Buy commodity tools when the vendor owns the artifact. Build or hire a firm-owned context layer on CRM, custodian, and the archive, scoped to one internal workflow, when off-the-shelf cannot file the packet. For US RIAs between $1B and $10B AUM.
How to build a custom AI system for a wealth management firm is not a platform program. It is a context layer on your CRM, custodian, and compliance archive, scoped to one internal workflow, with a licensed model and a human on the approval step. For US RIAs between $1B and $10B AUM, you build that layer when off-the-shelf tools stop at a single task and the work has to cross systems and survive an exam.
Buy the note-taker and the document tool when the vendor already owns the audit trail. Build or hire the context layer when the job has to read CRM fields, custodian positions, and the archive, then write a packet a person already owns.
Do not start with a firm-wide custom platform. Start with one painful internal process: onboarding, compliance review, fee billing, or trade exceptions. If that workflow does not run in production, you have not built a system.
howtheF designs and operates that internal layer. We do not sell a client-facing chatbot.
Off-the-shelf is the right buy when the problem is universal and the vendor already retains the artifact. Meeting transcription, basic document scan, and template drafts are that category. Jump, Zocks, and similar products do one job well.
Off-the-shelf fails when:
Schwab's public guidance is that firms are moving from experiment to strategy. Strategy here is composition, not a seventh SaaS login. Hartford Funds and WealthTech Today both treat operations and data discipline as the reason AI sticks. That is the diagnosis. Custom work is the treatment when the commodity layer cannot file the packet.
A longer treatment of the composition choice is here: Build vs buy AI for RIAs.
The context layer is the firm-owned retrieval and policy shell between licensed model access and your books-and-records systems.
You do not build the model. You license model capability through a contracted API. You build retrieval over the fields that workflow needs: CRM household data, IPS documents, meeting notes, portfolio snapshots, fee contracts. You keep embeddings and indexes inside your security boundary. A policy shell redacts NPI before a model sees a payload. The model drafts. A person who already owns the process approves. The artifact lands in the system your examiner already knows.
That is custom AI at a wealth firm. It is not a chatbot trained on your brand voice. It is not a client-facing assistant.
Do not paste households into unmanaged ChatGPT while you figure out custom later. That pattern is covered in Can financial advisors use ChatGPT with client data.
Implementation is not a pilot login. It is a workflow with a source of truth, a retention rule, and a human approval step.
Good first scopes: onboarding packets, advertising or communications review, fee exceptions, trade breaks, internal reporting assembly. Bad first scopes: a firm-wide knowledge bot, a client advice assistant, or "AI everywhere" on the CRM.
You can buy the note-taker in week one. You should not fund a second custom workflow until the first has a replayable audit trail.
The five-step sequence is the same as How to implement AI at an RIA. Custom work is scoped per workflow, not as a platform program.
Most $1B to $10B RIAs do not have a standing engineering team that can own model upgrades, connector breakage, and exam evidence. That is the default case for hiring.
Ask five questions. If the answers are vague, keep looking.
howtheF's refusal list is short and public: no client-facing advice bots, no NPI in plaintext prompts, no "set and forget" agents on regulated work.
The longer partner screen is here: Who helps RIAs implement AI.
We start with one internal workflow, not a firm-wide custom platform. The first deliverable is a production path: data stays put, the model is contracted, a person signs off, and the artifact lands in the system your examiner already knows.
If the first workflow does not run in production, we do not expand the scope.
When the work crosses CRM, custodian, and the compliance archive, and no vendor owns the exam artifact for that packet. Universal jobs (notes, transcription, template drafts) stay off-the-shelf.
No. You need a mapped source of truth for the fields that one workflow uses. Warehouse programs are how first implementations die.
US RIAs between $1B and $10B AUM. That is the band with enough operational complexity to need a context layer and usually without a standing engineering team to own it alone.
Time-to-value should be measured in weeks on one workflow, not a multi-year platform program. If a partner cannot name the first production artifact, they are selling strategy, not a system.
Custom internal AI for wealth firms. The product is the operating system around your existing stack: retrieval, policy, approval, and the workflow UI your team uses every day. No client-facing chatbot.
Automate RIA client reporting as internal ops: retrieve verified performance data, draft commentary, require human review, and publish to the portal you already have. Not a client-facing advice bot and not a new reporting platform.
An AI CRM is not a native feature on a vendor roadmap. Keep Wealthbox, Redtail, or Salesforce. Buy the note-taker. Own a context layer across CRM, custodian, and the compliance archive, with human approval before anything writes to a household record.
AI tools for SEC compliance help a CCO review ads, email, and books-and-records. They are not a chatbot that signs off. Buy commodity surveillance when the vendor owns the exam artifact. Build or hire the context layer when work crosses CRM and the archive. The human stays accountable.
See how HowTheF.ai can help your firm implement AI that actually works.