A smaller review surface
Avoid adding a mandatory third-party client-ledger store to every control questionnaire.
Open Portfolio · for wealth platforms
Run AI over wealth data without building another client-ledger warehouse.
You keep your login and your client data where it already lives. We turn broker and portfolio records into a small approved summary — and only that summary goes to the AI model.
What you get
- You keep the store
Your login and approved storage stay yours. We are not another place to park the client book.
- 19 dedicated broker adapters
Open-source importers normalize broker exports on the user side before anything reaches AI.
- AI sees a summary, not the ledger
The model call receives an approved aggregate — not a raw dump of client holdings for open-ended retention.
- Pocket Portfolio is the live proof
Our consumer app runs the same import pipes under real broker-file mess — so enterprise buyers are not buying a slide.
Keep your cloud
Plumbing fitting, not another vault.
Wealth platforms should not have to copy the client ledger into a new vendor cloud just to add AI. We fit ingestion and the AI hop around the stores and login you already trust.
- 01 · A second store creates a second problem
New retention, access, subprocessor, and incident-response work lands on your security and legal teams.
- 02 · Most AI jobs do not need the whole book
Advice and portfolio questions usually need a few facts or totals — not every underlying row.
- 03 · Generic SaaS asks the wrong question
The review becomes “Can this vendor hold our client book?” instead of “Can this tool work over data we already approve?”
Pocket Portfolio is the live consumer app on the same pipes. Enterprise pilots use your stores — we do not become the warehouse.
Read the reference architecture →How the path worksOpen the full architecture map →
What crosses to the model — and what does not.
Broker and portfolio files are cleaned up where the user or platform already works. You approve what may go into the summary. Only that summary reaches the model call.
Broker files and your stores
Local / edge import
Your login and storage
Approved summary
Model call
Streamed answer
The raw client ledger stays behind this line
Live product proof
Pocket Portfolio proves the pipes. It is not the enterprise store.
Pocket is the consumer app that exercises the same importers under real broker CSV chaos. How Pocket stores data for retail users is separate from the enterprise path, where your platform keeps the book.
View Pocket Portfolio →- Real broker-file import
- Works in the browser first
- Adapters tested on messy CSVs
- Same AI summary path as enterprise
How a design partnership starts
Fit the path around the stack you already run.
We start by mapping your login, approved stores, data sources, what may enter the AI summary, and how the model call is controlled.
Map what you already trust
Confirm identity, storage, jurisdictions, operators, and existing data flows.
Agree the summary contract
Lock exactly which fields and totals may reach the model.
Prove it in a scoped pilot
Test import, visibility, what persists, and what procurement needs to see.
Why platforms buy this
Less warehouse. Clearer diligence.
Copying the book into every AI vendor expands your subprocessor list, your contracts, and your cloud bill. A smaller approved summary removes what you never needed to store twice.
A clearer vendor story
Security, legal, and procurement can see what crosses to the model and what persists.
A scoped path to AI
Move from open-ended experiments to a controlled processor model over approved context.
Regulatory context
Where the data sits is a commercial decision.
DORA, the EU AI Act, GDPR, and vendor-risk review make architecture and processor scope board-level issues — not afterthought paperwork. We do not claim certification you do not hold.
GDPR Art. 83(5)
€20M or 4% of worldwide annual turnover
Maximum administrative fine — what your DPO already weighs against every new data store.
REG-01
EU AI Act Art. 99
€35M or 7% of worldwide annual turnover
The legislation enterprises will be audited against from 2026 onward — architecture and processor scope are diligence inputs, not automatic compliance.
REG-03
Avg. financial-services breach · 2025
£4.45M (~$5.56M USD)
IBM Cost-of-a-Data-Breach 2025. Perimeter width is a commercial risk, not a slogan.
CODB-01
Built in the open
Live product. Public architecture. Operator background.
Pocket Portfolio is the live wealth-tech app that proves the importers. Open Portfolio is the enterprise path on the same pipes. Founder track record: Built a digital portfolio platform supporting decision-making across £7B in energy assets. (National Grid Ventures, 2023–2025) — prior role credential, not an Open Portfolio customer claim.
- Live consumer app
- Public architecture map
- Open import proof
- Approved-summary design
- Founder enterprise track record
Verified receipts (operator-grade)
Curated feedback from the live harness — focused on compliance, procurement velocity, and boundary clarity.
Loading receipts…
Common questions
Straight answers for sharing inside your team.
What do you mean by sovereign / controlled AI?
In plain terms: you keep the client book and the login. AI only sees a small summary you approve. We supply the fitting — import and the model hop — not another vault for the ledger. Deeper map: /architecture.
Does Open Portfolio store client portfolio data in its cloud?
Enterprise path: no — you keep approved storage; we are not the warehouse. Pocket Portfolio is a separate consumer app used to prove importers under real files; that is not the default enterprise contract.
What reaches the AI model (Gemini, OpenAI, or yours)?
Only an approved summary assembled before the call — not the raw client ledger for open-ended retention on a vendor warehouse path. What is allowed is agreed in architecture review.
How is Pocket Portfolio different from Open Portfolio?
Pocket is the live consumer product that stress-tests broker imports. Open is the enterprise boundary for platforms that keep the book in their own stack. Same import ideas; different commercial contract.
When should we use Open Portfolio instead of Plaid or a portfolio data API?
Use Open when you need AI over wealth data without parking the ledger in another vendor — inspectable import and a controlled model hop. Not when you need bank linking or payments. Not a Plaid replacement.
What happens in a design partnership?
We map login, storage, sources, the summary contract, and model controls; then run a scoped pilot. Start at /tier1designpartner or book a diligence call from this page.
Talk to us
Book a diligence call.
Bring how you log people in, where client data lives, what you import, and the AI use case. We will map whether a scoped pilot makes sense. We reply within one working day.