Employment Tribunal case preparation for law firms
Your clients bring the facts. Your team does the law.
ClaimOS sits in between. Invite each client into a secure workspace for their matter, see every tribunal matter and key date across your department, and keep fee earners on the work that needs a lawyer.
Document-automation software licensed to law firms. It describes what the documents say; it does not assess merits, predict outcomes, or give legal advice, to your firm or to your clients.
Client · 6 of 9 items receivedAK · approved 2 · rejected 1Limitation 24 Dec 2026 · ERA 1996 s.111(2)(a)
Appeal outcome sentQuotation not found on the cited page. Rejected.C/9
Fictional matter · illustrative only
Documents arrive through the client workspace and are logged on receipt.
PayslipsContract of employmentGrievance letterDismissal letterET1 claim formET3 responseAcas early conciliation certificateMeeting notesCall recordingsAbsence recordsAppeal outcome
See it work
One file, seven stages, one source
A fictional matter moving through the seven stages, from intake to hearing bundle. Each stage leaves an artefact behind, and the chronology, the dates and the bundle are all built from the same vault, so they cannot drift apart. Press play, or pick a stage.
Fictional ET matterWorking file
Illustrative product view
Fictional data and references · no output shown here has been approved · every output is working material for a fee earner to check
Uses AITranscription · Chronology · Review lenses
Each stage, and what it leaves behind
StageWhat happensWhat it leaves behind
InIntake
Intake
Documents and audio arrive on the matter, from your team or through the client workspace, and are logged on receipt.
Receipt log
02TranscriptionUses AI
TranscriptionUses AI
Audio becomes searchable text, held against the recording it came from.
Transcript
03Evidence vault
Evidence vault
Indexed, versioned, searchable, access-logged. The single source for everything after it.
Matter index
04ChronologyUses AI
ChronologyUses AI
AI proposes dated events from the vault, each kept with the passage and page it came from.
Cited chronology
05Review lensesUses AI
Review lensesUses AI
Focused checks report procedural, evidential and consistency findings; a separate comparables view adds context.
Review findings
06Deadlines
Deadlines
Working shown and rule cited, for a fee earner to check.
Key dates
OutBundle
Bundle
Paginated and indexed from the vault, not assembled by hand.
Hearing bundle
Illustration · every output is for a fee earner at your firm to check before it is used
Where the time goes
A tribunal file arrives in pieces.
Payslips in one email, a grievance letter in another, a timeline half-remembered on a phone call. Someone has to chase it, log it, date it and put it in order before anyone can think about the case. Little of that needs a lawyer's judgement. Much of it lands on a fee earner's desk anyway.
Chasing documents by email
Re-keying dates from letters
Building the chronology by hand
Working out what is still missing
How it works
Your client contributes. ClaimOS organises. Your team decides.
01
Your client contributes
Your firm invites the client into a workspace for their matter. They see plain-English requests for the documents you need and the questions you want answered, and upload from their phone or computer. They see only their own matter, and only what your team chooses to share.
02
ClaimOS organises
Everything is logged on receipt into the matter's evidence vault. ClaimOS proposes a source-linked chronology, reports evidential and consistency points for review, and shows its working for every key date.
03
Your team decides
Fee earners review what has come in, approve what belongs in the case, and decide what goes back to the client. A client sees the requests your firm sends, and sees what ClaimOS produces only if your firm chooses to share the whole workspace on that matter.
One matter, two views
What your client sees. What your team sees.
The same matter, from both sides of the screen. Your client sees a short list of what is needed and what has arrived. Your team sees everything that has come in, what it says, and what still needs a decision.
Acting for an employee
Client workspace
Your Firm LLP · Your matter (0142)
6 of 9 items received
What we need from you
Payslips, July to September 2026requested 29 Sep
Messages with your manager about the meeting on 18 Septemberrequested 29 Sep
When did you first raise concerns at work? An approximate date is fine.
Received (6)
Contract of employment30 Sep
Dismissal letter30 Sep
Grievance outcome letter1 Oct
+ 3 earlier items
Your solicitor reviews everything you send. For questions about your case, contact your solicitor directly.
Firm matter view
Matter 0142 · Unfair dismissal · Acting for employee
Limitation (before early conciliation): 24 Dec 2026
Effective date of termination 25 Sep 2026 is before 1 Oct 2026, so three months less one day applies
ERA 1996 s.111(2)(a)Working shown · fee-earner check required
From client, received (3)
Contract of employment4 pp · logged 30 Sep
Dismissal letter2 pp · logged 30 Sep
Grievance outcome letter3 pp · logged 1 Oct
Proposed chronology entries, awaiting review (4)
18 Sep 2026Meeting with line managerSource B/2
25 Sep 2026Dismissal letter datedSource A/1
plus 2 more
Shared with client
Request list (3 items) · shared by AK · 29 Sep
Try it: send an item as the client, or accept an entry as the fee earner — the other view updates.
Acting for an employer
Client workspace
Your Firm LLP · Your matter (0157)
5 of 8 items received
What we need from you
Absence management policy in force in March 2026requested 24 Sep
Notes of the return-to-work meetingrequested 24 Sep
Who attended the appeal hearing, and in what role?
Received (5)
Contract of employment25 Sep
Absence records25 Sep
Outcome letter26 Sep
+ 2 earlier items
Your solicitor reviews everything you send. For questions about your case, contact your solicitor directly.
Firm matter view
Matter 0157 · Disability discrimination · Acting for employer
Response (ET3) due: 20 Oct 2026
28 days from the date the claim was sent (22 Sep 2026)
Working shown · fee-earner check required
From client, received (3)
Contract of employmentlogged 25 Sep
Absence recordslogged 25 Sep
Outcome letterlogged 26 Sep
Proposed chronology entries, awaiting review (3)
Shared with client
Request list (3 items) · shared by RT · 24 Sep
Try it: send an item as the client, or accept an entry as the fee earner — the other view updates.
Fictional matter · illustrative only · invented dates, not a calculation of any actual time limit
Language models help with the parts of case preparation that require reading language. Deterministic code handles citation verification and date arithmetic. ClaimOS shows the distinction so a fee earner can inspect the work rather than take an answer on trust.
AI workflow 01
Build a source-linked chronology
ClaimOS proposes dated events from the processed file and carries the supporting passage with each one. The quotation is machine-checked against the stored page; a citation that cannot be verified is rejected rather than presented as fact. A fee earner decides what belongs in the working chronology.
Flexible-working request sent12 Jan 26A/3
Written response received28 Jan 26A/7
Grievance submitted04 Mar 26B/12
Illustrative chronology · references not approved
AI workflow 02
Use focused review lenses
Procedural, evidential and consistency checks run in the review panel, and comparable published decisions sit in their own search view. ClaimOS says where model-assisted reading, deterministic checks and public-record retrieval are used. It reports material for a lawyer to inspect — never a case score, prediction or view on merits.
Ask a factual question in ordinary language and receive an answer grounded in the documents, with quotations linked back to their pages. Completed answers retain the question, answer, available citations, user and timestamp in the matter record. The tool describes the file; it refuses requests for advice, strategy, prospects or recommendations.
When was the flexible-working request answered?
The response letter is dated 28 January 2026.
response-letter.pdfquoted passagep.1
Completed answer and citation retained
Illustrative answer · question and sources retained
AI workflow 04
Keep an AI record of every model operation
Each matter carries its own AI record: the questions asked of the evidence with their answers, citations and verification state; each review run and who commissioned it; and a ledger of every model-assisted operation with its model, prompt version, tokens, cost and outcome. It records sources and verification state, not hidden model reasoning, and can be exported with the matter.
Select the heads being claimed and enter the figures. The basic and compensatory awards, notice pay, unpaid wages, holiday pay and injury to feelings are calculated as plain arithmetic, each figure shown step by step with the rule applied, using the statutory caps in force on the effective date of termination. Which heads apply, and what a tribunal would award, are questions it does not answer.
Basic awardERA 1996 s.119
Compensatory awardERA 1996 s.123
Notice payworking shown
Illustrative schedule · figures for fee-earner review
01 / The evidence, connected[ + ]
01p.3Request
02p.7Response
03p.12Grievance
Working chronology
12 Jan 2026Request sentSource · p.3
28 Jan 2026Written responseSource · p.7
04 Mar 2026Grievance submittedSource · p.12
For fee-earner review
Every entry has a source.Every source stays in view.
Fictional references · illustrative only
Integrations
Clio Manage, linked matter by matter.
A firm administrator connects the firm's Clio Manage account once. A fee earner then links a ClaimOS matter to its Clio matter, and sends that matter's dates and exports to it. Clio stays the system of record.
Status:Built, and not yet switched on. It is switched on after our data-protection review of sending matter data to Clio; a firm's administrator then connects the firm's Clio account.
MattersClio to ClaimOS
The firm's Clio matter list is searched by display number or description to make the link.
DatesClaimOS to Clio
The matter's dates, as its calendar download carries them, become entries on the Clio calendar of the person who connected Clio for the firm, against the linked matter. Sending again updates the same entries. ClaimOS sets no reminders.
DocumentsBoth ways
The chronology and finished hearing bundles are uploaded to the linked matter's documents when a fee earner sends them. A fee earner can also choose a document in the linked Clio matter to bring into the vault; nothing comes in unless they choose it.
ContactsNot exchanged
No contacts or correspondence come in from Clio, and nothing is kept in sync. A client's uploads reach Clio only inside an export your firm chose to send.
ClaimOS produces inspectable working materials. It does not act on a finding. Material reaches a client only when your firm sends or shares it. Fee earners can follow citations back to the source, review the retained activity history, and decide what is approved for use.
What ClaimOS is not
Not a law firm. ClaimOS does not give legal advice.
Not a claims management company, and not regulated by the FCA.
It does not assess the merits, value or prospects of any claim.
It does not take instructions, advise your clients, or make or receive referrals. It contacts a client only to deliver an invitation or request your firm has sent.
Every output — every date, every chronology entry, every bundle — is working material for a fee earner at your firm to check before it is used.
Built for a firm that will be asked about it
The matter database is hosted in London; evidence files are stored in Western Europe. Evidence text is also processed by the AI and document services named in our privacy notice.
PromptWorks AI Ltd is registered with the Information Commissioner's Office, reference ZC229196.
Client data is not used to train models.
Row-level tenant isolation, enforced in the database and tested as a build gate.
Two-factor authentication is mandatory. Evidence views and downloads are logged.
Clients see only their own matter. Isolation is enforced in the database itself, by row-level security, as well as by the application.
Files uploaded through the client workspace are scanned for malware before anyone can open or download them, and before they are processed or put in a bundle.
Where data goes, and which sub-processors handle it, is set out in the privacy notice.
Deadlines
The arithmetic just got harder
Three time-limit rules now each have two versions, and which one applies turns on a date in the matter — not on the date the claim is prepared.
Heads of employment. Every tribunal matter, key date and outstanding client request across the department, in one view.
Fee earners. Client material arrives logged and organised, with a source-linked chronology to review.
Your clients. A clear, private view of their own matter and what you need from them.
How does ClaimOS fit with the systems we already use?
ClaimOS does not replace Access, Clio, Leap or Proclaim, and does not need your accounts, billing or client ledger. It handles the evidence-heavy preparation work those systems were not built for, while the system of record stays where it is.
The firm view is a tribunal-preparation view (dates, evidence and review), not a practice-management system. Time, billing and conflicts stay where they are.
Does ClaimOS work with Clio?
A Clio Manage integration is built, and not yet switched on. When it is, a fee earner links a matter here to its Clio matter and sends the matter's calculated dates, chronology and finished bundles to it. A document comes in from Clio only when a fee earner chooses it, and nothing is kept in sync.
It is switched on after our data-protection review of sending matter data to Clio. Until then, ClaimOS sits alongside Clio without connecting to it.
How much does ClaimOS cost?
Plans start at £450 a month, billed annually, for 5 users and 60 tribunal matters a year. Every plan includes every feature and unlimited client workspace users. See Pricing. Pilot discounts are available on request.
What does a pilot involve?
A pilot runs for 90 days on 5 to 10 of your live or recent matters, under a short written agreement. We set it up with your team and review the results with you at the end. If you go on to subscribe, the pilot fee is credited against your first year.
Who is responsible if a calculated date is wrong?
Checking each date is your firm's responsibility. ClaimOS calculates each date from the published rule and the dates on the matter, and shows the rule and its working so a fee earner can check it.
It does not give legal advice, and every output (every date, every entry, every bundle) is for your firm to check before it is relied on. A firm's use on client matters is under its written agreement with us, which governs the licence, service levels and data processing.
What will our insurers and the SRA think?
ClaimOS is software your firm licenses. It isn't a law firm and it doesn't give legal advice, so your firm keeps professional responsibility for the work, just as it does with any practice-management or document tool. Every AI output links back to its source document. A citation that can't be verified is rejected, and each AI operation is logged in the matter's AI record, so a fee earner can check and evidence what was relied on. That record is something you can show your insurer or your COLP. We're happy to answer your insurer's or your risk team's questions during a pilot.
Is there a data processing agreement, and who are the sub-processors?
Where a firm licenses ClaimOS, the firm is the controller for its own client data and we act as its processor, under the data processing terms in its agreement with us.
The application database is hosted by Neon in London, and evidence storage is on Cloudflare R2 in its Western Europe location. Where the application processes evidence text with a model, that request goes to Anthropic's API; audio transcription and document text extraction use Microsoft Azure services. The sub-processors are set out in full in the data processing terms provided to licensing firms.
Which AI does ClaimOS use, and what does it see?
Evidence text is processed with Anthropic's models, through Anthropic's API: for the chronology, the review lenses, questions asked of the evidence, and dates read from a document, which are proposed for a fee earner to confirm. Audio transcription and document text extraction use Microsoft Azure services.
The deadline arithmetic, the vault and the bundle use no model. Each model run on a matter is listed in that matter's AI record, with what it cost.
Who built this
Built by PromptWorks
ClaimOS is a product of PromptWorks AI Ltd, a UK software product studio. Founder-led, clearly scoped, built with the people who use it.
We'll walk through a complete fictional matter in the firm view and the client workspace: the requests, the uploads, the chronology, the review lenses, the deadline working and the bundle. If it fits your practice, we agree the scope of a controlled pilot.