AI automation that takes over the busywork.
DVN Solutions brings AI automation to teams across the DACH region. Repetitive work in inbox, sales, marketing, support, ops, engineering, and cyber runs as an approved workflow. You keep control, your team gets the hours back.
Connects with what you already use
Connectors across the stack you already run.
- SlackChannels, alerts, approvals
- HubSpotCRM records + deals
- Google CalendarScheduling + prep
- GitHubIssues, PRs, releases
- CloudflareSecurity + DNS
- StripePayments + invoices
- Gmail
- Outlook 365
- MS Teams
- Telegram
- Pipedrive
- Salesforce
- Close
- bexio
- MS Calendar
- Cal.com
- Linear
- Notion
- 1Password
- Proton
- Airtable
- QuickBooks
- Xero
- Wise
Comms, CRM, calendar, engineering, security + finance · custom integrations scoped per engagement · Microsoft Defender + enterprise SIEM on request
What runs underneath
One private operating layer, not another AI tool.
DVN turns repeatable department work into a managed loop: capture the request, route it to the right Skill Pack, draft the output, ask for approval, and log the receipt. Hermes, OpenClaw, Claude Code, and Codex are the mechanism underneath; the business outcome is approved work your team can trust.
Orchestrator
Hermes
The routing brain
Decomposes each request, routes it to the right skill pack and agent, manages handoffs, and holds the long-term memory of your operations. Decides which agent runs and in what order, and enforces every approval gate before work proceeds.
Cockpit
OpenClaw
Your operator control surface
The single pane of glass where you watch, approve, and audit every agent action before it touches your business. Full evidence trail, branded to your team, a cockpit, not a generic SaaS console. Hermes runs the work; OpenClaw is where the human stays in control.
Reasoning
Claude Code
Deep + long-context agent · Anthropic
The right agent for careful, code-aware, long-context work, multi-step drafts, large-document analysis, cyber summaries. Deliberate over fast. Never acts on its own; every output routes back through approval.
Reasoning
Codex
High-throughput agent · OpenAI
The right agent for fast generation and second-opinion audits, high-volume drafts, code scaffolding, and cross-checks on Claude's work. Speed and breadth. Never acts on its own; every output routes back through approval.
What an agent action looks like
It drafts. You approve. The receipt proves it.
Every action lands as a draft in your inbox. You approve or edit. The receipt (timestamp, target, fingerprint) lands in the audit log.
Three customer emails about delayed shipments. Draft replies referencing tracking data, then book follow-ups for next week.
operator · 14:31
Drafted 3 reply candidates · pulled tracking numbers from Shopify · scheduled 3 follow-ups in Calendar for Tue/Wed. Ready for your approval.
atlas · 14:31
Approve replies 1 + 3. Edit reply 2, I want a softer tone.
operator · 14:32
Sent replies 1 + 3 from sales@example.ch · saved reply 2 as draft for your edit · 3 follow-ups added to Calendar with 'Shipment check-in' titles.
atlas · 14:32
Audit receipt
- session
- session #ac7f...
- action
- send_email · 2 messages · sales@example.ch
- target
- 3 customer threads
- approved by
- operator-12 · 14:32 GMT+1
- fingerprint
- sha256: 9c1b…f04e
Pick a department, ship a workflow
Seven department workflows. One private operating layer.
Mix and match. Volume discount on additional pre-built packs (-10% / -15% / -20%). Custom packs built with you in your environment.
Inbox + Morning Brief
Email/calendar/task triage · drafted replies · daily business brief · follow-up reminders.
Sales Pipeline
Lead capture · CRM cleanup · drafted outreach + follow-ups · meeting-prep one-pager.
Marketing Ops
Content repurposing · campaign calendar · weekly funnel digest · approval-gated campaign changes.
Support Triage
Ticket classification · drafted replies · account-context lookup · weekly voice-of-customer themes.
Ops + Finance
Document/expense classification · reconciliation drafts · ops digest.
Engineering / DevOps
Blocked-PR summaries · sprint digest · incident context packs.
- Premium · Detect + Escalate
Cybersecurity Operations
Defender / Sentinel / Tenable alert triage · auto-classify + dedup · remediation ticket drafts · Incident Response module included · DVN escalation gate for critical findings.
What we lock in by default
Boundaries you can show your board.
Swiss-first data residency
Lead + chat data stored in Switzerland (Zurich, eu-central-2). LLM processing is disclosed separately on /privacy. Subprocessor list disclosed before launch.
Approval-safe by default
No autonomous emails, payments, public posts, CRM writes, or security remediation until you explicitly authorize that mode.
Audit-ready evidence
Every agent action ships with timestamp, target, approval chain, and a cryptographic fingerprint on the receipt.
Questions, answered
What teams ask before they start.
How fast does the first workflow ship?
Typically around 14 days from kickoff, scoped to one workflow, not a big-bang rollout. We never promise a fixed date; we move as fast as your approvals allow.
What if a Skill Pack draft gets something wrong?
Nothing ships without you. Every agent action is a draft until a human approves it, and the audit trail shows exactly what was proposed, by which agent, and who signed off.
Can we leave whenever we want?
We work in 3-month minimum cycles so the build has time to pay off. The setup fee is fixed and non-refundable, it covers the work done upfront. After that you are month-to-month, and you keep your workflows and audit data. The managed runtime stays ours: keep it running on a monthly plan, or take a paid handover to self-host. Commit yearly and you save 30%.
Where is our data stored?
In Zurich, Switzerland, Supabase, eu-central-2. Your lead and workflow data stays on Swiss soil with no transatlantic transit. LLM calls (Anthropic, OpenAI) are a separate, transient processing path, disclosed in full on our privacy page.
Do you connect to the tools we already run?
Around two dozen connectors today across communications, CRM, calendar, engineering, security, and finance, and more on request. Custom connectors are typically built in 2–4 weeks.
What does it cost?
Pricing is modular and scoped to your setup, packs, seats, and runtime. Pricing is on request: start chat and we'll confirm the first workflow and the exact numbers together.
Is incident response really included in the Cybersecurity Pack?
The Cybersecurity Pack is scoped for continuous alert triage when connected to your security tools. It drafts first-response actions and escalates critical findings to DVN Solutions for human-led review. We are not a full MSSP: large, sustained, or complex incidents are scoped and quoted separately, and round-the-clock managed response requires a dedicated Swiss MSSP partner.
How does this scale as our team grows?
The managed runtime scales with you, capacity expands as your workload grows, and pricing tracks usage. No re-platforming, no migration project. And if you ever want full ownership, the stack is yours: self-host on a Swiss VPS with complete handover docs.
What does AI automation actually do for an SMB?
It takes the repetitive work your team does by hand (triaging the inbox, chasing follow-ups, entering data, drafting reports) and runs it as a workflow. The agent drafts, your team approves, the work ships. You get hours back without adding headcount, and nothing goes out without a human sign-off.
How do we start without an IT department?
You don't need one. We scope one workflow, connect the tools you already use, and run it on our managed Swiss runtime, so there are no servers for you to maintain. You bring the process knowledge; we handle the build, the connectors, and the approval layer.