A Local LLM for Solicitors, Tested on Your Files
Legal documents are long, structured and unforgiving of a model that paraphrases. The model that drafts a good client email is not always the one that finds clause 14 in a lease.
£500 per month for the software and the hardware rental. You rent the machine — you don’t buy it.
A local LLM for solicitors is a language model running on a machine in the firm rather than a provider’s cloud, chosen and tested on the firm’s own matter files, correspondence and precedents. It searches, summarises and drafts with citations, never gives legal advice, and is installed, kept current and supported by us.
Find out which model suits the job
A short call with a partner or the COLP. Bring your engagement terms; we will walk through where the model runs and what it sees.
One machine in the firm, an open-weight model chosen on legal documents, and the chat and document tools around it. Matter files never leave the office.
Cloud Agents for Intake, Local AI for the Matter Files
New-client calls and the general enquiries inbox carry little that is privileged, so firms run those on our hosted agents and keep client files on a Local AI machine in the office.
- ☁ Cloud-based AI Receptionist
- ☁ Cloud-based Email Manager
- ☁ Cloud-based Leads Outreach
- 🏢 On-premise Local AI
Do It Yourself, or Have It Done
Running a local LLM is not hard for an afternoon. Running one for law firms, every working day, with backups and someone to ring, is a different job.
| DIY on a spare machine | Cloud LLM subscription | Managed local LLM in the firm (ours) | |
|---|---|---|---|
| Where prompts and files go | Your machine | The provider’s servers | Your machine |
| Choosing local LLM models | Trial and error, evenings | No choice | Tested against your documents by us |
| Hardware | Whatever is spare | None | Specified for the team and rented to you |
| Reads your document library | If you build it | Only what is pasted in | Document spaces with citations |
| Model upgrades and updates | You, when you remember | Automatic | Us, as part of support |
| When it breaks | You | Their status page | Restored from backup by us |
Why Legal Documents Are the Hardest Test for a Local Model
Legal work is the strongest case for a local LLM and the hardest test of one. The case is strong because the material — privileged correspondence, counsel’s advice, client files under restrictive engagement terms — cannot go to a cloud provider, and a model in the firm means it never does. The test is hard because legal documents are long, precisely worded and structured, and a model that paraphrases where it should quote is worse than no model at all.
So the model choice is made on your documents. Open-weight models differ in how well they handle long contracts, how faithfully they cite the passage they relied on, and how they draft against a house-style precedent; the best one for a litigation team assembling chronologies may not be the best for a conveyancing team asking what a lease says. We shortlist candidates on a call with a partner or the COLP, specify a machine that runs the largest of them well, and test the shortlist against a sample of the first practice area’s real files once the machine is installed.
What fee earners then have is the model inside the tools they need: document spaces organised by matter or practice area that return a chronology with every entry cited to the page, a private chat assistant for first drafts from the firm’s own precedents, and custom assistants for the recurring memo. Which fee earners see which matters follows the Microsoft 365 or Google Workspace sign-in the firm already administers.
And what the model never does is advise. Every output is a starting point for a qualified person, and the training says so. Model upgrades are made with the firm as part of support, because a new model changes how drafts read and a litigation team does not want that discovered mid-trial. The case management system stays the system of record.
The Model Is the Engine. This Is the Car.
A bare model is a demo. Law firms need the machine, the tools around the model, the sign-in and someone to maintain it.
A model matched to the work
An open-weight local LLM chosen and tested against your own documents, on a machine specified to run it well for your team size.
Document spaces around it
The model on its own answers from what you paste in. Spaces let it search, summarise and compare across hundreds of your files, citing the page each answer came from.
Chat and assistants on top
A private chat assistant for the team and custom assistants for the recurring jobs, all running on the same machine as the model.
| Document | What the fee earner asks for | What stays where |
|---|---|---|
| Correspondence on a matter | A chronology, every entry cited | On the machine, in the office |
| Bundles and disclosure | A summary; the passages on a point | On the machine, in the office |
| Leases, contracts, agreements | What clause 14 says; the differences between two versions | On the machine, in the office |
| Precedents and house style | A first draft, for a supervisor to review | On the machine, in the office |
| Counsel’s advice and privileged notes | Searchable within the matter, by permitted staff | On the machine, in the office |
| Case management system | Not connected — remains the system of record | Your existing system |
From a Partner’s Call to a Model in the Firm
Weeks, not months. Most firms start with one practice area and widen once it is in daily use.
A call with a partner or the COLP
Practice areas, headcount, which team goes first. We specify the machine from that and walk through the data flow.
Install and import
The machine goes on the firm’s network. Sign-in connects to your Microsoft 365 or Google Workspace. Files, correspondence and precedents for the first team are imported into spaces.
Train fee earners, then support
We train the people who will use it, including what it is not for. Updates, model upgrades and support follow, from the people who installed it.
Where Local AI Is the Wrong Answer
We would rather lose the enquiry than the trust. Three things we tell every prospect before they sign anything.
It is not a frontier model
The largest cloud models are still ahead of any local LLM on hard, novel reasoning. If your work needs that, a local model is the wrong tool and we will say so. For summarising, drafting and answering questions about your own documents, most offices are happy with it.
It is slower than the big cloud models
Open-weight models on a single machine are capable for summarising, drafting and answering questions about your own documents. For frontier reasoning on hard, novel problems, the largest cloud models are still ahead. That is the trade.
It is a weeks-long install, not a sign-up
A cloud agent is live in days. Local AI needs the machine specified, delivered, installed on your network and your documents imported. Weeks, not months — but not tomorrow.
We Run Our Own
Chosen against your files
Model selection is tested on a sample of your own documents, not a public leaderboard.
Sized with the model
The machine and the model are specified together. A model the hardware cannot run well is the commonest DIY mistake.
Upgrades made on purpose
Model upgrades are part of support, so someone is deciding when the answers your team relies on are allowed to change.
UK data protection built in
No third-party AI processor and no restricted transfer for the AI step. UK GDPR, the Data Protection Act 2018 and the ICO as regulator, with a data processing agreement for the support access you grant us.
Which Page Answers Your Firm’s Question
| If your question is… | The short answer | Read more |
|---|---|---|
| How is the model chosen in general? | Tested against your documents, not a leaderboard | Local LLM |
| We want the Local AI overview for solicitors | The head page for law firms | Local AI for solicitors |
| Is this a private LLM for law firms? | Yes — private by construction | Private AI for solicitors |
| What does the COLP need to see? | No processor, no transfer, for the AI step | GDPR Compliant AI for solicitors |
| We would rather host it ourselves | It sits in your building; we maintain it | Self-Hosted AI for solicitors |
| We are a City firm | Managed local LLMs across the capital | Local LLM London |
| Why on-premises at all? | The case for Local AI, in full | Local AI product page |
Local LLMs for Law Firms: Common Questions
What is a local LLM?
A large language model — the kind of model behind chat assistants — running on hardware you control rather than accessed through a provider’s service. Your prompts, uploaded documents and chat history stay on that machine in the firm.
Which local LLM models do you install?
It depends on the job. Drafting and summarising, answering questions across a document library, and structured extraction each favour different open-weight models, and the machine has to fit the model. We shortlist and test candidates against a sample of your own documents during setup.
How do you test a local LLM on our matter files without seeing them?
The testing happens on the machine in your office, after it is installed and your first practice area’s files are imported. We work alongside a fee earner who knows the files; nothing is copied off site, and the support access we use is covered by the data processing agreement.
Can the model cite where an answer came from?
Yes, and for legal work it must. Document spaces return answers with a citation to the page they came from, so a fee earner can check the source before relying on it. A model that cannot do this reliably does not make the shortlist.
Does it give legal advice?
No. It drafts, summarises and finds. Every output is a starting point for a fee earner, and anything constituting legal advice stays with a qualified person. We say so in the training and we would say so to your insurer.
Can it read our case management system?
It does not connect to it. Documents and correspondence are exported into spaces on the machine, organised by matter or practice area. The case management system stays the system of record.
How much does Local AI cost?
£500 per month for the software and the hardware rental. The dedicated machine is rented to your business, not sold: you never buy the hardware. It is installed in your building and runs the private chat, document spaces and assistants. Local AI is for business customers only.
Test the Model on Your Own Files
Tell us your practice areas and which team would test it first. One short call to shortlist the model and size the machine, with your COLP welcome on the line.
Prefer email? sghaith@businessaiagents.co.uk