Every customer capability the product carries, grouped by the work it supports, each line with its real status — checked against the code, generated from the source, never the brochure. You make every call; nothing acts on its own.
Roadmap items aren't listed here as if they were done — book a demo and you'll see exactly what runs.
Your matter stays anchored to its own files, jurisdiction and audit record. Local-first document access, a confidentiality gate on every external route, market-walled legal logic, explicit human approval and a signed record sit underneath every capability. Those foundations are explained separately because they protect every action; they are not another list of buttons.
See all product foundationsSRA Code + LPP framing, in a solicitor's voice
N1, N9, N9A, N9B, N180, N181, N215, N244, N265 — reachable, previewable and downloadable as real PDFs (keyless demo included), and "Prefill from matter" now populates the real AcroForm fields (claimant, defendant, claim details) from the selected matter. Verified against the actual HMCTS PDFs, not stub maps.
On-demand jurisdiction-native client matter-update letter/email — summarises key developments since last communication, outstanding items, next steps, billing-estimate update; tone matches solicitor's configured voice; client data stays inside the firm boundary (reconciler-gated + egress-gated).
German and Spanish statutory deadline and costs calculations, on the matter screen the way the German fee calculator already is
Point at a case folder, get a sorted and ranked matter back
Reading happens on the solicitor's own machine, ranked by evidential authority, and the app states its own coverage instead of implying it read everything.
Every import ends in a Word file on your letterhead, not a screen that closes
The intake artefact states its own coverage and its own limits inside the document, and it is produced without the case leaving the building — the platforms that do whole-folder ingestion at this quality are cloud eDiscovery suites priced for litigation departments.
MLR sanctions screen at the intake gate against the OFSI (HM Treasury) UK consolidated list — import-bound: the firm imports gov.uk's list and re-imports to refresh it (no auto-fetch / server proxy), so every new matter is screened against the last import, and the result honestly shows "not performed" until a list is imported (never a false clear).
One-click deep-link to `sra.org.uk/consumers/register/` for solicitor-status verification at intake (verify regulated, not struck off, no practising-cert conditions). SRA does NOT expose a public JSON API by design — the on-screen register IS the canonical source for the verification fields. Skill `lookup.sra-register` builds the URL with the firm/solicitor query pre-filled, audits the lookup, surfaces structured verification fields. Same manual-lookup helper pattern used wherever a register offers no public API.
Search / profile / officers via same-origin proxy at `/api/registry/companies-house/`
Entity-suffix-normalised scan against firm's existing matters at intake gate
Select a supported file or paste a brief → extract source-grounded facts → review and correct → open the persisted matter
The bench is 15 specialties, each written natively for this market's own law — never carried across from another market and not checked by a practising solicitor. Your firm keeps the ones it does and unticks the rest:
Corporate Law · Employment Law · Family Law · Criminal Law · Property Law · Tax Law · Public Procurement · Immigration Law · Medical Law + Clinical Data · Dispute Resolution · Debt Recovery · Restructuring and Insolvency · Private Client, Wills and Probate · Data Protection and Cybersecurity · Intellectual Property and Technology
Tick the specialties your firm actually practises; everything downstream obeys the list
Every specialty for your market starts ticked; you untick what the firm does not do. A firm working in two legal systems curates each market separately. The list is a gate rather than a menu — a specialist you have not got cannot be put on a matter, and one you withdraw after a matter already had it stops reaching that matter's drafting, chat and adversary review.
Pick the specialists for a matter yourself; the record shows who picked them and when
One lead specialist, chosen by a person, plus any supporting ones — recorded against whoever chose them and when. The practice card beside it may say what the matter looks like; that guess never becomes specialist context and never reaches a prompt. The same chosen context is what intake, drafting, chat, adversary review and the reasoning ledger read.
Add your firm's own material to a specialist, permanently labelled as yours
On Enterprise a firm adds its own document kinds, terms of art, citation forms, deadline dates, amounts and concerns to a specialist it already has on its list. The anchor statutes are deliberately not amendable — a firm adds what it checks for, it does not restate the law. What it adds stays marked as the firm's own on screen, in the prompt and in the exported ledger, and can never be shown as something a lawyer checked.
6-bucket urgency tiles + closest-3-to-expiry + drill-down, with a coverage strip naming the active matters that carry NO limitation cutoff at all. Each market's own limitation vocabulary and law — Limitation Act 1980, KC art. 117–125, §§ 195/199 BGB, art. 3:306–3:310 BW, arts. 1964/1968 CC. A deadline shows the statute the record carries, or says the record carries none; it is never inferred from a parenthetical. Deadlines update live as you edit matter facts, with a manual Refresh button if you want the recount immediately.
Customer-facing 7-named-agent panel (OpponentScout / PrivilegeGuardian / IntakeAgent / CasePreparer / Scrutineer / ComplianceOfficer / agentRegistry). Buyer sees 7 named agents, not 1 chatbot.
Tags privileged docs at upload; static checklist gate before AI invocation
Gates the matter on first intake — extracts parties + jurisdiction + claim type + key dates from raw paste; runs conflict-check + sanctions-check + counterparty registry lookup before matter is created
Assembles draft pleadings and supporting documents from the matter facts, evidence index, and workflow library
Keeps the audit trail — categorises every AI invocation per SRA Code 2.1 + RODO art. 22 + EU AI Act art. 50/52, writes to per-matter audit log + AI disclosure bucket
Pre-ship review of every draft before export — surface-level checks (form validity, party-name consistency, statute presence).
Reads the draft like opposing counsel would — runs attack-vector lenses against the freshly-generated draft and returns a prioritised "what they'll hit on" list sorted by severity. Reviews the draft specifically, as distinct from the matter-level review that scrutinises the whole file.
Per-AI-decision append-only chain-signed record in the browser's local IndexedDB store (bytes never leave the machine), one signed row per AI call (timestamp + skill + provider + model + tokens + cost-usd + input-fingerprint + output-excerpt + chain-link SHA-256, each row additionally HMAC-signed under the firm key). Distinct from the audit-log hash chain, which signs at the audit-row level; this one signs at the AI-decision level with richer per-call metadata. Verifier walks the chain + flags first-broken row. Solicitor-facing panel in the Audit Trail view shows row count + chain-verified status + 4 buttons (Export PDF / MD / CSV / Re-verify). System Check proves the tamper-evidence guarantee through the shared `audit-chain` crypto self-test — the reasoning and audit chains share one sha256 chain primitive, so a single probe covers both.
Composite per-matter risk score (0-10) aggregated from the limitation clock, privilege-egress events, adversary-review findings, the red-flags register, and sanctions hits — recalculated live as the matter changes, with a manual Recompute button. When an input is missing or a calculation fails it reports no score at all and names what is missing, rather than showing a low one.
Identifies the opposing solicitor from the counterparty dossier, cross-references a growing pattern library, and surfaces known filing patterns, preferred strategies, win/loss record in similar matter types, and argument templates used. "Know how they'll fight before the first letter."
Specialised research mode — case-law, statute, and regulatory-guidance synthesis with cited grounding. Where the drafting and review agents write and check, this one answers legal-research questions with citations.
Per-practice-area template catalogue with placeholder-fill + clause library. Mirrors W-01 workflow library shape. UK + PL native libraries.
Ask the same questions across many matter documents at once and get an answer grid — one source-linked answer per document, with privileged files automatically withheld from the AI.
Bulk document review as a source-linked grid — review a stack of exhibits against the same questions in one pass, privilege-gated per document.
Upload a Word document and get it back with the AI review findings written into the margin as inline comments - your original formatting untouched.
The AI findings land as real Word comments in the solicitor own document, not a separate report.
The cross-exam plan, evidence outline, chronology and settlement memo a prep run produces are kept on the matter, signed off by a named fee-earner, and flagged the moment a fact edit, a new document or a contradiction leaves them out of date.
The pack tells you when it has stopped matching the matter, and refuses to be signed off while it does — most tools hand you a generated document with no idea whether the file moved under it.
LBA / Part 36 / N225 / N228 / Witness Statement / N244 (6 sub-types) / N265 / Precedent H / CMC/CCMC/PTR / Mediation Brief / Part 8 / Freezing/Search/Bayer / new-matter-intake / prepare-for-hearing / cross-examine-opponent / prepare-for-trial / close-matter / compliance-snapshot
`is_lip` flag + `litigantInPersonBlock` helper auto-prepended to all 18 drafters; first-person voice, no firm letterhead, CPR 46.5 LiP-rate-cap branch on Precedent H
Extracts the verification marker in every market — `[VERIFY …]` (UK) / `[DO WERYFIKACJI …]` (PL) / `[ZU PRÜFEN …]` (DE) / `[TE VERIFIËREN …]` (NL) / `[POR VERIFICAR …]` (ES), plus the draft and completion markers — holding the citation whole, brackets and all, so the marker names the authority exactly as pleaded; per-marker verification progress; gates downloads/signing until the solicitor ticks every citation. A document in which no authority was found reports "nothing was verified", never a clean pass
After a matter closes, with solicitor consent, approved pleadings are stripped of client PII and added to a searchable firm-wide precedent library; future matters automatically surface the three closest precedents on Overview. The bank grows more valuable with every matter, and it never leaves the firm.
Solicitor calendar with 4 toggle-able colour layers — court dates / statutory deadlines / client meetings / internal deadlines. Month + week views (toggle in header). Click event → matter or task. Reads from existing matter dates + limitation dashboard + audit-log scheduled-actions.
Drag-card-to-reassign Kanban across the firm caseload — 7 litigation-workflow columns (intake → pre-action → claim-issued → defence-stage → directions → trial-prep → closed/settled). Each matter's stage is derived from its status plus its drafts and is overridable by dragging a card, persisted to the matter so the move survives reload. Native HTML5 drag-and-drop (no external drag library). Card shows matter ref, jurisdiction, client, matter type, risk and last activity; click a card opens the matter. Runs at the firm-dashboard level.
Horizontal-axis events per matter across 5 lanes (matter/statutory dates, compliance gates, drafts filed, evidence in, AI consultations), with start-to-end range bars for dated windows. Most of what a Gantt chart gives you, without the project-management baggage.
Auto-generated entity graph per matter — parties / witnesses / counsel / court / claim / dates / documents / evidence as nodes; opposes / represents / mentions / filed / received / cites as edges. 4 clusters with collapse toggles. Click node → highlights incident edges + detail panel.
Withhold ANY privileged matter file from AI / CRM egress — the privileged file never leaves the firm boundary, while the assistant still answers from the matter's disclosable, PII-pseudonymised documents (and names the withheld files). PL absolute (tajemnica zawodowa art. 6 PoA + art. 3 uoRP); UK strong (privileged + unknown). Unknown class treated AS IF privileged across every external destination — solicitor must classify the file or explicitly confirm per-egress.
59 identifier classes across all five markets, reversible in the browser, applied to the message, the system instruction, retrieved documents, every tool result and the replayed history — with two named exceptions (dictation clean-up, screenshots)
Per-matter log of every AI invocation; export disclosure pack to client (SRA 2.1 + RODO art. 22 + EU AI Act art. 50/52)
One-click "Generate DSAR Pack" → machine-readable JSON (art. 20 portability) + statutory PDF cover
The assistant reads every document you add, and this is where that reading adds up — a chronology, the issues, and which document speaks to each. It tells you at the top how much of the file it actually got through, and names anything it could not read or was not allowed to send.
It states its own coverage before its conclusions, so a picture built over a half-read bundle can never be mistaken for a complete one — and in Poland, Germany, the Netherlands and Spain the privileged documents are analysed on the solicitor's own device and never sent for assembly at all.
Every matter keeps its own append-only audit log, approvals record, and change history
Per-matter `AUDIT.hashchain.jsonl` — one SHA-256 chain link per AUDIT.md row (`{seq, prev_hash, content_hash, chain_hash}`). Daily HMAC-signed seal in `AUDIT.dailyseal.jsonl` per UTC day. Verifier walks chain + flags first-broken row + reason ("row edited / row removed / chain reordered"). Non-destructive — AUDIT.md unchanged for solicitors / auditors. SoC-grade tamper-evidence. Distinct from P4 Reasoning Ledger (lighter scope, ships now).
Every document moves DRAFT → REVIEWED → APPROVED → SIGNED → FILED → SERVED with role-gated transitions and an append-only approvals trail; one status across every view
Per-matter audit log → filter (time / source / severity / free-text) → CSV + PDF export. The PDF carries the regulator header for the firm's own jurisdiction — for an England & Wales firm, ICO + SRA + COLP with the 6-year retention basis (SRA Code 6.4 + Limitation Act 1980 s.5) — plus **tamper-evidence integrity status** (✓ verified intact / ⚠ broken at row N because X) and a signature footer. Solicitor-signed evidence ready for regulator scrutiny.
3-figures form (bank / cash book / sum of client ledger) + live discrepancy classifier + 5-week mandate countdown + COFA PDF
Firm-wide unpaid-invoice tracker, 4 age buckets (current ≤30d / overdue 31-60 / 61-90 / severe >90)
Derives the matter's strength tier, scans similar matters with definitive outcomes, and surfaces "X of Y similar matters settled — Z%", with an insufficient-cohort fallback
Per-fee-earner / per-firm analytics: matters worked this period, AI invocations used, estimated drafting time saved, utilisation rate vs target. Quantifies ROI for the renewal conversation ("your team saved X hours this {period}"); all data local, zero telemetry.
Drop a folder or a zip — 22 formats are read (PDF, Word, Outlook .msg, .eml, Excel, RTF, XML, HTML, WhatsApp exports, scans via OCR) and the 9 a browser cannot read say so by name and tell you what to do instead
Drop a folder or ZIP into the matter and its readable contents become searchable case material for source-linked drafting, answers and review. Scanned PDFs go through OCR rather than disappearing as empty pages; formats a browser cannot safely read are named, refused and paired with the next practical step.
Zero vendor domains visible in solicitor browser; live E2E validator 7/7 pass
Solicitor downloads filled PDF → opens signing portal new tab → signs → attaches signed PDF back
33-statute lookup + anchor+walk-back parser + lookup-aware truncation + self-check probe
Solicitor drafts inside Word; drafter, adversary pass, opponent intel, firm-playbook check, tracked-change redlines and filing to the firm's document system are all callable from the Word task pane against the chosen matter. The matter's own jurisdiction picks the drafter, so an English, Polish, German, Dutch or Spanish matter each produces its own court document rather than an English one. Output writes back to the selection only on an explicit click, and not at all if the response came back refused or incomplete or the selection moved while it ran.
Word remains the solicitor's working surface while Avantwerk brings in the selected matter's jurisdiction, evidence and firm rules. It can draft, challenge a selection, apply the firm's playbook, create tracked-change redlines and file the open document — but writes back only after an explicit click and never from a refused or incomplete response.
Adds a fourth bring-your-own-key AI provider for chat drafting, alongside the existing providers (separate from the on-device voice-transcription option that already ships)
Solicitor sees Avantwerk inside Outlook on read-message / compose-message / calendar-attendee / calendar-organiser surfaces, each handled as itself. Match this email or calendar entry to a matter (an uncertain match is offered, never pre-selected), file it to correspondence once however many times you click, draft a matter-anchored reply in the matter's own language, and read deadlines out of the text in that market's own date forms for confirmation before any reach the timeline. The pane opens on the item in front of you; it does not watch your mailbox.
Outlook works on the message or appointment the solicitor has opened. It proposes a matter without silently choosing an uncertain match, files the item once, drafts a reply from that matter's context and turns detected dates into proposed deadlines that still require confirmation.
Excel ribbon → "Open Avantwerk" task pane → choose the market → select a column of that market's counterparty identifiers → each row is checked against that market's own register (Companies House for the UK with the firm's own credential; Biała Lista VAT plus KRS officer enrichment for Poland, both open) and, if the box is ticked, screened against the consolidated sanctions list the firm has imported → results written beside the selection. The German, Dutch and Spanish company registers are not connected — those rows say so and link to the register, and every row that could not be checked is flagged, so a blank never reads as a pass. Uniquely Excel-shaped — neither Word nor Outlook handles spreadsheet-driven batch CDD.
Excel handles the job that belongs in rows rather than prose. Select a column of counterparties, run the registers available for that market and optional sanctions screening, then receive the result beside each row; an unavailable check is visibly unavailable, never an empty cell that looks clean.
Pay by card inside the app — checkout opens on a secure page hosted by our payment processor, card details never touch the product, and the subscription or top-up activates the moment payment completes. Referral credits and prepay top-ups run on the same billing spine.
On a UK VAT number the product checks the format (9 or 12 digits), then opens HMRC's own public "Check a UK VAT number" service so you can confirm the registration and record the registered name, address and date checked back onto the matter. It does not look the number up for you: the check is yours, made on HMRC's page, and the product keeps the record of it. Automatic resolution of the registered name and address is not available yet - it needs production access from HMRC, which is still under review with them, and until that lands this stays a guided check rather than an automated one.
Tier-1 / mid-market firm enterprise DMS — UK iManage is dominant; many large firms refuse SaaS without DMS connector. Near-term build — pulled into the active F1 phase (operator 2026-07-08); no longer first-firm-gated. Surfaced 2026-05-05.
VPS-isolated single-tenant deploy for firms refusing multi-tenant SaaS. Hard procurement gate at Tier-1. Surfaced 2026-05-05.
Bridge to Microsoft Copilot ecosystem so firms standardising on Copilot don't see Avantwerk as competitive. Read-only context exchange. Surfaced 2026-05-05.
The honest way to know what runs is to watch it. The populated demo takes a fictional matter through it in minutes. No card, no contract — a work email and your firm’s name, so we know who we’re talking to.