Role Design
Each agent is scoped like a hire: a role, inputs, outputs, tools it may use and decisions it may make. You approve the job description before anything is built.
AI agents for business: software that does the job.
Overview
A chatbot answers questions. An agent does work. Custom AI agents read your inbox, research your prospects, draft your documents, match your payments and file your reports, autonomously, inside the tools you already run, with human checkpoints where judgement matters.
Tensor Partners designs, builds and operates agent systems for businesses in the US, Australia and New Zealand. Our own operation runs on them daily, from publishing to finance, which is why we scope them by the work they replace, not the technology they contain.
Capabilities
What we deliver
Tool + System Access
Agents are wired into your stack, CRM, inbox, accounts, calendars, ad platforms, with scoped permissions. They act through the same systems your team uses, auditable at every step.
Multi-Agent Teams
Bigger workflows get a roster: specialist agents for research, drafting, checking and reporting, orchestrated so each does one job well. The same model that runs our own agency.
Human Checkpoints
Anything customer-facing, financial or irreversible routes through an approval queue. The agent drafts and prepares; a human clicks send until trust is earned.
Memory + Context
Agents keep durable knowledge of your business: clients, preferences, history, house style. Output improves with every cycle instead of starting cold.
Operations + Tuning
Deployed agents are monitored, measured and tuned. As models improve, your agents improve, without a rebuild.
Go deeper
Specialist services
Each one can run on its own or as part of a wider engagement.
01Custom AI AgentsBespoke, single-purpose agents built for one business's exact workflow, not an off-the-shelf tool bent to fit.Learn more02Workflow AutomationThe repetitive processes your team runs the same way every week, handed to software that does not get bored, sick or behind.Learn more03Document ProcessingAgents that read invoices, statements, forms and contracts, extract the data and act on it, without manual re-keying.Learn more04Internal Knowledge AssistantsAI agents that answer your team's questions from your own documents, policies and history, not the open internet.Learn more
Use cases
Where it pays for itself
- 01
Lead handling
An agent that researches every new enquiry, drafts the first response and preps the call notes before your morning coffee.
- 02
Finance ops
Transactions matched to invoices, payments recorded, exceptions queued for review. Continuous bookkeeping instead of month-end panic.
- 03
Content + publishing
Research, writing, imagery and publishing run as an agent pipeline to a fixed calendar, at editorial quality.
- 04
Reporting + monitoring
Agents that watch your numbers, assemble the weekly report and flag anomalies with the context already gathered.
- 05
Internal assistants
A knowledge agent that answers your team's questions from your own documents, policies and history, not from the open internet.
Case studies
Live systems, real-world application
Autonomous publishing system
Content Pipeline
A scheduled agent pipeline that researches live industry trends, writes and edits an article, generates its imagery and publishes to two production websites three mornings a week. Runs unattended, to a fixed schedule, at editorial quality.
Finance operations agent
Reconciliation Agent
An agent that reads live bank transactions, matches every credit against outstanding invoices, records the payments and flags exceptions for human review. Real money and real ledgers, reconciled automatically.
FAQ
Frequently asked
01What is the difference between an AI agent and a chatbot?
A chatbot waits for a question and answers it. An agent is given a job and completes it: it reads, researches, uses tools, takes steps in sequence and delivers an outcome, escalating to a human when it hits something outside its brief.
02Can agents work with our existing systems?
Yes, that is the point. Agents act through your existing CRM, inbox, accounting platform and documents with scoped permissions. If a system has an API or an interface, an agent can usually work with it.
03How do we stop an agent doing something it should not?
Permissions and checkpoints. Each agent has an explicit list of tools it may use and actions it may take, and anything financial, customer-facing or irreversible routes through human approval. Every action is logged.
04How long does it take to deploy a custom AI agent?
A single well-scoped agent typically goes from mapping to production in weeks. Multi-agent systems are sequenced so the first agent is earning its keep while the next is being built.
Start here
- What you get
- A straight answer on whether AI pays off in your business, and what it would take.
- Office
- 2 Wakefield Street, Auckland CBD, Auckland 1010, New Zealand
- Response time
- Within one working day
Start here
The role is defined. Deploy the agent.
30 minutes, no pitch deck, no obligation. Tell us what's eating your team's time and we'll tell you straight whether AI is the answer, and roughly what it would take.