The source stays attached
The document, ticket, file, message, or code change stays connected to the answer or draft.
Private AI for sensitive work
Foundry helps your business use AI on documents, client intake, support, internal knowledge, and code review — from equipment you control, with records your team can check, and approval before important outputs move forward.
Book a Foundry Fit ReviewSee how it keeps a record
Built for owners and partners who want AI to save time without turning client files, business records, or staff judgement into a black box.

The everyday problem
Documents arrive. Client files need checking. Support tickets need first replies. Staff search old folders for answers. Code changes wait for review.
AI can help with the first pass, but only if the business can still answer simple questions: What did it read? What did it produce? Who checked it? What was approved?
Foundry is built around those questions. It helps with repeat work, keeps a record of what happened, and makes sure your team checks the important outputs.
A client pack arrives. Foundry reads it, lists what is missing, links back to the source documents, drafts the chase message, and stops for review. Nothing important goes out until a person approves it.
Example work progress for agreed jobs.
Foundry runs on Apple hardware controlled by your business. It sits beside the tools you already use — folders, inboxes, helpdesks, case files, knowledge bases, or code repositories — and helps with agreed jobs.
It is not a public chatbot. It is not a pile of DIY AI tools. It is a managed setup for repeatable work: reading, sorting, extracting, searching, drafting, flagging, recording, and queueing for review.
Shows its working
For sensitive work, speed is not enough. Owners and partners need a record.
Foundry is designed to show what came in, what it read, what it found, what it drafted, what it flagged, and who checked it before the next step.
The document, ticket, file, message, or code change stays connected to the answer or draft.
Your team can see what is waiting, what needs checking, what has been flagged, and what has already been handled.
Client replies, document outputs, exceptions, and sensitive actions can pause until the right person checks them.
The system can keep a practical record of the job: what arrived, what Foundry did, what it produced, and what your team approved.
Foundry changes “the AI said so” into “here is what happened.” This can help with accountability and internal governance. It is not legal advice and does not automatically make a workflow compliant.
The right tool for each part of the job
A business process has different steps. Some are simple and repetitive. Some need careful reading. Some need a specialist tool to read a scanned page or pull fields from a document. Some need a stronger AI step to draft a careful summary for review.
Foundry can use the right local tool for each part of the job, instead of pushing everything through one big general-purpose AI model.
Sorting, tagging, routing, checking formats, spotting missing fields, and preparing work for review.
Reading scanned pages, extracting fields, handling tables, and keeping the answer tied to the source document.
Summaries, comparisons, client-reply drafts, code-review notes, and exception explanations for a person to check.
The point is not more AI. It is using the right tool at the right step.
Some client documents, legal files, financial records, support tickets, and code should not be copied into a general AI service. Foundry gives suitable work a more controlled route.
Foundry keeps the source, draft, flags, queue, and approval step together so managers can check the path.
Cloud AI can be useful, but prices, limits, models, accounts, and policies can change. Foundry helps bring suitable work under your own control.
Local AI is only useful if it is managed. Foundry covers setup, workflow design, review queues, support, and sensible guardrails.
Foundry is strongest where the work is repeatable, sensitive, review-heavy, and expensive to handle manually.
Invoices, purchase orders, contracts, renewal notices, forms, and quotes can be handled as agreed jobs. Foundry keeps the original file, pulls out key fields, flags issues, keeps a record, and creates a review queue.
For matter packs and intake files, Foundry can identify documents, check completeness, flag missing evidence, and draft chase messages for approval. It supports the admin workflow; it does not give legal advice or make professional decisions.
Foundry can read tickets for agreed support queues, check the knowledge base, draft replies, route bugs, and pass complex issues to people with context already summarised.
Foundry can search selected internal files and answer questions with references to source documents where possible. If it cannot find a source, it should say so.
Foundry can review code changes for security, tests, edge cases, consistency, and obvious performance issues before a senior engineer spends their attention.
reduction in document handling time in a structured document-processing workflow.
faster matter readiness in the conveyancing intake example.
routine support response time in the client-support workflow.
to find internal knowledge in the search example.
PR review path in the code-review workflow.
Figures are from example case-study workflows and demo evidence, not promised results. The Fit Review checks what is realistic for your volume, tools, data quality, hardware, AI tools, and approval process.
For agreed jobs, Foundry runs the AI work on equipment your business controls, reducing the need to send sensitive material to a general cloud AI provider.
Foundry can show the source, draft, flags, queue status, and approval step.
It can read, sort, draft, search, and flag. It should not make important legal, financial, client, HR, medical, or compliance decisions on its own.
As better local AI tools become practical, the setup can be reviewed and improved without throwing away the process.
For work moved into Foundry, the business is less dependent on one cloud AI account, one price change, one service outage, or one platform rule change.
Practical visibility
Foundry should make the work visible enough for a manager to understand without asking an engineer.
Example work progress, not live customer telemetry.
More control over the process
Cloud AI tools can be useful. They are also outside your control. Prices can change. Limits can appear. Models can be retired. Accounts can be restricted. Services can have bad days.
Foundry does not remove every dependency. You still rely on hardware, software, support, electricity, networks, and the systems your business already uses.
But for suitable work moved into Foundry, more of the process is under your control: the equipment, the workflow, the review steps, and the record of what happened.
Book a Foundry Fit Review. We will look at the work, the data, the risks, the current cost, the approval steps, and whether a private AI setup is sensible.
If it is a fit, you will know where to start. If it is not, we will say so.
Book a Foundry Fit ReviewSee example workflows
No generic AI pitch. Just a practical answer.