Applications

ExpertForge is a browser-based software platform that gives a professional, an engineering practice or a company a team of AI-based experts grounded in its own documentation. Manuals, datasheets, standards, procedures, project notes, scientific papers, web content: everything that makes up the knowledge assets of the people doing the work becomes the basis on which the experts answer, search and produce.

What sets it apart from a generic conversational assistant is not only the quality of the answers, but what happens after the answer: ExpertForge takes the user all the way to the finished document — a report, a manual, a presentation, a video presentation, an image, an interactive application — without leaving the application and without having to rework the material downstream.

Anyone who works with technical documentation knows three recurring frictions. The first is the time spent finding information you know you have: an application note read months ago, a paragraph of a standard, a figure in a five-hundred-page manual. The second is the unreliability of generalist assistants when the question is specialised: they always answer, but you cannot tell on what basis. The third is the distance between the answer and the actual work: even correct answers still have to be composed, rewritten, laid out, illustrated and turned into a deliverable document.

ExpertForge tackles all three together. The knowledge base solves the first, because search works by meaning (semantic search) as well as by words; source grounding and precise citation solve the second, because every statement can be traced back to the document and the page it came from; the document generator solves the third, because the answer is not the destination but the raw material of the deliverable.

The application is built around a simple path that holds for any activity: you bring in the sources, you question the team of experts, you collect the results into a deliverable. Everything produced can in turn re-enter the user’s reusable assets.

This is the daily entry point. A conversation can involve from one to three simultaneous experts, each with its own language model, its own temperature and its own knowledge-base grounding mode. The answers appear side by side and are immediately comparable.

Conversations are saved with an automatically generated title, and are searchable, renameable and exportable.

This is the heart of the product: the private archive of documents on which the experts base their answers. Ingesting a document is not a simple upload but a processing chain that preserves its structure — headings, tables, formulas, images — and makes it searchable by meaning.

You can upload PDFs, Word documents, presentations, spreadsheets, text, tables and images; the maximum size of a single file depends on the plan. At upload time you choose the processing options: retention of the original, automatic generation of a topic index and description, optical recognition for scans, advanced extraction of formulas and tables for technical documentation.

An investigation environment over external sources, not a plain search engine. Ordinary search queries two engines simultaneously and merges the results, removing duplicates; the selected sources can be summarised by an expert with custom instructions on tone, length and structure.

This is the function that turns knowledge into a finished product. A guided path leads from choosing the document type all the way to export, with the option to step in manually at every stage: automation never takes control away from the author.

Two steps deserve particular attention. In the outline the application does not merely propose a structure: section by section it checks how much knowledge is actually available in the selected sources, flagging the parts that would remain uncovered. In the content step the author composes an editorial draft with hand-picked text, excerpts and images: this material takes precedence over automatically retrieved material, and is the way to stamp your own mark on the result.

In the editor the document can be refined by asking the expert for a new draft following your own instructions, and enriched with an assistant that analyses the text and proposes images, diagrams, charts and formulas where they are needed — with a maximum spending budget you set before generating them.

Everything the AI produces — images, charts, diagrams, code snippets, notes, summaries — converges into a personal archive filterable by title, type, date, category and tag, with an immersive viewer offering zoom, rotation, full screen and download. In the same area you generate new images by specifying the model, the format, the number of alternatives and optionally a reference image; the credit estimate is shown next to the generate command.

A function that exploits the models’ ability to write code: analysis dashboards, animations, simulations, dynamic charts, interactive presentations and small calculation tools, built by conversing with the expert and based on your own knowledge-base data. The applications run in a sandboxed environment inside the browser, can also be edited by hand, keep a version history and can be captured as an image or recorded as video to enrich your documents and complete your videos.

It is a rapid-prototyping tool, not a development environment: the applications live in a single file and use a limited set of libraries. Within those limits, though, they make a concept immediately demonstrable — and exportable for a client or a presentation.

A project is a logical folder grouping conversations, documents, resources, applications and reports that share a purpose — a client, a product, a piece of research. Selecting the active project from the top bar filters every view of the application accordingly, and new items flow into it automatically. It is the way to keep different working contexts separate without duplicating anything.

ExpertForge is a modern web application: the user works in a browser, and the sensitive logic lives entirely on a server. The design follows three criteria — independence from AI providers, configurability without touching code, and a clean separation between what the browser may see and what must stay on the server.

The web origins allowed to talk to the backend are configurable, preventing the application from being embedded in third-party sites. Payments are handled by a specialised operator: no payment-card data passes through the application or is stored in it.

The plans differ in usage volume, the power of the accessible models and the tools enabled. The table summarises the main differences.

The availability of image and video generation, the document types that can be generated and the monthly credit allowance are defined plan by plan and listed on the pricing page. Plan configuration can be changed from the administration panel: the values shown are those in force on the document’s issue date.

Subscription and credit-pack prices are published and always up to date at expertforge.boardesign.it/pricing. They are subject to change and are therefore not reproduced here; an early-subscriber offer is currently running on the paid plans.

Alongside the credits included in the subscription, additional packs can be bought at any time. Subscription-plan credits renew each period and, if unused, expire at the end of the subscription period; purchased credits (credit packs) do not expire and accumulate: they are only drawn on once the plan’s credits are exhausted.

The estimate is deliberately conservative: since it involves content production by AI models that cannot be predetermined, it cannot be computed exactly. The final figure, by contrast, is computed on the actual usage reported by the provider. If the balance is not enough to cover the estimate, the operation is stopped.

Plain web search, by contrast, consumes no credits: a cost only arises when the results are reworked by a model. Likewise, navigating the application, browsing the archive, organising by project and exporting already-generated documents are free operations.

Alongside the standard version available directly online, an Enterprise version is planned for organisations that do not want their documentation to reside on external infrastructure. The user experience stays identical — corporate users work from the browser on their own PCs, exactly as today — but the application and the services that make it up reside on the customer’s server instead of on BoarDesign infrastructure.

In this configuration documents, the knowledge base, generated resources and conversation history never leave the corporate perimeter. The only outbound communication that remains is the one to the language-model providers, made with keys registered and billed directly to the company — and if the organisation has local models served internally, the provider catalogue is configurable in that direction too.

ExpertForge brings together three things that usually live in different tools: a conversational assistant able to compare several models, a document archive searchable by meaning, and a generator of delivery-ready documents. The combination produces an effect none of the three parts would give on its own: accumulated knowledge stops being material to consult and becomes productive capacity — enriched by web research, image and infographic generation, and the generation of small applications such as dashboards, animations and calculators, with adjustable levels of grounding in your own knowledge base.

The application is reachable at expertforge.boardesign.it, where you can sign up and evaluate it on the free plan. For a guided demonstration, for customising the instance with your organisation’s visual identity, for assisted construction of a corporate knowledge base, or to evaluate the Enterprise version, BoarDesign is available at its usual contacts.

We are working on the A3 Platform. This tab will host the functional description, the screenshots and the downloadable document, in the same structure as the ExpertForge tab.

If the subject interests you and you would like to be notified at release — or want to flag a specific requirement now — get in touch: input gathered at this stage genuinely shapes the implementation choices.

We are working on the CIA software. This tab will host the functional description, the screenshots and the downloadable document, in the same structure as the ExpertForge tab.

If the subject interests you and you would like to be notified at release — or want to flag a specific requirement now — get in touch: input gathered at this stage genuinely shapes the implementation choices.