Skills

Skills are one layer of the platform Syntur is building, not the product. One core, specialist capability on demand. You should not have to keep a different AI assistant for each kind of work — Varra reasons broadly by default and reaches for a specialist Skill when a general answer is not good enough.

Where a Skill sits

A Skill is one layer of Varra, not the whole of it. It is worth seeing the stack before the catalogue, because a page listing twenty specialisms reads as though specialising were the point, and it is not — it is one of the things the system does on the way to something useful.

• General intelligence — understands the task. Live
• Router — works out what the task actually needs. Built, staged
• Work Items — hold one piece of work together across steps. Live, nothing automatic
• Specialist Skills — understand the domain. Modes live Packs staged
• Company context — understands how one business runs. Planned
• Varra Assurance — works out what can be trusted. Partly live
• Runtime — eventually carries out approved work. Planned

A Skill makes the Skills row good. The rows around it are what turn a good answer into work somebody can act on — and most of them are not built yet.

What a Skill is

A Skill is a named body of domain instruction the engine can load for one request: what the domain's failure modes are, which figures have to be checked rather than recalled, what a competent answer in that field is expected to contain, and where the answer must stop. It is not a separate model and not a separate product. The same Varra identity answers either way — a Skill changes the depth and the discipline of the answer, not who is answering.

That matters for the thing most people actually do with AI, which is ask about whatever is in front of them. A question about a rebate letter, a rafter span and a cash-flow statement should not require three tools and three logins.

Two mechanisms, honestly described

Specialist depth reaches Varra by two routes today, and they are at different stages. Both are described here as they actually are rather than merged into one claim:

• Work modes — live. Sixteen domain modes ship in the product now. You pick one, and it replaces the working context for the conversation: its own system prompt, its own cautions, and for three of them a verified reference table that overrides the model's own figure. This is what you can use this afternoon.
• Skill packs — built, staged. Six Skill packs live in the engine's routing registry with their own eval set. They are selected by the router rather than by you, they can compose with one another, and they sit behind a switch that reads as off unless it is deliberately turned on. Nobody using Varra today is getting one.

The direction is for the second to absorb the first: you describe the work, and Varra determines what the work needs. Choosing your own mode is the current state, not the destination.

The catalogue

Everything below either exists in the engine now or is a genuinely planned direction, and each entry says which. Nothing on this page is listed because it would round out a category.

Live in the product today   Staged built and tested, behind a switch that is off   Planned intended, not built

Business & operations

First, because this is where the built Skill packs actually are. It is the work where a specific document, a specific number and a specific company rule all have to agree before an answer is worth anything — and where getting it wrong is a cost somebody absorbs rather than an embarrassment.

Build & field

The deepest domain coverage, and the reason for it: this is work where a confidently wrong number costs a callback, a failed inspection, or an injury. It is where the verified reference data starts, not where the product stops.

Technology

Everyday

Varra is a generalist first. Two of the live modes are here to make that concrete rather than to fill out a trades list, and General mode — research, writing, analysis, planning, ordinary questions — is the default every conversation starts in.

Skills compose

Real work rarely sits inside one domain. A disputed construction invoice is a document question and a finance question at once; a piece of equipment that keeps tripping a breaker is an HVAC question and an electrical one. So a Skill pack declares which other packs it can run alongside, and the registry carries those relationships rather than leaving them implied:

• Construction Invoice Audit composes with Document Evidence Analysis.
• Document Evidence Analysis composes with Financial Analysis and Investment Research.
• Financial Analysis and Investment Research compose with each other.
• Electrical Troubleshooting composes with HVAC Diagnostics.

That is the mechanism a single assistant needs and a folder of separate assistants cannot have. It is also worth being precise about what it is not: composition means two Skills can inform one answer to one request. It is not a workflow engine, it does not run a sequence of steps on its own, and nothing in the registry executes anything. A piece of work that spans several domains is still one conversation with a person in it.

A Skill decides how a task should be handled and what a good answer contains. It is not how Varra reaches a system, and it is not what carries out an action — that is Varra Runtime, a separate execution layer that is planned rather than built. Eventually a Skill may end in a real action rather than only an answer; today every Skill ends in an answer.

Asked plainly. Answered carefully.

Two real questions, asked the way somebody would actually type them, and what came back. Neither one asks for a checklist or a risk level — the shape of the answer is the Skill's, not the prompt's. They sat on the home page until it was cut back to one product story; two specialist transcripts are a Skills demonstration, so this is where they belong.

Both are the model's own reasoning inside a mode. Neither is an independently verified answer — that is a much shorter list, and it is on the Assurance page.

What is deliberately missing

A mode existing is not the same as that mode's numbers being checked. Three of the sixteen carry verified reference data and only two are published, which is the honest version of the coverage claim — the full accounting is on the Assurance page. The automotive torque figures are the clearest example: they are in the engine, they are not published, and they should not be, because the values held today carry no vehicle. A fastener torque without the vehicle it belongs to is exactly the confidently wrong number this whole approach exists to prevent.

Filling that in properly is 02 — Specialize on the roadmap.

Try Varra now →