Intelligent process automation and agentic automation
Robots for the steps that must be exactly right, AI agents for the steps that need judgement, and people for the decisions that matter, all running as one orchestrated process. Flex Consulting delivered New Zealand's first UiPath Maestro go-live.
What is intelligent process automation?
Intelligent process automation (IPA), now often called agentic automation, combines three things in one process: robotic process automation for the deterministic steps, AI for reading documents and making judgement calls, and orchestration that hands work between robots, agents, systems and people. Where RPA automates a task, IPA automates the whole journey, including the parts that used to stop and wait for a person.
The difference matters. RPA is deterministic: it does a set thing, in a set way, every time, which is exactly what you want when you are paying money or placing a trade. AI agents are probabilistic: they are right most of the time and can handle variety a robot cannot. IPA uses each where it fits, with clear rules for when a person must check or decide.
What IPA changes
Whole processes, not single tasks
A request moves from intake to outcome across systems, robots, agents and people without anyone chasing it.
Judgement without the wait
Agents read emails and documents, classify requests and recommend the next step, so routine decisions no longer queue for a person.
Control where it matters
Steps that must be exact stay with robots and rules. People approve anything outside the limits you set, and every hand-off is logged.
Change without a rebuild
Processes are modelled visually, so a new rule, approver or system is a change to the model, not a new project.
Where IPA fits
Processes with a mix of fixed steps, unstructured inputs and human decisions:
Approvals and requests
Contract, purchasing and HR requests routed through finance, procurement and approvers, with status visible to everyone involved.
Case management
Long-running work such as claims, applications and special consideration, where each case takes a different path.
Customer and staff enquiries
Requests read and classified by an agent, answered or actioned by robots, and escalated to a person with full context when needed.
Document-heavy processes
Invoices, forms and correspondence extracted with intelligent document processing and passed straight into the next step.
Replacing old workflow tools
Unsupported workflow platforms retired and their processes rebuilt as orchestrated, role-based applications.
Exception handling
Robots process the standard cases and agents triage the exceptions, so people only see the work that genuinely needs them.
RPA or agents? Usually both
Use robots when the answer must be right every time: payments, trades, regulatory submissions. Use agents when the input varies and a well-supervised decision that is right most of the time is good enough: reading an email, classifying a document, drafting a reply. Put a person wherever the cost of a wrong answer is high.
Orchestration is what makes the mix safe. UiPath Maestro models the process, runs each step on the right worker, applies business rules, and sends exceptions to a person, with an audit trail of who or what did each step. Read about AI agents
Platforms
UiPath Maestro
Orchestration of robots, UiPath and third-party agents, systems and people in one governed process, modelled in BPMN with DMN business rules and human tasks for exceptions.
UiPath robots and agents
Unattended and attended robots for deterministic steps, agents for judgement, and document processing for unstructured inputs.
Microsoft Power Automate
Cloud flows, desktop flows and AI Builder for organisations working mainly in Microsoft 365.
How we deliver IPA
Map the journey
We map the end-to-end process with the people who run it and mark which steps are fixed, which need judgement and which need a person.
Design the controls
We agree the rules, limits and approvals for each step, so agents act only where it is safe and people decide where it matters.
Build and orchestrate
Robots, agents and human tasks are built and joined up in the orchestration layer, then tested against real cases.
Run and improve
We monitor the process in production, tune the agents and rules, and extend the model as the process changes.
Our work
University of Auckland
Three critical University processes moved off an unsupported workflow platform onto secure, role-based applications orchestrated by UiPath Maestro.
1st UiPath Maestro go-live in New Zealand
Craigs Investment Partners
A closed trading platform with no APIs turned into one guided adviser experience, returning 40,000 hours a year to client-facing work.
40k adviser hours returned a year
Why Flex Consulting
We are a New Zealand automation consultancy with more than 150 projects delivered across over 15 industries, for organisations including Foodstuffs North Island, SkyCity, Craigs Investment Partners, Contact Energy, Chorus, the University of Auckland and Mercury.
Every engagement starts with a free, no-obligation consultation to find your highest-return candidates, followed by a pilot you can measure before you commit to scaling.
Questions we are asked
What is intelligent process automation?
Intelligent process automation (IPA) combines robotic process automation, AI and orchestration so a whole process can run automatically: robots handle fixed steps, AI handles documents and judgement calls, and people handle exceptions and approvals.
What is the difference between IPA and RPA?
RPA uses software robots to repeat fixed, rule-based steps exactly the same way every time. IPA adds AI and orchestration around those robots, so the process can also handle unstructured inputs and decisions, and hand work to people when needed.
Is agentic automation the same as IPA?
Largely, yes. Agentic automation is the newer term for the same idea: AI agents, robots and people working together in one orchestrated process, with the agents handling the steps that need judgement.
Find out which of your processes needs more than a robot
Bring a process that stalls waiting for decisions or documents. We will show you where robots, agents and people should each take over, and what it would save.