# Flowdesk Systems — full reference > Flowdesk Systems (Flowdesk Systems Ltd) is an Auckland, New Zealand-based AI automation agency for small and medium businesses. It builds workflow automation, AI infrastructure and custom SaaS software using a Map → Build → Ship process, with fixed NZD quotes. Clients own all code, accounts and credentials — no lock-in. ## Key facts - Website: https://flowdesk.systems/ - Location: Auckland, New Zealand; works with clients across New Zealand (Aotearoa) - Founder: Lachlan Tonner - Email: hello@flowdesk.systems - Phone: +64 20 400 45080 - LinkedIn: https://www.linkedin.com/company/flowdesk-systems - Book a free 30-minute discovery call: https://flowdesk.systems/book/ - Primary stack: n8n (automation), OpenAI and Anthropic Claude APIs (AI), React, Supabase and Node.js (custom software) ## Process: Map → Build → Ship 1. Map — sit with the team, chart every manual step, and score each for automation ROI. 2. Build — build, test and wire the automation into existing tools; no rip-and-replace. 3. Ship — go live with monitoring, a dashboard and a human on call. ## Services (https://flowdesk.systems/services/) ### Workflow automation Replaces repeatable manual work — invoice processing, employee onboarding, daily reporting, follow-ups — with reliable, hands-free systems. Triggers from email, forms, CRM updates and calendar events. Full audit trail and error alerts. Typically saves 8–20 hours per week, per team. ### Custom SaaS builds Internal tools, customer portals and dashboards shaped around how a business works, replacing spreadsheets and manual tracking. Documented, tested, maintainable, and integrated with the existing stack from day one. ### AI integrations and AI infrastructure AI document processing (contracts, invoices, applications), intelligent email routing and response drafting, data extraction from PDFs, images and unstructured text, and custom AI agents for repetitive cognitive tasks — connected by API to the live business stack. ### n8n pipelines Visual workflow automation at scale: CRM sync, lead routing and enrichment, multi-step approvals and notifications, data transformation and reporting. Self-hosted option for full data sovereignty. ## Pricing (https://flowdesk.systems/pricing/) All prices NZD, excluding GST. Final fixed quote confirmed after the free discovery call. - Starter — from $3,500: one workflow automated end-to-end, 1–2 tool integrations, 2–3 week delivery, handover docs, 30-day post-launch support. - Growth — from $6,000: multiple workflows or a full SaaS build, integrations across the stack, 4–8 week delivery, admin dashboard, 60-day support, team training. - Enterprise — custom: phased delivery, dedicated engineer, priority support SLA, monthly reporting, quarterly strategy reviews. - Optional support retainer — from $500/month. No lock-in; clients own everything. ## Example builds (https://flowdesk.systems/case-studies/) Illustrative examples of what Flowdesk builds: AI receptionist (24/7 voice AI that answers calls, books and summarises), document & invoice processing, lead capture & CRM autopilot, client onboarding automation, and a custom ops dashboard. ## Frequently asked questions (https://flowdesk.systems/faq/) ### How long does a typical project take? A simple automation — like a single workflow end-to-end — typically takes 2–3 weeks from kick-off to go-live. A full custom SaaS build is usually 6–10 weeks depending on complexity. We'll give you a realistic timeline during the scoping call, and we stick to it. ### Do we need to replace our existing tools? No. We wire automation around the tools you already use. Changing your stack is expensive and disruptive — our job is to make what you have work better, not replace it. If something genuinely needs replacing, we'll tell you honestly. ### What if the automation breaks? Everything we build includes error handling, alerts and monitoring from day one. If something fails, your team gets notified immediately so nothing slips through undetected. If you're on a support retainer, we're on it within hours. ### Do you work with non-technical teams? Yes — most of our clients are non-technical. We handle all the technical implementation. You just need to know your business and your processes well. We translate the rest. ### What tools do you build with? Our primary automation platform is n8n. For AI integrations, we use the OpenAI and Anthropic Claude APIs. For custom software, we typically use React, Supabase and Node.js. We always use whatever fits best for the specific problem, not just what we're comfortable with. ### How much does it cost? Projects start from $3,500 NZD for a simple automation. Larger builds — multi-system integrations or custom SaaS — are typically $8,000–$25,000+. Discovery is always free. See the pricing page for full details, or just book a call and we'll give you a ballpark. ### Do you offer ongoing support after launch? Yes — we offer a monthly support retainer from $500/month NZD. This covers monitoring, minor updates and priority access if something needs fixing. It's completely optional. Once we hand over, you own everything and can run it yourself. ### Can you integrate with [our specific software]? Almost certainly. If it has an API — and almost everything does — we can connect to it. We've integrated with Xero, HubSpot, Salesforce, Shopify, Google Workspace, Microsoft 365, Trade Me and dozens of industry-specific platforms. Just ask. ### How do we get started? Book a free 30-minute call. We'll ask about your biggest time drain, walk through your current workflow and come back with an honest assessment of what's worth automating — and what isn't. No sales pitch. ### Where are you based? We're based in Auckland, New Zealand. We work remotely with clients across Aotearoa — and occasionally further afield for the right project. ## Blog posts (https://flowdesk.systems/blog/) ### Why most NZ businesses are overpaying for manual work > Most business owners know their team spends too much time on repetitive tasks. What they don't know is exactly how much it costs — and how cheaply it can be fixed. - URL: https://flowdesk.systems/blog/why-most-nz-businesses-are-overpaying-for-manual-work/ - Published: 2026-03-12 - Author: Lachlan Tonner, Flowdesk Systems - Category: Automation Ask any business owner if their team wastes time on repetitive admin and they'll say yes without hesitating. Ask them how much it costs and you'll usually get a shrug. That gap — between knowing the problem exists and knowing its price — is where most of the waste hides. #### The hidden cost of "that's just how we do it" Manual work rarely shows up as a line item. It's spread across a dozen small habits: re-typing data from one system into another, chasing invoices by hand, building the same report every Monday morning. None of it looks expensive in isolation. Multiplied by 52 weeks and however many staff touch it, it adds up to real money — usually more than the cost of fixing it. We see this pattern constantly: a task that takes 20 minutes, done by someone earning $35 an hour, five times a week. That's $3,000 a year for one task, for one person. Most businesses have five to ten of these running in parallel without anyone adding them up. #### Doing the maths The exercise is simple. List the repetitive tasks your team does. Estimate the time each one takes and how often it happens. Multiply by an hourly rate. You don't need to be precise — even a rough number is usually enough to make the case for fixing it. What surprises people isn't that the number is big. It's that the fix is small. Most of these tasks are pattern-matching: if this happens, do that. That's exactly what automation is good at, and it's usually cheaper to build than a single quarter of the manual cost. #### Where to start Don't start with the flashiest idea — start with the task everyone complains about. It's usually the one that's been "temporary" for two years, the one new hires get told "yeah, we know, we've been meaning to fix that" about. That's the highest-ROI place to begin, because the annoyance is already quantified in the way people talk about it. Fixing one workflow properly tends to be more valuable than automating five badly. Start narrow, prove the maths on a real task, then expand from there. --- ### Why we use n8n for production automation > We've built production workflows on just about every platform out there. Here's why n8n is the one we keep coming back to. - URL: https://flowdesk.systems/blog/why-we-use-n8n-for-production-automation/ - Published: 2026-02-28 - Author: Lachlan Tonner, Flowdesk Systems - Category: Tools We've built automation pipelines on a lot of platforms over the years. Every time we evaluate the landscape, we end up back at n8n. Here's why it's our default — and what it does better than anything else in the space. #### Built for real business logic Most automation tools are designed for simple trigger-action flows: new form submission → send email, new row → post to Slack. That's fine for basic use cases, but real business processes are rarely a straight line. They're full of conditional branches, error handling, retries, and custom data transformations. n8n handles all of that natively — you can drop into JavaScript or Python whenever the visual nodes aren't enough, and the workflow engine supports complex branching, merging, and error-recovery strategies out of the box. #### Open source, self-hostable, no per-execution billing n8n is open source and self-hostable, which matters for two reasons. First, there are no per-execution billing surprises when a workflow scales from 100 runs a month to 10,000. Second, client data stays inside infrastructure we control — no third-party cloud processing sensitive business information. For NZ businesses with data sovereignty requirements, this is often a deal-maker. #### The node ecosystem is genuinely deep n8n ships with 400+ native integrations covering CRMs, databases, email platforms, cloud services, and APIs. But the real power is that when there isn't a native node for something — and there often isn't for bespoke internal tools or legacy systems — you're not stuck. HTTP Request nodes, webhooks, and custom code nodes mean you can connect to literally anything with an API. No fighting the tool, no awkward workarounds. #### Production-grade from day one n8n has built-in error handling, retry logic, execution history, and alerting. You can set up Slack or email notifications for failed workflows, inspect every execution with full payload visibility, and debug issues without digging through logs on a server. For client-facing automations where reliability is non-negotiable, this matters a lot. #### The verdict Almost everything we build ends up needing at least one of: conditional branching, a webhook from a system with no native integration, custom data transformation, or a retry-on-failure strategy for something that occasionally times out. n8n handles all of that without fighting the tool. It's the platform we trust for production workloads — and the one we recommend to every client who asks. --- ### How to audit your workflows in an afternoon > A simple framework for finding your highest-ROI automation opportunities — no consultant required. Walk away with a prioritised list of what to fix first. - URL: https://flowdesk.systems/blog/how-to-audit-your-workflows-in-an-afternoon/ - Published: 2026-02-14 - Author: Lachlan Tonner, Flowdesk Systems - Category: Guides You don't need a consultant or a six-week discovery process to find your best automation opportunities. Most businesses can do a useful audit in an afternoon with a spreadsheet and some honesty about how the week actually goes. #### Step 1: List every repeating task Get your team (or just yourself, if you're small) to write down every task that happens more than once a week and follows roughly the same steps each time. Don't filter yet — invoicing, data entry, report generation, follow-up emails, scheduling, all of it. Aim for at least 15-20 items; most teams are surprised how long the list gets once they actually sit down and write it out. #### Step 2: Time it, multiply it For each task, estimate two numbers: minutes per occurrence, and occurrences per month. Multiply them together, then multiply by an hourly rate for whoever does it. This gives you a rough monthly cost per task. You're not aiming for accounting-grade precision — a five-minute estimate per line is enough to separate the big items from the noise. #### Step 3: Rank by pain, not size Sort the list by cost, but don't treat it as gospel. A cheap task that everyone hates doing is often a better first fix than an expensive one nobody minds, because morale and error rates matter too. Cross-reference the dollar ranking against a quick gut-check: which of these, if it broke tomorrow, would actually cause a problem? Those are your real priorities. You'll usually end up with three to five tasks that are both expensive and painful. That's your automation roadmap — not a hundred nice-to-haves, a short list you can actually act on. One afternoon of honest accounting beats months of vague intent to "get around to automating stuff eventually". Do the exercise, pick the top item, and go fix it. --- ### The 5 automations every service business should have > Invoice follow-up, onboarding, reporting, booking and CRM sync. If you're doing any of these manually, you're leaving money on the table. - URL: https://flowdesk.systems/blog/five-automations-every-service-business-should-have/ - Published: 2026-01-30 - Author: Lachlan Tonner, Flowdesk Systems - Category: Automation Not every business needs the same automations, but service businesses tend to share the same five bottlenecks. If you're doing any of these by hand, here's what to fix first — roughly in order of impact. #### 1. Invoice follow-up Chasing unpaid invoices is one of the most avoided tasks in any business, which is exactly why it gets neglected and cash flow suffers. A simple automation that checks invoice status daily and sends a polite, escalating reminder sequence recovers money faster than any human ever gets around to doing it, without anyone having to feel awkward about it. #### 2. Client onboarding New client onboarding is usually a checklist scattered across email, a spreadsheet and someone's memory. Automating the intake form, document collection, and account setup turns a multi-day process into something that runs itself in the background — and it means nothing gets missed because someone was on leave that week. #### 3. Reporting If someone spends part of every Monday manually pulling numbers into a report, that's a standing automation opportunity. Pulling data automatically from the tools you already use and assembling it into a dashboard or scheduled summary removes a recurring task and gives you fresher numbers, since it can run daily instead of weekly. #### 4. Booking Manual scheduling — back-and-forth emails, double bookings, missed confirmations — costs more time than it looks like. A booking flow that checks real availability, confirms automatically, and sends reminders removes an entire category of admin and reduces no-shows as a side effect. #### 5. CRM sync Leads and customer data scattered across email, spreadsheets, and whatever tool sent the enquiry means nobody has a single source of truth. Syncing every channel into one CRM automatically — with proper tagging and routing — means nothing falls through the cracks between systems. None of these five are exotic. They're the boring, recurring tasks every service business already has, and they're exactly the kind of thing that pays for itself within a few months of being automated properly. --- ### What custom SaaS actually costs (and what it saves) > The price of a custom internal tool sounds scary until you compare it to the cost of your team living in spreadsheets for another 12 months. - URL: https://flowdesk.systems/blog/what-custom-saas-actually-costs/ - Published: 2026-01-15 - Author: Lachlan Tonner, Flowdesk Systems - Category: SaaS "Custom software" sounds expensive the moment you say it out loud. It conjures images of six-figure enterprise projects with a project manager, a Gantt chart, and a launch date that slips twice. That's not what most businesses actually need, and it's not what it actually costs. #### The sticker price A focused internal tool — a dashboard that replaces five spreadsheets, a booking system tailored to how your business actually operates, a portal that gives clients visibility without a phone call — is a fundamentally smaller build than a public SaaS product. It doesn't need to handle millions of users, multi-tenant billing, or every edge case a general-audience product has to plan for. It needs to solve your specific problem well, which is a much narrower and cheaper target. #### What it replaces The real comparison isn't "custom tool vs nothing" — it's "custom tool vs the current cost of not having it". That cost is usually invisible because it's distributed: an extra hour a day reconciling spreadsheets, a report that's always slightly out of date, a client who calls because they can't check something themselves. Add up twelve months of that and it's frequently more than the build cost of the tool that would have removed it. #### When it's worth it (and when it isn't) Custom software makes sense when a process is core to how you operate and off-the-shelf tools force you to bend your business around their assumptions instead of the other way round. It doesn't make sense for a problem three off-the-shelf tools already solve well — in that case, wire those tools together instead of rebuilding them. The honest test: if you're maintaining a workaround, a spreadsheet, or a manual step specifically because no existing tool fits how you work, that's usually the signal that a small custom build will pay for itself faster than people expect. --- ### AI agents in production: lessons from real NZ deployments > What actually goes wrong when you put AI agents in production — and how to build them so they fail gracefully rather than catastrophically. - URL: https://flowdesk.systems/blog/ai-agents-in-production-lessons-from-real-nz-deployments/ - Published: 2025-12-10 - Author: Lachlan Tonner, Flowdesk Systems - Category: AI AI agents look effortless in a demo. Production is a different environment entirely — real customers, real edge cases, and real consequences when something goes wrong. Here's what we've actually learned putting agents into live use for NZ businesses. #### What goes wrong The failures are rarely dramatic. An agent doesn't usually go rogue — it quietly misreads an ambiguous request, gives a confident answer to a question it should have escalated, or gets stuck in a loop when a tool call returns something unexpected. The risk isn't Hollywood AI going haywire, it's a small, plausible-sounding mistake that nobody notices until a customer flags it. The second common failure is scope creep: an agent built to answer FAQ-style questions gets asked something it was never designed for, and instead of saying "I don't know", it improvises. Improvising is exactly the wrong behaviour for anything touching bookings, money, or commitments made on the business's behalf. #### Designing for graceful failure The fix isn't a smarter model — it's tighter boundaries. Every agent we deploy has an explicit list of what it's allowed to do and a hard rule to escalate to a human the moment it's outside that list, rather than guessing. Every action that changes real-world state — sending an email, confirming a booking, updating a record — goes through a defined tool call with validation, not free-form text generation. That turns "the AI might say something wrong" into "the AI can only do things we've explicitly allowed it to do". Logging matters just as much as the guardrails. Every conversation, every tool call, and every escalation gets recorded, so when something does go sideways, it's a five-minute fix instead of a mystery. #### What we've learned deploying in NZ Local context matters more than people expect. An AI receptionist needs to understand Kiwi phrasing, local business hours conventions, and the fact that callers will happily talk over a scripted greeting. Generic, overseas-trained assumptions about how a conversation should flow don't always hold here, and the agents that perform best are the ones tuned against real local call transcripts, not a generic template. The businesses getting the most value aren't the ones chasing the most ambitious agent — they're the ones that scoped a narrow, well-defined job, built proper guardrails around it, and let it run reliably before expanding what it's trusted to do.