Custom Agents
Your workflows, running themselves.
The repetitive work in your fund, handled end to end. An agent watches for its trigger, runs the steps the way your team runs them, and hands you the finished work to approve.
IT approvedData secureHuman in the loop
How We Build It
How your agent gets built.
Scoped before we price, built on your real systems, maintained after launch.
Step 1 · Scope
Understand the workflow end to end, with the person who runs it.
Step 2 · Build
Built on your systems, tested on your real work.
Step 3 · Deploy
Live in your tools, behind your sign-off.
Step 4 · Maintain
Watched, corrected and improved as things change.
The Benefits
Why funds are moving repetitive
work to agents.
Faster, safer, always on, improving with every run. Your team’s time goes to the work that needs them.
Days of work,
Your team freed
for expert work
Nothing missed,
start to finish
Always on,
nights and weekends
One standard,
every run
More business, with
the same team
Every run logged
and auditable
Better with
every run
The Ideas
What could your agent be?
Every box is a repetitive workflow from credit and equities funds that an agent can run end to end. Click one to see what it does.
These are just examples of what we can build, whatever the complexity.
Pricing & Inclusions
Scoped to your workflow.
We map the workflow before we price it, and agree a fixed number before we build.
Every build
Custom pricing
The specific price depends on the complexity of the workflow. Agents range anywhere from $5,000 to $50,000, and complex builds may include a scoping fee.
Not just the agent
- End-to-end build and rollout
- Training for the team that runs it
- Usage dashboard
- Full audit trail on every run
- Ongoing maintenance and improvement
Case Study
Loan monitoring, run end to end by an agent.
A real build for an Australian private credit fund that now runs its loan monitoring completely through agents.
*The information in this demo is for demonstration purposes only and does not reflect the actual fund, which stays anonymous.
The goal
Monitor more loans, more efficiently and with better analysis.
The process
- Receive the company’s financials
- Transpose them into the model
- Analyse the month’s numbers
- Draft and approve the report
- Complete the cycle, with a full audit trail
“The month used to disappear into monitoring. Now each loan is an hour of review, and the analysis is sharper than what we produced by hand.”
Associate Credit Analyst
7 hrs
Saved per loan, each monitoring cycle
100%
Of covenants and trends analysed, every cycle
3x
More loans monitored, same team size
Get Started
Book a strategy call.
See where AI fits in your fund.
We'll show you where AI lifts efficiency and performance across your organisation.
Common questions.
Answers to some frequently asked questions.
Our workflow is messy and half of it lives in people's heads. Can you still automate it?
Yes, and that is normal. Scoping starts by sitting with the person who actually does the job and mapping the workflow as it really runs, not as the manual says it runs. The messy parts usually become the agent's rules. If a step genuinely needs human judgement, it stays with a human and the agent works around it.
What kinds of workflows make good agents?
Anything that repeats. If the steps are roughly the same every time and the inputs live in your systems, it is a candidate: logging enquiries, screening deals, monitoring covenants, reconciling the book, drafting the quarterly report. The scoping call usually finds more than teams expect.
Does a human stay in the loop?
Yes. Sign-off is built into the workflow, and you set the level. Some teams want approval on every output; others only want to see the exceptions. Anything outward-facing, like an email leaving the building, only moves once a person has approved it.
What happens when an agent gets something wrong?
Every run is logged, so a miss is caught rather than buried. It is captured as feedback, grouped into themes and used to improve the agent, so the same mistake does not come back. That improvement loop is part of every build, not an optional extra.
Our tools do not have off-the-shelf integrations. Does that rule us out?
No. When a tool has no ready-made connector, we build a custom connection so the agent can still read from it and write back to it. If your team relies on it, we find a way to wire it in.
How is this different from Fund Brain?
Fund Brain answers questions; agents do work. The brain is the single place to ask about anything your fund knows. An agent takes one specific workflow, like logging enquiries or reconciling the book, and runs it end to end. They are built to work together: agents often draw on the brain, and each is scoped on its own.