Skip to content
Start a project Sign in
Build with Specicon

Design, engineering and intelligence, in one team.

Specicon designs and builds websites, commerce experiences, portals and AI products. Where a decision matters to your customers, we build the reasoning into it, and let them see it.

Start a project See the work How we deliver
Who it’s for

Four kinds of team we build with.

01

A founder building an AI-native business

You need a product that works on the first day and a decision core you can explain to customers and investors.

First stepA scoped first release: the smallest product that proves the idea.

02

An established company rebuilding its site or store

The site sells but does not guide. We rebuild the journey around the decisions your customers actually make.

First stepAn audit of the current journey and a rebuild plan, page by page.

03

A product team adding discovery or decision support

You have the product and the users. You need recommendations or eligibility rules that explain themselves.

First stepOne decision, scoped: data, rules, where it appears, how it is measured.

04

An enterprise needing a controlled portal or application

Access control, audit trails and review steps matter as much as the interface does.

First stepRequirements for roles, data handling and review, before any design.

What we build

Three kinds of build.

AI goes where it improves a real decision or workflow. Where it would not, we leave it out and build the rest well.

Websites & commerce experiences

Company and brand sites, storefronts, product pages, guided finders and quizzes, and content structured so search engines and AI answers can read it.

Typical deliverablesDesign system, page templates, content model, storefront integration, release audit
Where intelligence helpsProduct discovery, finders, comparison, what to show whom
Seen in the Specicon site

AI applications & portals

Customer portals and self-service accounts, internal operations consoles, guided advisors and assistants, and review, approval and routing workflows.

Typical deliverablesRoles and permissions model, application screens, integrations, audit trail, admin tools
Where intelligence helpsRouting, triage, eligibility, next best step, summarising for a reviewer
Seen in the Specicon portal and console

Decision engines & integrations

Recommendation and ranking with reasons, eligibility and constraint rules, holdout assignment for measurement, and integration into the stack you already run.

Typical deliverablesDecision design, input and provenance map, explanation format, measurement plan, working prototype
Status todayMatching, ranking and eligibility engines, each returning its result with the reason.
Seen in the Decision Inspector · the engines

We design for many sectors. The industry pages describe how decisions differ by sector.

Services

What you can hand to us.

Strategy and scope
What to build first, what to leave out, and how success will be measured.
User research
Interviews and journey reviews with the people who will use it, before design starts.
Interaction design
Flows, screens and states, tested as a clickable prototype.
Brand implementation
Your identity turned into a working design system: type, colour, motion, components.
Frontend engineering
Fast, responsive pages and applications, checked at every screen width.
Backend engineering
APIs, data models and services behind the interface, with tests.
Integration
Commerce platforms, CRMs, identity, payments and the systems you already run.
Accessibility
Keyboard routes, visible focus, contrast, reduced-motion paths and screen-reader labels.
Security considerations
Access control, data minimisation and review of what leaves the browser.
Launch
Performance, search and consent checks, redirects, monitoring and a release audit.
Analytics
One named event per step of the journey, sent only with consent.
Iteration
Measured changes after launch, planned from what the numbers say.
How we deliver

Six stages, the same ones every time.

Each stage ends with something you can review and a decision you make. How long each takes is set in the scope, once we know what is being built.

  1. 01

    Discover

    The business problem, the people it affects, the decision the product has to win, and what would count as proof.

    You receiveProblem statement, decision map
    You decideWorth scoping?
  2. 02

    Define

    Sitemap or feature list, surfaces, content model, data and integrations, success measures. Agreed in writing before design starts.

    You receiveWritten scope, success measures, estimate
    You decideApprove scope and engagement
  3. 03

    Design

    A visual system that is yours, the key screens, and a clickable prototype of the riskiest part, tested with real users where possible.

    You receiveDesign system, prototype
    You decideSign off the prototype
  4. 04

    Build

    Frontend, backend and integrations on a staging site you can use throughout. QA at every screen width, with accessibility checks in the same pass.

    You receiveStaging build, QA report
    You decideReady for launch review?
  5. 05

    Launch

    Performance, accessibility, search and AI-visibility checks. Redirects, sitemap, consent and monitoring in place before the switch.

    You receiveRelease audit, live site
    You decideGo live
  6. 06

    Grow

    Measure against the success measures agreed in Define. Where a decision is involved, compare against a holdout before claiming any effect.

    You receiveMeasurement readout, next plan
    You decideContinue, change or stop
Engineering and AI

How intelligence goes into a build.

Every recommendation or rule in a Specicon build is designed as a decision: its inputs and where they come from, the rules that bound it, the reason it shows, and how its effect will be measured.

Available today Design and engineering of the experience; decision design (inputs, rules, explanations, measurement plan); a working productDecision Inspector; integration with rules or models you already run. Live The Specicon decision runtime and API. The written scope for your project says which of these it uses.
  • 01Reasons by defaultA recommendation carries the factors that produced it, written for the person who sees it.
  • 02Rules a model cannot overrideBudgets, eligibility and policy are applied before ranking, and every exclusion is logged with its rule.
  • 03Provenance on every inputDeclared, observed, inferred, derived or unknown. Unknown inputs contribute nothing instead of being guessed.
  • 04A holdout from the startA share of sessions follows a reference ordering under the same rules, so the effect can be measured.
  • 05Fast, accessible pagesStreaming pages that paint early, keyboard routes, visible focus and reduced-motion paths.
  • 06Consent before analyticsEvents are named per step and stay in the browser until the visitor consents.
Ways to engage

Three ways to work together.

There is no price list here on purpose. Every engagement is quoted against a written scope, and no project is given a fixed timeline before that scope exists.

01

Discovery and definition

Suits you whenThe problem is clear but the answer is not, or you need a scope before you can set a budget.
Typical deliverablesProblem statement, decision map, written scope, success measures, a prototype of the riskiest part.
How scope is agreedA short, fixed engagement agreed up front. The output is yours whether or not we build it.
02

Scoped build

Suits you whenYou have a defined site, store, portal or product and a written scope for it.
Typical deliverablesDesign system, the built and tested product, a release audit, and the launch itself.
How scope is agreedPriced against the written scope and paid at milestones. Changes are quoted, never absorbed silently. The timeline is set in the scope.
03

Continuing product and engineering team

Suits you whenYou ship continuously and want design and engineering capacity that already knows the product.
Typical deliverablesContinuous design, build and measurement, with the care plan below included.
How scope is agreedA set monthly capacity. Priorities are reviewed together each month.

After launch: the care plan

Included in a continuing team · optional after a scoped build
MonitorUptime, errors and asset health, with alerts that reach a person.
UpdateDependencies, browsers and security changes applied on a schedule.
ChangeContent, campaigns and new pages. New surfaces go back through Define.
ReviewA regular review of what the measurements say and what to build next.
Compare

Which kind of build do you need?

Each is the right answer for someone. If yours is the first column, you probably do not need us.

A site builderAn agency buildAn intelligence-backed product
Best forA straightforward site your team edits itselfA custom-designed site or campaignA product where a decision is central: what to buy, which plan, who to route to
Speed to startFastest; templates are readyDepends on the scope and the agencySlower to start: the decision has to be designed before it is built
Custom logicLimited to the builder’s features and pluginsWhatever is scoped and builtDecision design with inputs, rules and reasons
Explains its suggestionsNot usuallyOnly if it was specifiedBy default, from the same factors that made the decision
Measuring the effectPage analyticsAs agreed with the agencyA holdout comparison planned from the start
The wrong choice whenThe site has to guide complex choicesNo one owns the logic after launchYou need a simple brochure site quickly; a site builder is cheaper and faster
Start here

Bring the problem. Leave with a written scope.

Start with a short brief. If there is a fit, a discovery engagement turns it into a scope you own, whether or not we build it.