AI services and platform
AI consulting, agents and automation—built around your business.
We help New Zealand businesses choose the right opportunity, build the working solution, connect it to existing systems and keep it useful after launch. Start with one expensive or repetitive workflow. Prove the value. Expand only when it makes sense.
AI opportunity and workflow consulting
You do not need to arrive with a technical specification. Bring us the process that is slow, repetitive, inconsistent or too dependent on one person.
We examine:
- what starts the process and what a completed result looks like;
- which people, systems, documents and decisions are involved;
- where time, rework, delay and errors accumulate;
- which steps need AI, which need standard automation and which should remain human;
- what information the solution can access and what it must never access;
- how value and quality will be measured before and after launch.
- What you receive
- A current-state workflow, a proposed workflow, opportunity and risk notes, a recommended first implementation, success measures, scope and price.
- Best fit when
- You know work is being wasted but are not yet sure which AI use case deserves investment.
Custom AI agents
An AI agent is an AI worker with a defined job, approved information, tools it is allowed to use and boundaries around the actions it can take.
A useful agent might review an inbox, classify a request, find the relevant customer and job records, draft the next action, update a system and ask a person to approve the result. It does not simply answer a question and disappear.
Every agent design defines:
- its job and success criteria;
- the sources it may trust;
- the systems and actions it may use;
- validation and exception rules;
- human approval points;
- logs, alerts and escalation;
- ongoing review and improvement.
- Best fit when
- The work requires reading or judgement as well as a repeatable sequence of actions.
AI workflow automation
Workflow automation moves work between people and systems without relying on copying, pasting or memory. AI is added where a process needs to interpret an email, document, image or conversation.
Example workflow:
- A quote request arrives by email.
- The AI extracts the products, quantities, location and deadline.
- The workflow checks customer and price information in the relevant systems.
- A draft quote is created and added to the CRM.
- A person reviews and approves it.
- The customer receives it, and a follow-up is scheduled if there is no reply.
- Best fit when
- People can describe the job as “first this, then this, then this”, but the inputs are too messy for a simple rule-based integration on its own.
AI knowledge assistants
Turn policies, manuals, product information, project records and other approved documents into a knowledge system your team can ask in plain English.
Answers are grounded in your sources and can link back to the relevant document or record. Access can follow existing roles, so people see only the information they are permitted to use.
Common uses include staff onboarding, internal support, product and warranty questions, technical guidance, tender preparation and finding information across large document sets.
- Best fit when
- The answer exists, but people lose time looking for it or depend on the same experienced colleague to find it.
Document and data automation
AI can read invoices, forms, contracts, reports, scanned records and email attachments, then turn unstructured information into useful data and actions.
The workflow can extract fields, classify the document, check information against a purchase order or business rule, route exceptions to a person and write approved information into the right system.
- Best fit when
- Documents arrive in several formats and a person currently reads and re-enters the same kinds of information at volume.
AI chatbots and voice workflows
Customer-facing AI should have a clear job and a reliable handover—not pretend to know everything.
We build chat and voice workflows that can answer approved questions, collect structured details, check availability, book an appointment, create a request and pass complex or sensitive conversations to a person with context attached.
- Best fit when
- Slow response times lose opportunities, or your team repeatedly handles predictable enquiries that can be safely bounded.
Integration, implementation and ongoing support
The value usually comes from making existing systems work together. We connect the agent or workflow to the tools involved, test it with real examples, train the people using it and support it after launch.
Potential connections include MYOB, Microsoft 365, Outlook, Gmail, Google Drive, SharePoint, HubSpot, Shopify, Slack, WhatsApp, CRMs, job-management software, databases and spreadsheets.
We continue to monitor errors, exceptions, model performance, cost and user feedback. A workflow is operational software, not a one-off demonstration.
The platform
WorkerAI platform
Our platform is the operating layer for your AI workers. It brings models, business knowledge, integrations, rules, approvals and monitoring together so the solution can perform a real workflow rather than remain an isolated chat window.
No build cost on standard setups
If the requirement fits a standard agent, knowledge assistant, chatbot or supported integration pattern, there is no build fee.
Fixed-price custom workflow design
Custom workflows are scoped and quoted before work starts. You know the deliverable and price without an open-ended hourly build.
No lock-in contract
Service is month to month. We need to keep earning the next month by keeping the solution useful.
Usage-based monthly pricing
The monthly AI component adjusts to actual use. A quiet month should not be priced like a busy one.
Model choice
Different work deserves different models. Use a high-capability model for complex reasoning and a faster, lower-cost model for predictable high-volume tasks. Model choices can change as performance, price and requirements change.
Hundreds of integrations
Agents can work across the accounting, CRM, email, file, job, customer-service and reporting systems your business already uses.
On-premises option
For organisations with strict control, sovereignty or infrastructure requirements, the platform can be deployed in your environment. We agree the model, storage and integration architecture against the required data boundary rather than treating “on-premises” as a vague security label.
Useful autonomy, with people in control.
Example workflow: overdue account follow-up
- Trigger
- The agent reviews aged receivables each weekday morning.
- Information
- It reads the account balance, invoice history, CRM notes and approved communication rules.
- Action
- It drafts a message appropriate to the customer and stage of the account.
- Control
- A person approves messages above an agreed value or risk threshold.
- System update
- The approved communication is sent and the CRM record is updated.
- Evidence
- The activity is logged, including the records used and the action taken.
- Escalation
- Disputed invoices, vulnerable customers and unusual cases go to a person.
This is the difference between “adding AI” and designing a reliable business workflow.
How an engagement works
-
Free consultation
Identify the business problem and decide whether it is worth investigating.
-
Scope and quote
Map the workflow, controls, systems, deliverable and expected value. Provide an upfront price.
-
Configure and connect
Build the agent or automation using the platform and required integrations.
-
Test against real work
Compare outputs with the current process, tune rules and confirm exceptions.
-
Launch with training
Put the workflow into controlled use and show the team how to work with it.
-
Monitor and improve
Review performance, cost, exceptions and changing requirements.
When AI is—and is not—a good fit
AI is worth investigating when a workflow happens often, consumes meaningful time, uses digital inputs, follows recognisable rules and has an output that can be checked.
AI may not be the right first answer when the process rarely occurs, the process itself is not understood, source information is unreliable, mistakes cannot be safely detected, or a standard feature in existing software already solves the problem.
The free consultation is designed to make that distinction before money is spent.
Services and platform questions
Can we start with one workflow?
Yes. One well-chosen process is usually the best place to begin. It gives the team a clear result to assess and creates the evidence needed before expanding.
Can the platform run on our infrastructure?
Yes. An on-premises deployment option is available. The exact design depends on your infrastructure, model choice, integrations and data-boundary requirements.
Can we choose which AI model is used?
Yes. You can use different leading models for different tasks and change them as performance, cost or requirements change.
How long does implementation take?
It depends on the workflow, systems, data and approvals involved. You receive a scoped set of milestones before work starts rather than an undefined implementation window.
How do you measure whether it is working?
We agree the measure before building. Depending on the workflow, that may include time per case, turnaround time, exception rate, rework, accuracy, cost per completed task, missed follow-ups or user adoption.
What happens if our needs change?
Workflows, rules, integrations and models can be adjusted. Ongoing monitoring and improvement are part of the service.
Start with the job, not the technology.
Bring one process that takes too long, creates avoidable errors or keeps getting pushed to the bottom of the list. We will tell you what can be improved, whether AI belongs in the answer and what it would cost.