Funding a large redemption

The scenario and numbers in this demonstration project are simulated data.

Industry: Discretionary investment management
What was built: Agents, skills, semantic models and write tools, built on demonstration data
Where the knowledge came from: The redemption instruction and an investment policy excerpt, plus demonstration data from the CRM, portfolio, market data, order and risk systems

Separate desks before the cut-off

An institutional client sends a large redemption instruction for T+2 settlement. The portfolio does not hold enough cash.
The investment desk, the trading desk and risk each used to run their own round of calculations before the redemption cut-off.

The redemption instruction and an investment policy excerpt sit in Drive. The agent reads the instruction and checks the authorised signatory, the amount and the settlement account against the client record in the CRM.
The agent reads the position limits in the investment policy, and checks portfolio positions and each holding's trading volume through semantic models.

The agent uses the liquidity ladder skill to sort holdings by how fast they can be sold: cash, money market funds, government bonds and liquid stocks. The agent works out how much each layer can provide, and confirms that T+2 settlement holds.
Selling too fast moves the market. The agent uses the market impact estimate skill to estimate selling pressure and splits the block order into VWAP batches.
An ex-dividend date, an index change or a major earnings call can make liquidity unusual on the execution day. The agent checks on the web whether any of them falls on that day.
For the limits risk cares about, the agent uses the mandate compliance check skill to verify sector weights, single stock weights, the cash floor and tracking error after the sale.
The decision pack is in Sindri. It holds the liquidity ladder, the impact estimate and batch plan, the post-trade compliance checklist and a draft client confirmation.
The block order and the client confirmation cannot be undone and are regulated. Both need approval from the chief investment officer and the head of risk.
After approval, the agent writes the liquidity ladder to the portfolio system, the batched order to the order system, the client confirmation to the CRM and the internal report to the risk system.
Every amount in the decision pack traces back to which agent read which record in which system.

Get started

AI feasibility review of a live workflow