FAQ

Not in the traditional sense — but the platform is the system of record for the client relationship, and that matters more. You log calls, meetings, documents, and notes against a client, hold both system and custom fields on the record, and everything a flow produces lands there too. If you already run a CRM, integrate it: render client information from the data you hold, so what the platform produces reflects your records.

No. Products can be used without a timeseries — you define the relevant values manually instead. Mapping a product to a dataset is what unlocks calculated risk, return, and correlations, sustainability datapoints such as SFDR fields and EET data, and access to product documents.

You create and manage your own custom timeseries. A product mapped to a custom timeseries is handled the same way as one mapped to a Morningstar dataset.

Using the Morningstar Funds, ETFs, Index, and Categories datasets requires a Morningstar licence, held with Morningstar directly. Custom timeseries and manually defined products do not.

The platform generates the orders and keeps their status; execution happens where it happens today. Orders can be read over the API by whatever executes them — an order management system, a custodian, or your core banking platform — and the status on each order follows it, so the outcome of the advice stays visible on the client.

Yes. An account can be created by a flow in the platform, or mirrored from your portfolio management system with its positions and cash. Each account carries the vendor it came from and an external ID, so the two stay matched, and the client record shows when accounts were last synced.

A wrapper type says what an account is — a pension account, an ordinary investment account, a tax-favoured wrapper — and which products may be held in it. When a flow produces an investment, the platform uses the wrapper type to place each product in the right account automatically, so allocation follows your rules rather than an advisor's memory.

Because instructions are guidance the model applies when it decides to, and published benchmarks measure the gap. In SysBench (2024, arXiv:2408.10943), the strongest model tested followed about 87% of the individual rules in its instructions, but kept every rule for a whole conversation only 54% of the time. In ManyIFEval (EMNLP 2025, arXiv:2509.21051), with ten simple rules to hold at once, the best model managed 78% and several widely used ones fell below 50%. Newer models do better, but any rate below 100% is the point. In the platform, suitability rules, product limits and approvals are checks in code, applied to every flow whoever runs it — advisor or agent — and every action is logged. That is the difference between a rule that is probably followed and one you can evidence to a regulator.