Services
AI research agents, built into your investment workflow.
Deskside agents read, compare, and flag across every source your team relies on. Each deployment starts with one real workflow and one real user, and goes live on your coverage in weeks.
Proactive agents
Not every useful agent waits to be asked. We build agents that run continuously in your environment, watching filings, prices, data feeds, positions, and inboxes. They act when something changes: they do the first pass, attach the evidence, and bring it to the right person. You set what they watch, when they escalate, and which actions always need a human to approve.
- Filing and news alerts on your positions
- Pre-open briefs, built overnight
- Risk and exposure limit monitoring
- Late or anomalous vendor-data detection
- Trade-break triage with drafted fixes
- Inbox agents for fund-admin and broker email
Research & analyst tools
Your analysts spend hours reading so they can spend minutes thinking. We build agents that do the reading — across filings, transcripts, broker research, expert-network calls, and your own internal notes — and bring back what changed, with citations.
- Earnings-call and filing diffing
- Coverage-universe morning briefs
- Search over internal research notes
- Expert-call summarization and tagging
- Live call flags from your note-taker
- Model-update assistants for Excel
- Thesis tracking and catalyst calendars
Data & alternative data
Structured signals hide in unstructured sources. We build pipelines that extract, clean, and validate data from documents, websites, and vendor feeds, and put it where your quants and fundamental teams already work. Our quant research background means we check point-in-time correctness, look-ahead bias, and coverage before anyone builds a signal on it.
- Cross-vendor signal triangulation
- Automated cohort and retention analysis
- Document-to-dataset extraction
- Vendor-data evaluation and onboarding
- Entity resolution and ticker mapping
- KPI tracking from company disclosures
- Data-quality monitoring and alerting
- Natural-language access to your warehouse
Operations & investor relations
The middle and back office run on email, spreadsheets, and tribal knowledge. We automate the repetitive parts with human review built in, so your ops team spends time on exceptions — not on copying numbers.
- Reconciliation and trade-break triage
- DDQ and RFP response drafting
- Investor letter and report prep
- Counterparty and fund-admin email handling
- Onboarding and KYC document review
- Internal knowledge assistant
Compliance, risk & AI platform
Before AI scales across a firm, it needs rails. We help your technology and compliance teams stand up secure model access, evaluation, and monitoring — and build the compliance tools that benefit from AI themselves.
- Secure, logged internal LLM gateway
- Evaluation suites on your own examples
- E-comms surveillance triage
- Restricted-list and MNPI checks
- AI usage policy and controls
- Vendor and model risk assessment
Engagement models
Start small. Prove it. Expand.
Scoped, fixed-fee engagements with clear outcomes. Pricing depends on scope — we'll give you a number on the first call.
Desk Diagnostic
We sit with your team, map workflows, and deliver a ranked list of AI opportunities with a working prototype of the top one.
- Workflow shadowing
- Opportunity map with ROI estimates
- Prototype on your data
Deskside Build
One or two engineers embedded with your team to take a workflow from idea to production tool the desk relies on.
- Production deployment in your environment
- Evaluations and monitoring
- Full code and documentation hand-off
Embedded Retainer
An ongoing forward-deployed engineer who works as part of your team, shipping new tools and maintaining existing ones.
- Dedicated engineer capacity
- Roadmap set with your leadership
- On-site or hybrid
Questions
Common questions
What is a forward-deployed engineer?
An engineer who works directly with the people using the software instead of from a distant backlog. They learn the workflow first-hand, build quickly, and adjust based on what they see at the desk. It's how the best AI products actually get adopted.
What's a proactive agent?
An agent that runs on its own instead of waiting for someone to type a question. It starts from a schedule or an event, like a new filing, a data file, a limit breach, or an email. It does the first pass and brings the result to the right person. Anything with real consequences, like booking a fix or sending something outside the firm, waits for a human to approve it.
Where does our data go?
Nowhere you haven't approved. We deploy inside your cloud account or on-prem, use model providers under enterprise terms with zero data retention where available, and can run open-weight models entirely in your environment when needed.
Can you work with the data we already license?
Yes, and that's usually where we start. We build inside your environment and within your vendor agreements. Your data never leaves your control, and we never pool or reuse one client's data for another.
Do you share anything between clients?
No. Code patterns and engineering experience carry over; your data, positions, prompts, and outputs do not. Everything we build for you is yours.
Do you work on-site?
Yes, when it helps. Most engagements are hybrid: in person for the first week of shadowing and key milestones, remote with daily contact the rest of the time.
Which models and tools do you use?
Whatever fits your constraints. We're model-agnostic and work across the major frontier and open-weight models, and we build on your existing stack — Python, your data warehouse, Excel, Bloomberg, your OMS — rather than replacing it.
What happens when the engagement ends?
You own everything: code, prompts, evaluation sets, and infrastructure. We document it and train your team to run it. If you'd like us to stay on, the retainer model covers that.
Will you sign an NDA?
Of course — before the first substantive conversation. We're used to the confidentiality expectations of investment firms.
Next step