tools.digitalbridge.ie — a 50-tool applied AI platform

What problem was this solving?

Most firms claiming AI capability have slide decks. We wanted an estate we could point at, where the inference runs server-side, the costs are metered, and the scoring logic is ours rather than a wrapper around someone else's answer.

What did we build?

Fifty-plus shipped tools on one platform, including two proprietary scoring algorithms and production generative AI running server-side.

• AI Product Angle Generator — production generative AI turning a product into positioning angles.
• AI Batch Image Generator — high-volume generative imagery with queueing and cost control.
• Website Worth Analyzer, carrying the proprietary Website Worth Index scoring algorithm.
• Business Profile Authority Ranker, carrying the proprietary DigitalBridge Authority Score.
• A shared server-side inference layer so no model key is ever exposed to a browser.
• Crawl and enrichment pipeline combining Firecrawl retrieval with Gemini reasoning.

What does it integrate with?

The integration surface technical buyers ask about.

• Server-side model invocation with keys held in edge-function environment, never client-side.
• Firecrawl for structured retrieval, Gemini for reasoning over the retrieved corpus.
• Shared scoring library so both indices stay reproducible between runs.
• Rate and cost controls at the platform layer rather than per tool.

How does the system actually work?

The decision path in order. Client-confidential detail is deliberately excluded.

• 1. Input — A URL, a business profile or a product description, submitted by the user.
• 2. Retrieve — Firecrawl gathers the structured source material the score depends on.
• 3. Reason — Gemini interprets the corpus server-side; no key ever reaches the browser.
• 4. Score — A deterministic scoring library produces the Website Worth Index or Authority Score.
• 5. Return — A reproducible figure plus the reasoning behind it, not an opaque number.

What made it hard?

The real engineering constraints on this build.

• A score is only useful if it is reproducible — both indices are deterministic given the same inputs, which rules out free-form model output as the final figure.
• Generative image work is expensive at volume, so batching and queueing were design requirements, not optimisations.
• Public tools attract abuse; inference had to be gated without making the tools useless to genuine users.

What was the outcome?

The platform is live and public. It is also the single clearest answer to the question 'have you actually shipped AI in production' — a prospective client can use the tools themselves before speaking to us.

What capability does this demonstrate?

Engagement band: Applied AI engineering band.

• Two proprietary, named scoring algorithms
• Production generative AI (text and image) at volume
• Retrieval-augmented reasoning over crawled corpora
• Server-side inference with zero client key exposure

Are the scoring algorithms just a model prompt?

No. The model does the retrieval-side reasoning; the score itself is computed by a deterministic library so the same inputs always produce the same figure. A score that changes between runs is not a score.

Can we have a tool like this for our own business?

Yes — a customer-facing AI tool of this shape is one of the more common briefs we take. It sits in our applied AI engineering band, and you own the source code and the prompts at the end of it.

Where does the inference run?

Server-side, in EU-hosted edge functions, with provider training disabled. No model key is ever present in browser code.

A note on what is published here

Case studies on this page describe work Digital Bridge delivered. Client-confidential detail, credentials and internal data structures are deliberately excluded. Where a client has not authorised publication of measured figures, outcomes are described qualitatively rather than quantified — we do not publish numbers we cannot evidence. Last reviewed 2026-08-18.

Related reading

• AI integration services
• How we de-risk a production AI build
• AI engagement bands and pricing

Other builds in the estate

Continuous AI incident triage across four Irish depots: natural-language and image severity classification, Eircode geolocation, dispatch routing and duty-principal escalation. A full web-to-print platform with a live design configurator, print-file automation and twenty-nine-plus structured product pages. Generative redesign from a single photo, paired with a live multi-retailer price engine that keeps the result inside the user's budget. Jo McAteer, Forever Home Sanctuary, Riversdale House, AI Triangulate, Equi Consignment, RCP Psychotherapy and Time To Change — grouped by what the engineering actually required.

Could this pattern work in your operation?

Every production build starts with a seven-day shadow-mode pilot, so you see the system's decisions against your own data before it touches anything live.

What is the Digital Bridge AI platform used for internally?

This is the internal tooling we use to run parts of our own studio: drafting first-pass content for client review, checking site structure against the technical rules we apply on every build, and tracking the delivery detail across active projects so nothing depends on one person's memory. It is included in this portfolio because it is a genuine working system rather than a demo, and because clients reasonably ask whether we use the kind of tools we build for them — we do, and this is what that looks like in practice.

Where does human judgement stay in the loop?

Nothing the platform drafts is published without a person reviewing it first; it produces a starting point, not a finished decision, which is the same rule we apply to every client-facing AI build. No client data from one project is used to inform another client's work, and any data the platform handles is kept under the same GDPR-consistent retention rules we set for client systems. The honest limit is that internal tooling like this saves drafting time, not judgement time — deciding what is true, what is worth publishing and what a client actually needs still takes a person doing the thinking, which is exactly why the platform is a support tool and not a replacement for the studio's work.

What made building our own AI platform, rather than reselling one, worth the effort?

A fair question, since plenty of studios simply resell an existing AI vendor’s platform under their own branding. We chose to build our own for a specific reason: reselling means you’re constrained by whatever features and pricing structure the underlying vendor decides on, and clients end up paying for capability they don’t need or missing something they do. Building our own platform meant we could design it around how we actually scope and deliver AI projects for Irish small businesses — starting from a discovery assessment, then configuring the system to match a specific business’s workflow rather than forcing a business to adapt to a generic template. It also means we understand the platform at a technical depth that a purely reseller-based studio can’t match, which matters when something needs troubleshooting or a client’s requirements shift after launch. This doesn’t mean every client needs a fully custom AI build — for plenty of businesses, a simpler chatbot or automation is the right call — but having our own platform means the option exists to go further when a project genuinely warrants it, without being boxed in by someone else’s product roadmap.

How do you decide when a client needs a custom AI system versus an existing tool?

This decision happens during the discovery assessment, and it’s driven by the complexity and specificity of what a business actually needs rather than any preference for building things ourselves. If an off-the-shelf chatbot or automation tool genuinely covers a client’s requirements, we say so, because there’s no point charging for custom engineering when a simpler, cheaper option does the job just as well. Custom builds tend to make sense when a business has workflows, data structures or integration needs that don’t fit neatly into existing platforms — situations where forcing the business to adapt to generic software would actually cost more in lost efficiency than building something tailored. Our own platform gives us the flexibility to build genuinely custom logic when that’s the right call, without starting from zero on every project. We’re upfront that AI systems built this way are quoted individually after the paid discovery phase, because the scope varies too much between clients to price generically, and we’d rather a client pay for exactly what their business needs than a package that doesn’t quite fit.