The number that leaves your firm should be able to explain itself.
Monitoring, research, reconciliation, reporting and workflow signals for PMS and AIF managers, family offices and wealth managers — built in your environment, on your data, and handed over.
Three systems, three answers, one deadline.
The custodian file, the administrator's book and the internal sheet disagree. Somebody finds the break by reading rows on a Saturday. The number goes out on Monday and nobody can reconstruct how it was arrived at.
- The month-end runs on a personNot a process — a person, and their spreadsheet, and their memory of last quarter's exception.
- Breaches are found lateMandate drift, concentration and limit breaches surface at review, not at the moment they happen.
- Provenance is missing“Where did this figure come from?” has no answer inside the file that contains it.
Monitoring is where we start. Research is what makes it worth keeping.
Monitoring only works on a reconciled book, which is why the two are sold together. Research over your own documents is the one that changes how the desk works every day.
Five kinds of system, not a service list. Each one is a thing you are handed, and each one is already running somewhere.
- Steady stateCustodian, broker and administrator files are reconciled into a book of positions, then checked continuously against the mandate and its limits.
- Most daysNothing to report. That is the correct output, and it is what a monitoring system looks like when it is working.
- When something deviatesAn exception is raised carrying the record that produced it — the position, the limit, the file and the timestamp — and it goes to a named owner, not to a dashboard.
Continuous checks on your own book, mandates and limits. Three things you receive, each one a system rather than a feature:
- Held-name monitoringExposure and concentration against the mandate, limit and covenant breaches, and data-quality anomalies on the positions you actually hold.
- Corporate actions and filing changesCorporate actions on held names, and filing and disclosure changes, resolved to the instruments and entities they touch.
- Mandate-derived exception signalsSurveillance and compliance alerting derived from your own mandate, not from a market view.
Private, client-hosted research over the documents and instruments you actually hold. Extraction, search and monitoring across your IC memos, deal files, custodian packs, filings and disclosures for held names — every figure carrying the page it came from. Ask what the last four IC memos said about a holding and get the answer with its paragraph. Watch a filing source and get told when something changes on a name you own. For example — a research agent that answers a question about your own documents and cites the paragraph it came from.Buyer: head of research · CIO
Custodian, broker and administrator data reconciled to the books; exception queues with source lines; securities master and corporate-action correctness; the pipelines everything else runs on. For example — the custodian file against the administrator's book each morning, breaks queued with the source line that caused each one.Buyer: COO · head of operations
Client statements, investor and LP reporting, fact sheets, the SEBI cadence, NAV and fee computation checked against the mandate before anything is issued. For example — the PMS monthly offsite submission assembled from the reconciled book, with every figure traceable to its source file before it is signed.Buyer: COO · compliance officer
Signals derived from your own book, your own mandate and your own rules — not from the market. A concentration threshold crossed. A covenant approaching a limit. A corporate action on a held name that changes an assumption in a memo. A data-quality break that would make tomorrow's number wrong. Each one arrives with the record that produced it and a named owner, never as a view on a security. Buyer: CIO · COO · principal officer
Where a system produces signals, they are signals about your own book, your own mandate and your own documents — never a view on a security for anyone else.
Underneath all five: data pipelines, point-in-time correctness, audit logs, replay, an exception queue and a kill switch. That is the list a firm has to demonstrate about any system it relies on, so we build it first.
Most days the output is nothing.
That is the correct output, and it is drawn at the same weight as the other one. The value is in the day a rule is crossed, and in whether the signal arrives with the record that produced it.
An alert you have to investigate from scratch is a second job, not a system.
The answer carries its paragraph, or it is not an answer.
A question about your own holdings, answered from your own files. What the system cannot find comes back as not in your documents, drawn at the same weight as the success path — the one that decides whether a research system is usable.
Nothing here reads the market. It reads what you already own.
Redact the decision, show the machine.
Order state, orchestration and risk are shown as built. Signal logic and exit rules are not — those are strategy, and the strategy is not the product.
Every system below runs on our own account. None of it is client work and none of it is offered as evidence of client work.
- Market event intelligence Own accountAn event graph linked to instruments, sectors and entities, so a filing or a corporate action resolves to everything it touches rather than to a headline.
- Risk & position monitoring Own accountA limit and exposure engine with pre-trade checks and breach alerting — risk controls, not order routing. The same shape as facet one above, and the reason that facet is not theoretical.
- Order state and reconciliation Own accountA broker-integrated order pipeline where every transition is recorded — placed, modified, cancelled, partial, filled, rejected — and the book is reconciled against the broker’s own record at the end of each session, so the two never quietly disagree.
- Backtest and research infrastructure Own accountA harness where a result can be reproduced from the same inputs months later — corporate actions applied, the data vintage pinned, and every run recorded with its parameters.
- Workflow orchestration Own accountA trigger graph — time, condition and event triggers wired to actions, with every run recorded and replayable.
- Operating view Own accountA unified position, exposure and workflow view. One place where the state of the machine is visible without opening four tools — the same artefact that ships with a build.
Named teams, not a firm directory.
The work lands with four people, and the one who calls owns the number that leaves the firm.
- OperationsOwns the close, the reconciliation and the exception queue.
- ComplianceOwns the cadence, the breach register and the evidence trail.
- ResearchOwns the documents and what can be found inside them.
- The principal officerOwns the consequence of any of the above being wrong.
PMS managers · AIF managers, Category III and Category I/II · family offices · wealth managers and RIAs with reporting obligations.
Also real, lower here for prominence rather than fit: hedge funds · VC and PE · research firms · brokers.
- You need investment advice, research recommendations or trading calls
- You need execution or order management
- You have engineers who build this in-house
What your vendor assessment will ask, answered up front.
You are obliged to assess us in writing before you engage us. Rather than make you extract it call by call: The full pack is one email away, and the detail is on Privacy and Terms.
We build systems our clients own and operate. We do not provide investment advice, research recommendations or trading calls, and we are not a SEBI-registered Research Analyst or Investment Adviser.
- Where it runsIn your environment, on your infrastructure and your accounts. Client-hosted by default.
- What leaves your networkNothing, by design. Where a third-party model is used, it is your account and your contract, and the boundary is documented before we build.
- What is loggedEvery input, every decision and every output, with replay. The audit trail is the product, not an add-on.
- Who touches itThe principal who scoped it. Named individuals, disclosed before the engagement starts.
- ContinuityThe code, accounts and documentation are yours from day one. If we disappear, the system does not.
- Sub-processorsListed in the pack, with what each one sees.
Monitoring & Data Diagnostic
Two weeks. We map one workflow end to end against your real files, and give you a written teardown: what breaks, where, what it costs when it does, and whether a system is worth building.
$2,500, fixed, credited against a build. The verdict can be no. The build itself is quoted after the diagnostic, because the scope varies with what the diagnostic finds.
- Week oneThe workflow as it actually runs — who touches it, what they open, where it breaks, what the last three exceptions were.
- Week twoThe system as it would be built, with the data foundation named, the failure modes named, and an estimate you can take to a partner.
- You receiveA written teardown, a target architecture, and a build/no-build recommendation with the reasoning attached.
- And in the build that followsAn operating view — what the system checked, what it flagged and why, what it could not reconcile, and the evidence behind each. A window onto the machine, not a substitute for one.