What a fund is made of
We started with a question most emerging managers avoid: what actually constitutes a fund management business? Not inventory, not factories, not working capital. A fund is its people, its investment process, its legal structure, its systems, and the capital entrusted to it. Three of those five can be built from a clean slate, and in 2026 they can be built by a small team using the same technology the fund invests in.
So we built them. Fund administration, performance and fee calculation, tax, accounting, foreign-exchange reconciliation, risk monitoring, investor relationship management, market data, reporting, research and valuation support run on systems we own and control. Work that once needed a team now runs continuously, consistently and auditably.
Where AI is used, and where it is not
Priori runs its back office on agentic AI connected to the firm's internal database. Tax accounting, fund administration, unit pricing and reconciliation run continuously rather than in a monthly batch. The two managers remain responsible for market research and execution, and review the entire operation.
- Manager-led: trade execution, position sizing, portfolio construction, research interpretation, investment decisions, risk appetite, and final approval of all reporting.
- AI-enhanced: research coverage and data gathering, market and position monitoring, routine risk analysis, fund metric tracking, scenario and sensitivity runs, document and filing review.
- AI-led: tax accounting, mark-to-market NAV calculation, unit pricing, trade reconciliation, fee and tax accrual, reporting data and design, and the audit trail.
The technology does not decide what the fund buys or sells. It gives the people who do decide more time to research, to manage risk, and to talk to investors.
Why it matters to an investor
A substantial proportion of repetitive back-office work is automated, which keeps overheads low and lets a small team spend its time on security selection, portfolio construction and risk management. Those savings are passed to investors through the absence of a management fee.
Administration runs against a single internal database, so valuations, unit prices, fee accruals and tax provisions are reproducible from source records rather than reconstructed at period end. Nothing is released without manager review. The model is designed to scale without a corresponding increase in headcount, so the cost of servicing an additional investor or sub-fund is close to unchanged.
Same thesis, both sides
We use frontier models every day, so we have a working understanding of their capabilities, their cost curves, their rate of improvement and their limits. That first-hand experience informs how we weigh the claims of model developers, infrastructure providers and commentators, and it is one reason our conviction in the AI buildout has held through periods when the market's did not. We spend less time operating a fund and more time managing one.