The full shipping log: every change that has landed in the product, newest first, each with its real ship month.
Vorenq is in invite-only beta: we onboard a small number of Swiss companies personally and ship weekly. There are no customer logos on this page. Instead: the company behind it, the hosting region, every sub-processor by name, and this log. What the product can't do yet, we tell you before you pay.
Search and filter results across all 21 industry tools, with complete totals and matching CSV exports. Download source records, review date and follow-up notes as a local work file to reopen later; findings are recalculated when opened. Compare two supplied reviews without treating absent rows as resolved work. Saving is explicit, with no autosave, new service or AI credits. Optional AI drafts still use normal credits.
Record the manual time a result replaced, then subtract your review and correction time. Weekly summaries and the Manager show measured savings separately from estimates, including work that took extra effort. Review recurring-work changes in the Manager before confirming them.
Track milestone handovers, materials gaps, unit contribution, prepaid sessions, fee milestones, harvest allocation or administrative request turnaround with your own records and rules. All local tools now support direct row entry, blank CSV templates and a CSV of every result alongside the report and action plan. No AI credits, new service or storage are added; optional AI drafts still use normal credits.
Each industry now has another tool: budget checks, quote lines, reorder quantities, appointment overlaps, document revisions, delivery settlements or administrative document receipt. Calculations use your supplied rules and records. Organise open findings into a downloadable action plan with owners, dates and review status. Everything stays in the browser without AI credits; plans do not update source systems or start AI work.
A first set of local CSV checks and 21 manual workflow drafts now covers services, trades, retail, salons, architecture, agriculture and practice administration. Review records in your browser without AI credits, then optionally prepare a sourced draft with an existing employee. No extra subscription or employee is required; AI execution uses your existing credits. Dedicated industry pages explain the workflows and their limits.
Employees now correct their own mailbox requests instead of stopping. Gmail invoice attachments are recognised reliably, one Gmail label can be listed on its own, and Gmail searches accept a date window. Moving an Outlook meeting onto another appointment is refused, like a new booking. Sent Outlook mail is confirmed even in busy shared mailboxes.
The Manager's current work, the shared to-do list and proposals now have distinct, simple areas. Approvals and questions link directly to the right decision. Details, the task board and history remain available.
Connected Studio and pinned apps can refresh their data automatically while open, with pause controls and clear freshness status. Previews stay visible during updates, unchanged data keeps your filters intact, and app search makes saved tools easier to find.
Outlook folder discovery now handles accidentally included message filters, sorting and empty category fields without stopping. Folder totals are clearly unfiltered; your requested filters still apply when selecting messages. Follow-up chats also attempt to recover from correctable read requests, and unresolved work stays marked incomplete.
Employees try to correct invalid requests to connected apps and continue after a successful correction. Genuine blockers and unfinished work remain visible. Finished tasks now fold their brief and work details into dropdowns you can reopen; results stay visible.
Ask your employees to sort Outlook mail by date, sender or subject, filter it precisely and continue through older messages. They can file mail in batches and change categories, follow-up flags and read status.
Vorenq researches legal company details on the web and fills the supported fields with source links. Review, undo and save the results. Unknown VAT methods and financial year-ends stay unchanged.
The plans are now cut by how much work you hand over: Solo for one use-case, Business for the whole firm, Scale for heavy work on every model. Standard and Pro are now Business and Scale, Solo includes the balanced Sonnet-class models, and every plan includes more monthly credits. Included credits stay usable for 12 months; top-ups don't expire while you're subscribed. Usage now carries a slightly higher margin, so a credit covers about 10% less model usage than before.
Anthropic's newest Opus now runs everywhere Opus 5 used to be the default: hired specialists, the manager, the accounting guardrail, scan reading, brand review and final polish. It costs a fifth less per token than Opus 5. Opus 5 itself is retired: employees who were on it move to Opus 5.5 automatically, at the lower price. Workspaces limited to EU processing run Opus 5.5 through AWS Bedrock's EU profile, where that provider is enabled.
Report from any connector card or its connection setup, even when it is unavailable. Your report includes the connector name and reaches the support queue with a reference. Failed submissions keep your text; returning from a report preserves setup inputs.
Swiss preparation recipes now open directly in the workflow editor for review. The operating plan also lists existing workflows with their latest linked run, completed steps and a history link. Missing evidence stays visible; opening a template does not save or run it.
Set the entity country, ledger currency, reporting standard, accounting basis and timezone. Swiss VAT guidance stays specific to Switzerland; international flows verify local obligations. Trial balances and bank reports use each currency’s ISO minor units, with profile checks across every stage.
Choosing a recurring service opens its existing job or saves a new draft before setup. The job page guides the next step, shows the actual rehearsal status and lets you recover from an interrupted request before trying again.
Accounting stages now require a ledger read after the final bookkeeping write. Before completion, Vorenq checks trial-balance totals and bank-to-books arithmetic using exact cents, rejects empty or unreadable deliverables, and records the files it checked. Actual discrepancies stop the flow; corrected reports can resume without repeating completed stages.
Needs you, In progress and History organize the work hub. The daily overview is now alongside your queue, available integrations come first, one text and voice composer handles requests or captured ideas, and one Save changes button saves company preferences and notifications.
Agents keep recent working details until another summary is needed. Long tasks carry the previous summary forward with new information and reuse available summaries before compressing other details. Your chosen model and reasoning settings stay the same. Exact repeated responses count only once before agents decide whether another summary is needed. Repeated details take less space, leaving more room for new information when a conversation needs summarising.
Enter a title and press Enter to add a task; open notes, date and priority when needed. Select visible tasks to complete them together, set a shared date or clear their dates. Each batch reports confirmed changes and leaves unconfirmed items selected for review.
Your operating plan now shows linked live runs, schedule failures and pending decisions. Open the relevant run, approval or question from the contract review, see the next Swiss local firing time, and pause future cycles when needed. Existing problems take priority over setting up more routines. Custom jobs and responsibilities outside the selected sectors are included, with direct review links and visible evidence coverage.
Find open to-dos by title or notes, then narrow the list to today, overdue, or undated work. The same filters work in card view. Add or edit a deadline with Today, Tomorrow, or In a week.
Agents share a company-knowledge foundation and load matching procedures when needed. Repeated information takes less space, while a complete copy of each reused result stays available. Identical results also reuse an approved summary within the same turn.
See pending decisions, expired permissions and standards due for review. Open each item at its existing controls, start company standards from editable drafts, and download a scoped reference snapshot. Reviewing or downloading never grants permission.
The operating plan and handover now use the same setup checks, including app access and contract requirements. See actual test progress, open the result, and fix a broken routine even when another is running. Completed tests still need review before you rely on them.
The workspace has quieter surfaces, labelled utility links and scrolling tabs that reveal the current page. Long filters use a compact picker, integration search is easier to find, and Settings reveals required details when they need attention. Search results now show previews and let you recover from failed searches.
Choose bexio, CashCtrl, Odoo, QuickBooks, Xero, KLARA, Banana or SmallInvoice. Each flow uses its supported operations, checks the selected ledger and keeps missing posting evidence unresolved. Xero and Banana expose complete read continuations; KLARA adds dated VAT profiles and current payable/receivable positions. Unavailable integrations are labelled before setup.
Set up a daily, monthly or fixed-period bexio accounting workflow with ten stages, evidence checks, recurring handovers and bounded recovery. Review interventions in run history. Bank release, statutory filing and final sign-off remain explicit external steps.
Find standing permissions, shared standards, employee procedures and company context together in the Library. Review what is in force, reaffirm a standard or revoke a permission. A saved procedure does not grant access to your tools.
Add sourced prospects, record replies, meetings and wins, and set the next follow-up directly in a campaign. Work priority brings conversations and overdue follow-ups forward while respecting future dates and opt-outs. The AI sees the same queue. Export the complete pipeline as CSV; reported stages remain separate from verified send receipts.
Choose target cantons and supported outreach languages for a campaign. The saved profile reaches every AI campaign brief and can be reviewed on the campaign card. Language preferences must still come from the prospect’s conversation or evidence; geography does not determine language. The profile guides research and drafts without activating ads or outreach.
Import a reviewed CSV or TSV with source evidence, reported stages and follow-up dates. Preview and choose up to 100 records; existing prospects are skipped and workspace opt-outs remain in force. Individual results distinguish saved, skipped and unconfirmed records. Importing does not send outreach or start a campaign.
Campaigns now offer three starting playbooks and a prospect pipeline with sources, stages and next actions. Configure a draft, run one cycle or launch recurring work. Outreach follows your permissions; campaign suppression is checked before supported Gmail and Outlook sends. Do not contact now applies across all workspace campaigns and survives campaign deletion. Campaigns and opt-outs are included in the workspace export.
Ticket workflows now keep one owner across failures and retries, preserve previous attempt evidence, and constrain writes to the current ticket, branch and stage. Stops and stale approvals cannot silently restart ended work. Uncertain writes require reconciliation before continuing. Pending CI is checked automatically for up to two hours, then a fresh review is required. Retry costs count toward the workflow spending limit.
New GitHub, GitLab and Bitbucket workflows connect ticket intake, a feature branch, a proposed fix, CI checks and a verified merge. Missing evidence stops the next stage. Repository CI runs the project’s tests; deployment and ticket resolution need their own confirmation.
Ask the manager for a starting plan, then prepare a recurring job with an existing employee. Three trades starters cover enquiries to offers, completed work to invoice drafts, and weekly office handover. Review and rehearse before owner activation. Home links to decisions and blocked work; weekly results record owner-reported workload, time and cost separately from estimates.
Connector reads now disclose more pagination and coverage limits, while supported writes check identifiers, receipts or readback before reporting completion. Uncertain responses stay unconfirmed, and planned integrations are labelled unavailable. A queued request is not a completed job.
Connected agents can prepare notebook and pipeline definitions, submit supported Fabric operations and inspect job evidence. Saved content, an accepted job and a completed execution are reported separately. Your Fabric access and workspace permissions still apply.
Account, MFA, invitation, credential and deletion operations recheck current authority and preserve cleanup work across failures. Launch journeys exercise both storage backends in CI. These are internal engineering checks; they are not an independent security certification.
The bexio Mahnlauf now applies the statute for you: which rung each overdue invoice stands on, the day the debtor entered default (Art. 102 OR), the 5 % default interest to date (Art. 104), the limitation horizon (Art. 127/128), when the Betreibungsamt may act at all (Betreibungsferien, Art. 56/63 SchKG) — and, for an exhausted ladder, the Betreibungsbegehren data sheet (Art. 67) with the fee, read from your own records. Prepared, never filed.
Every employee reads three federal sources directly, no account needed: the ESTV exchange rates for VAT (the day's rate for any date), the UID register (is this supplier's VAT number live, is this debtor still a going concern), and the SHAB gazette (bankruptcies, liquidations, register changes). The MWST worksheet converts foreign-currency invoices at the ESTV rate of their date and prints the rate beside the figure.
Vorenq now opens straight into the conversation with your manager; the team sits beside it in configuration. Whatever finished comes back into that same thread — a “since we last spoke” block that you and the manager read identically.
A third door on the brief field: Vorenq reads your brief back to you, asks two to five tappable questions and writes the finished brief from your answers — before a run starts and costs anything. A brief agreed that way is then allowed to work on its own for far longer.
Everything your employees have completed in one place: search it, filter by period, sort by cost, duration, department or person. Every row says whether the result was read back and confirmed afterwards, and what the run cost. Where nothing was measured, it says exactly that — never a zero.
You can grant one employee a standing permission for a kind of task, or clear a single action for everyone. Both are revocable at any time, and some things always ask: money, deletions, signatures, permission changes, and any run that has read outside text.
New logo, new marks, aligned accent colours and a consistent light and dark mode. Panels, cards and the sidebar now sit on frosted glass.
The platform went through a security review. Every finding is fixed or refuted with reasons — from blocking requests into internal networks, through the keys of the file store, to the OAuth sign-in sessions — and each fix is held down by a regression test that fails if the problem returns.
A marketing campaign now has a goal, channels, a spending ceiling and a cadence, and it persists. Every publication is filed against it, including one that only half succeeded. No paid channel goes live without a stated ceiling.
What employees record as findings is read back by their department: open items, topics, how long since anything was written down. Until now only accounting had a view like this. A person can close any item — and undo that.
Word, Excel, PowerPoint, OpenDocument, e-books: sixteen formats are read locally in milliseconds instead of being sent to text recognition. A PDF is classified from its own bytes first, so you only pay for the part that really was scanned.
A photo of an invoice is not a scan of it: Vorenq finds the page in the picture, takes the tilt out, removes shadows and judges the result. When it is unsure, it says so rather than confidently cropping the wrong thing. The original photo is kept, and the finished page is also filed under your receipts.
⌘⇧Space dictates anywhere in the app, not only in chat, and the text is tidied before it lands. Never into a password field. And a transcript is never lost: if it cannot be inserted, you get it to copy.
The new Anthropic flagship is now the heavy tier everywhere Opus 4.8 used to be: hired specialists, the manager, the accounting guardrail, scan reading, brand review and final polish. Same price per token as before, so nothing costs more. Opus 4.8 stays selectable for anyone who wants it.
A ready-made team for the construction back office: Regie reports become invoice drafts, site changes become documented Nachtrag offers before the work is done, plus Akonto billing, checking subcontractor paperwork, defect lists and tender summaries.
A team lead now splits a brief, routes each part to the model ranked highest for it, runs them in parallel, checks every result, and assembles one finished deliverable.
Opt-in for Pro workspaces: wider parallel fan-out, replies triple-checked by majority vote, and a second automatic retry on flagged parts.
A handbook inside the app, one chapter per part of the product. Every screenshot comes from a real workspace, with numbered markers that show exactly where to click. In German, French and English.
Every employee can look up verified Swiss rates and rules (VAT, payroll contributions, safe-harbour interest) and cites the legal basis in its answer. When it's really a question for your fiduciary, it says so.
Scan a bill with the in-app camera, several pages if needed. Your bookkeeper reads it, books it in bexio on your approval, and files the original with the entry.
It sees the whole workspace live, reshapes teams and automations, and tunes every employee toward lower cost, routing only to models rated at least as good on our benchmark.
Hands-free back-and-forth with the manager and every employee. Any written reply can be read aloud.
Email invites, owner and member roles. Every run shows who started it, and chats stay private per person.
Several employees combine on one goal; what one run learns, the next builds on.
Flag an employee and everything it sends waits for your approval, even where a recipient allowlist or a tool's “act freely” setting would let it through; only Monitor mode or an approval shortcut you set yourself lifts it. A pending approval waits for you instead of expiring.
150+ new operations across the catalog: Discord's full bot surface, X read and write, mail that files itself in Outlook and Gmail.
15 trades, fiduciary to sports club, as a ready team with its automations; under “Other”, the AI designs a custom package.
Every task runs on the model our benchmark table ranks highest for that kind of work, always overridable.
Built once by an employee; after that, the numbers refresh without another run.
Proven procedures every employee masters from day one, including on-brand social posts.
The QR section is decoded checksum-verified instead of AI-guessed. Amount and reference are always exact.