We ship & support it · nothing leaves your network
The answers are already in your files. Now you can find them.
FileFerret is private AI search across your firm’s documents, images, audio, and video
— delivered as a turnkey appliance we ship, install, and support for years. Search by
meaning, ask questions and get cited answers, without a single client file ever leaving your network.
Trytax documents from 2023ai_sentiment:negative**invoice*
Shipped, installed & supported. Indexing, search, and AI all run on a private appliance on your premises.
Hardware + software, delivered
Ongoing support & updates
Nothing in our cloud
What self-hosting solves
The cheapest way to pass an audit is to remove the third party.
A stack of regulations makes you responsible for vetting and
certifying every vendor your client data passes through — and every one of their
subprocessors. FileFerret keeps the data on a box you control, so the third parties
— and the obligations that come with them — simply aren’t there.
No vendor chain to vet or certify
Put client data in the cloud and it flows through hosting providers, AI vendors, and their subprocessors — and the rules make you responsible for diligencing and contractually binding every link in that chain. On a FileFerret appliance there is no chain: the data never leaves your network, so there is nothing — and no one — to certify.
Anyone who touches protected health information has to be under a signed agreement, and so does every one of their subcontractors. Self-hosting keeps PHI inside your walls, so there is no business associate to contract with, audit, or be breached on your behalf.
Defense and controlled-data mandates restrict where data may live, who may access it, and forbid uncertified or foreign-hosted clouds. Running on a box you physically control satisfies the boundary by construction — no cloud authorization package to inherit or maintain.
Confidentiality & supervision duties, met by design
Professional-conduct and safeguards rules require reasonable efforts to protect client information and to supervise any third party who handles it. With no third party in the loop, the supervision obligation is satisfied at the root rather than managed vendor by vendor.
Regulations are named for orientation only and vary by jurisdiction,
data type, and engagement; this is not legal advice. Confirm your obligations with counsel.
Why this matters
The problem isn’t the AI. It’s where the AI sends your files.
Consumer and cloud AI tools are powerful — but using them on confidential
client material means handing that material to someone else. Here is what that exposes a firm
to, and how an on-premises appliance we ship and support removes each risk at the root.
Client confidentiality
Problem
Upload a privileged file to a cloud service and strangers — vendor staff, subprocessors, support engineers — can technically reach it. You no longer hold sole custody of your clients’ secrets.
Solution
Indexing, search, and AI all run on the appliance we ship you, inside your own network. No client document is ever copied to an outside cloud or public AI service.
Cost of doing nothing
A single confidentiality breach can mean lost clients, bar or board complaints, and reputational damage no settlement undoes.
Effort to fix
Nothing to upload and nothing to opt out of — the files simply never leave the building.
Liability you can’t sign away
Problem
Cloud and AI terms of service routinely make you indemnify and hold them harmless — so when their platform leaks or misuses your client’s data, the liability lands on your firm.
Solution
With no third-party custodian in the loop, there is no one-sided contract shifting their breach onto you. You set the terms because you hold the data.
Cost of doing nothing
You can be left financially and ethically on the hook for a breach you neither caused nor could have prevented.
Effort to fix
No legal review of fine print, no risk acceptance — the indemnity trap is gone because the vendor is gone.
Your files, their training data
Problem
Many providers reserve the right to use what you upload to train their models or for marketing — in whole or in aggregated form — turning your clients’ confidential material into someone’s product.
Solution
On-premises by design: your documents are never an input to anyone else’s model, dataset, or marketing — because they never reach them.
Cost of doing nothing
Privileged work product and personal client data, irreversibly absorbed into a third party’s systems you can never claw it back from.
Effort to fix
Prevented by architecture, not by toggling a setting you have to trust them to honor.
Service disruptions & lock-in
Problem
Outages, suspended accounts, sudden price hikes, or a sunset product can cut you off from your own files at the worst possible moment.
Solution
FileFerret runs on the appliance we ship and support, kept on your own premises — it works even when the public internet doesn’t, on your schedule, not a vendor’s.
Cost of doing nothing
Lost billable hours and blown deadlines every time a provider goes down, changes terms, or locks your data behind a paywall.
Effort to fix
You control uptime and upgrades; no migration scramble when a vendor changes the deal.
Data offshore & in transit
Problem
Cloud data is often replicated to offshore data centers — permanently, temporarily, or sporadically — and travels the open internet, where it can be intercepted in a man-in-the-middle attack.
Solution
Your files stay physically inside your walls, under your jurisdiction. Nothing crosses the wire to an outside service, so there is nothing to intercept.
Cost of doing nothing
Loss of legal jurisdiction and physical control over client data, plus exposure to interception you can’t see or audit.
Effort to fix
No data-residency questionnaires and no network egress to lock down — the data has nowhere to travel.
Limited legal recourse
Problem
When a mega-vendor leaks your data, mandatory arbitration clauses and liability caps buried in the ToS leave you with little practical recourse.
Solution
No third party is ever entrusted with the files, so there is no one-sided contract — and no breach by an outsider — to litigate.
Cost of doing nothing
A breach you can’t meaningfully recover for, while your clients still hold you accountable.
Effort to fix
Removes the third party from the equation entirely, so recourse is never the question.
Runaway, unpredictable costs
Problem
Cloud and AI tools bill by subscription tier, per seat, and usage or pay-as-you-go metering — so the more your team and your data grow, the more you pay, and renewals reset the price upward.
Solution
We ship you the hardware and software together and support it for years: a fixed, predictable monthly cost with no per-seat, per-query, or pay-as-you-go metering — search as much as you like for the same price.
Cost of doing nothing
Budgets blown by surprise overage charges and annual price hikes you can’t forecast or control.
Effort to fix
One known line item instead of a metered bill — no usage dashboards to police and no overages to ration.
Durability & offsite backup
Problem
A single cloud copy you don’t control can vanish in one outage, ransomware event, or closed account — taking your only copy of the record with it.
Solution
Your index is stored redundantly, and you choose where encrypted offsite backups live — your home, another office, a trusted partner, or a best-of-breed secure-storage provider — so a recoverable copy always exists under your control.
Cost of doing nothing
Catastrophic, unrecoverable loss of client records — the kind of event a firm often does not survive.
Effort to fix
Automated redundant and offsite copies, configured once to the locations you already trust.
How we compare
Capabilities and cost, side by side.
How FileFerret stacks up against the tools firms usually weigh it against
— public cloud AI, cloud eDiscovery and review platforms, and cloud document
management. Representative products are named for orientation only.
Capability and cost comparison of FileFerret against public cloud AI, cloud eDiscovery, and cloud document management tools.
Capability
FileFerretShipped & supported
Public cloud AIChatGPT, Copilot, Gemini
Cloud eDiscoveryRelativity, Everlaw, DISCO
Cloud document mgmtiManage, NetDocuments
Runs on-premises on hardware we ship you
✓Yes
×No
×No
×No
No client file ever leaves your network
✓Yes
×No
×No
×No
Semantic search with cited answers
✓Yes
✓Yes
–Partial
–Partial
Searches audio & video (transcribed)
✓Yes
–Partial
–Partial
×No
Your files never train the vendor’s models
✓Yes
×No
–Partial
–Partial
Per-client ethical walls & full audit log
✓Yes
×No
✓Yes
✓Yes
Vendor can support the box but can’t read your files
Comparison reflects the standard, publicly documented offerings of representative products and is provided for general guidance; specific plans and configurations vary. All product names are trademarks of their respective owners.
How it works
From a shelf of files to a searchable practice — in three steps.
01
We ship & install it
We deliver a preconfigured FileFerret appliance to your office and point it at the documents, images, audio, and video on the storage you already run — all inside your own network.
02
Search and ask
Staff search by meaning or ask questions in plain language, and get answers drawn from your files with every source cited.
03
Nothing leaves
Indexing, search, and analysis all run on-premises. No client file is ever sent to an outside cloud or AI service.
A data-flow view of FileFerret. Every step happens inside your own network — no client file is ever sent to an outside cloud or AI service.
Capabilities
Everything a firm needs to put AI to work on confidential files.
Semantic search
Find by meaning, not just keywords.
Ask in plain language and get back the documents that actually match the idea — ranked by similarity, across every file you have indexed, with keyword and wildcard filters when you need precision.
Pose a question over a matter and get a written answer assembled from your own files — every claim linked to the exact document and passage it came from, so nothing is taken on faith.
FileFerret reads far more than text. Scanned pages are run through OCR, recordings and video are transcribed, and images are described — so a phrase buried in a voicemail or a scanned exhibit is just as findable as one in a Word file.
Turn a search into a case file. Group results into named projects, flag the files that matter with an interest score and a note, and export the tagged set as a clean spreadsheet for review or discovery.
Each client’s material lives in its own collection. Administrators decide exactly which staff can reach which client, so a search never crosses a boundary it should not.
Every login, search, file open, and admin action is written to an activity log you can review — the kind of accountability your clients, insurers, and regulators expect of a firm holding their most sensitive material.
Ask a question. Get an answer your partners can stand behind.
Pose a question over an entire matter and FileFerret assembles the answer from your own
documents — with every statement linked back to the exact file and passage it came
from. It reads scanned exhibits, transcribes recordings, and never sends a word to an
outside service.
We’ll scope your firm’s needs, ship and install a private FileFerret appliance,
train your team, and support you for years to come. No client data leaves your office — ever.