain, right to left.
Over 21,000 interface strings in each language, kept in step by the build, and 700-odd Arabic columns in the database — because the objectives, KPI names, minutes and certificate lines are content, not chrome. Write a field in one language and the assistant drafts the other, marked as machine-written until a person edits it; a human value is never silently overwritten.
Let’s engineer what comes next.
Bring the plan you have to report on this quarter. We will show you the chain from its top objective to a paid certificate.