AI Skills is the reusable library that makes every other pillar consistent — the data dictionary, the frameworks, the SROI method, the standards, the deterministic-output rule, and the brand voice, all packaged as skills. It isn't a course you take start to finish; it's the library every pillar reaches into, so a field means the same thing and an output reads the same way everywhere.
AI Skills is not a course you take start to finish — it's the reusable library every other pillar reaches into. A data dictionary, the frameworks, the SROI method, the standards, the deterministic-output rule, and the brand voice, each packaged as a skill so a field means the same thing and an output reads the same way in Case, Grant, Portfolio, and Survey Intelligence alike. Define the method once; apply it everywhere.
Key takeaways
AI Skills means turning the methods you'd otherwise re-derive every time — how a field is defined, how a framework is built, how SROI is computed, how the AI reads and renders — into reusable skills that every pillar calls. The method becomes the asset, not the person who happens to remember it.
If your organization runs more than one program — across sites, years, or funders — you've felt the cost of not having this: the same “employment status” means three different things, every analysis rebuilds the framework from scratch, and the AI gives a slightly different answer each run. Comparability, reuse, and consistency all break at once. AI Skills fixes the layer underneath, so everything built on top lines up.
The incumbent is scattered templates, one-off consultant methods, and whatever lives in an analyst's head. AI Skills makes the method a versioned, reusable asset. Three shifts make the difference:
The other pillars don't duplicate these — they call them, so a fix made here propagates everywhere at once.
Forget “AI feature” for a moment. In this library, a skill is a reusable, deterministic method Sopact's AI applies the same way every time — a data-dictionary definition, a Theory-of-Change build, an SROI value map, the house rule for how output is graded and cited. Each is written once, versioned, and called by every pillar and use-case — so the method travels with you instead of living in one person's memory.
Honest boundaries first. AI Skills is the standards layer under the pillars, not a standalone workflow. Reach for it when you need consistency across programs; don't expect it to be a start-to-finish product path on its own.
| Strong fit | Why |
|---|---|
| You run more than one program | One shared dictionary makes every program, site, and year comparable |
| You're building a framework | Theory of Change, logic model, logframe, results framework — build and audit |
| You compute SROI | The full method — value map, proxy, deadweight, ratio — as reusable skills |
| You report against standards | IRIS+, the IMP Five Dimensions, and ESG applied consistently |
| You're tired of AI drift | The deterministic-output standard: same input, same graded, cited output |
| You need on-brand output | The Sopact voice and component set, packaged as a skill |
| Not this | Why |
|---|---|
| A standalone product workflow | It's the layer under the pillars, not its own start-to-finish path |
| A one-off analysis you never reuse | The value is reuse — across programs, years, and pillars |
| A replacement for the pillar courses | The pillars call these skills; they don't compete with them |
| A generic AI chatbot | Deterministic and cited, not a free-form model that drifts run to run |
The rule of thumb: if you'd otherwise re-derive the same method in more than one place, it belongs in AI Skills — defined once and called everywhere.
You don't follow one organization through AI Skills — you see the same skills surface in every other pillar's demo: the graded Theory of Change in Case Intelligence, the standards mapping in Portfolio, the cited scoring in Grant, the themed open text in Survey. All of it runs on this library.
The same skills power every pillar's demo — deterministically, and cited to source.
One definition per field; Theory of Change, logic model, logframe, and results framework — build and audit.
Skill clusterThe full SROI method — value map, proxy, deadweight, ratio — plus IRIS+, the Five Dimensions, and ESG.
Skill clusterThe deterministic-output standard (same input, same graded output, cited), and the Sopact voice and component set.
Skill clusterAI Skills begins with the smallest possible unit: one field, defined once. Pick a metric you collect in more than one program — say “employment status” — and write its single definition and unit. That one line is the first entry in your data dictionary, and the reason your programs will finally be comparable. The data-dictionary skill builds it out.
Take one term two of your programs both use and reconcile it to a single definition. It's the smallest possible start on the standards layer — and the fastest way to feel why defining once, and applying everywhere, changes everything downstream.
Data leads, analysts, and the people who keep a whole organization's measurement consistent — anyone tired of re-deriving the same framework and getting a slightly different answer every time.
Frequently asked questions
The reusable library — data dictionary, frameworks, SROI, the deterministic-output rule, and brand voice — that makes every Sopact pillar consistent.
It's a reference library you dip into as needed and cross-link from every pillar, not a start-to-finish path.
The house standard for how Sopact's AI reads and renders: deterministic (same input → same graded output), evidence-graded, every claim pinned to source.
One definition per field so it means the same thing across every program, hospital, and year — the foundation of comparability.
Theory of Change, logic model, logframe, results framework; IRIS+, IMP Five Dimensions, ESG; and the full SROI method.
They call these skills — the Case demo's graded ToC, the Portfolio demo's standards mapping, the Grant demo's cited scoring all run on AI Skills.
Next: Build One Data Dictionary That Holds Across Every Program → · or Try Sopact Sense →
ChatGPT, Claude, and Gemini are fine for a quick test — but not for an answer you'll put in front of a funder or board. When it has to hold up, run it in Sopact Sense.
Try it in Sopact →
