Every Number Traces Back to Its Source

Institutional investment research, reimagined.

Built for long-short equity hedge funds. Every output cites the exact source it came from. Every model exports as an editable, formula-linked file, not a black-box answer. Deployed cloud, private VPC, or fully on-prem.

  • Page-cited answers, every time
  • Formula-linked, editable models
  • Earnings and guidance tracking
  • Regulatory filing ingestion
  • AI research agents
  • On-prem or private VPC deployment
  • Portfolio monitoring
  • SFC-aligned, audit-first design
StockLens LIVE
Avanis AI

Built for modern investment teams

Multi-market Intelligence

Analyze companies, filings, and market data across global public markets.

AI Research Agents

Automate document analysis, monitoring, and recurring investment workflows.

End-to-End Research

From data ingestion and financial models to valuation and portfolio monitoring.

Institutional Deployment

Flexible infrastructure built for long-short equity hedge funds, cloud or fully on-prem.

01. About

An auditable AI research platform, built for long-short equity.

Avanis AI develops software that transforms raw market information into structured investment intelligence, with every claim traceable back to its source.

From exchange filings and earnings calls to alternative datasets and company guidance, our platform automates the research process through specialized AI agents, quantitative analytics, and formula-linked, editable models instead of black-box answers.

Built for long-short equity hedge funds, our infrastructure helps analysts accelerate research and defend every number in an IC meeting. Deploy in the cloud, a private VPC, or entirely on your own infrastructure.

Alternative Data AI Research Agents Financial Models Valuation Earnings Intelligence Guidance Tracking Portfolio Monitoring Knowledge Graph Research Automation Workflow Automation Custom Deployments On-Premise AI
02. Team

Built from inside investment research.

Yash Dutt, Founder & CEO

Yash Dutt

Founder & CEO

  • BBA in Management, The Hong Kong University of Science and Technology
  • Equity research and automation experience across boutique investment firms
  • AI & Automation Intern at EY, advising on enterprise AI adoption strategy
  • Head of Equity Research at I-Circle
  • Leads product vision, finance domain expertise, and full-stack execution
Saanvi Shukla, Chief Technology Officer

Saanvi Shukla

Chief Technology Officer

  • BEng in Computer Science, The Hong Kong University of Science and Technology
  • Deloitte Hong Kong – AI & Data Consulting
  • Y Combinator AI Startup School alumnus
  • Top-14 finish in an Asia algorithmic trading competition
  • Leads AI architecture, infrastructure, and scalable backend systems
Antariksh Verma, Head of Engineering

Antariksh Verma

Head of Engineering

  • BEng in Computer Engineering, The Hong Kong University of Science and Technology
  • Published research in leading computer science venues
  • 15+ hackathon wins across AI and software engineering
  • Leads platform engineering, product development, and technical execution
Anish Khanna, Chief Operating Officer

Anish Khanna

Chief Operating Officer

  • BBA in Professional Accounting and Management, The Hong Kong University of Science and Technology
  • Equity and alternative investment research at Kaleidoscope Capital
  • Co-President, Dawn Advisory (HKUST Startup Consulting Club)
  • Leads operations, partnerships, and go-to-market execution
What We Build

Institutional-grade AI, without the enterprise drag.

Three core practices, one engineering standard: verifiable, auditable, production-ready.

Predictive Modeling

Leverage LLMs and neural networks to forecast market trends with institutional-grade accuracy, backtested, explainable, and tuned to your universe.

Autonomous Workflows

Automate back-office reconciliation, compliance checks, earnings triage, and reporting with AI agents that leave a full audit trail.

Risk & Fraud AI

Real-time anomaly detection systems that adapt to new threat vectors faster than legacy rule-based engines, with human-readable alerts.

Custom Builds

If you can describe it, we can build it.

Large funds pay six figures for this off the shelf. We build it around your exact workflow, for mid-market long-short equity funds trading Asia and beyond.

Scope a Build
01IPO IntelligenceDeal pipelines and scoring for new listings in any market you cover.
02Earnings IntelligenceGrade management guidance, score analyst accuracy, and see whether earnings momentum agrees with the revision trend.
03Sentiment & Influencer TrackingFollow the people who move your names across Substack, X, and interviews. Get flagged when their tone shifts.
04Regulatory & Filings PipelinesFilings pulled from HKEX, BSE, NSE, or the SEC the moment they post, parsed, structured, searchable.
05Thesis & Catalyst MonitoringA standing watch on every position: news, filings, and data checked against your thesis, daily.
06Local Transcription (Listenr)Webcasts, earnings calls, and investor days transcribed on your own machines. Nothing leaves the building.
07Workflow AutomationThe repetitive half of the job, email triage, report assembly, data pulls, finished before you sit down.
08Anything You Can DescribeTell us the workflow that eats your week. We build the thing that does it.
Under the Hood

One screen. Six tool-chains running beneath it.

Most desks juggle five disconnected tools for this. Click a feature, on the screen or below, to see what actually runs underneath, and what it replaces.

StockLens dashboard with interactive feature hotspots

Reads every earnings call and filing, then answers questions with page-level citations.

Replaces
NotebookLM + finBERT, juggled by hand
Runs underneath
LlamaParse parsing → FinBERT sentiment → LLM reasoning
Result
3–4 hrs → 25 min per earnings call

Shocks your holdings against rate moves, FX swings, and macro scenarios, before they happen.

Replaces
Perplexity-style deep research, run manually
Runs underneath
Multi-model reasoning over macro data + your live portfolio
Result
Portfolio stress test in minutes, not a weekend

Builds a working DCF with assumptions lifted straight from the filings.

Replaces
Claude + Excel, stitched together by hand
Runs underneath
Reasoning models + LlamaParse extraction → live valuation model
Result
2 hrs → 10 min per model

Turns a plain-English idea into a ranked shortlist across the whole market.

Replaces
Manual screening in spreadsheets
Runs underneath
Text-to-SQL + semantic re-ranking over 3,000+ names
Result
Full-universe shortlist in <10 seconds

A fund-specific agent that runs your research routines on schedule, built around your fund's own workflow rather than a generic template.

Replaces
OpenClaw / Hermes-style agent stacks
Runs underneath
Custom agent loop trained on your fund's workflow
Result
Grunt work done before market open

News triage across every market you cover, pushed as alerts on your names only.

Replaces
60–90 minutes of manual news reading
Runs underneath
LangGraph agent pipeline + LLM triage → push alerts
Result
<5 min morning routine across 6 markets
How We Build

One connected system, not five disconnected tabs.

Most research desks stitch these tools together by hand, copy-pasting between tabs. When we build you a custom solution, the same capabilities run as one pipeline. Watch the flow.

Stage 01 · Ingestion Documents In
NNotebookLM

Filings, transcripts, and decks ingested and indexed

Stage 02 · Analysis Models & Research
Claude GChatGPT PPerplexity FfinBERT

Valuation, deep research, and sentiment scoring

Stage 03 · Agents Routines Run Themselves
HHermes Agent OOpenClaw Claude Agents

Scheduled monitoring, triage, and alerts

Output Your Investment Intelligence

Signals, models, briefings, and alerts delivered into your workflow, cited and auditable

Tool-agnostic by design, we pick the right engine per stage and swap it out as better ones ship, so your workflow never depends on one vendor's roadmap.

03. Case Studies & Engineering

Systems we have designed and built.

Real platforms, built for real desks. Case study write-ups are coming as these engagements mature.

Quant Platform

StockLens Research OS

Research platform for long-short equity desks: earnings-call guidance extraction and credibility checks against prior-quarter actuals, citation-verified Q&A over transcripts, automated formula-linked DCF export. In active development, not yet deployed to a client.

Market Intelligence

Earnings Signal Engine

Signal aggregator fusing analyst revisions, transcript narrative shifts, and management guidance accuracy into a single conviction score per ticker.

IPO Intelligence

HKEX IPO Scout

End-to-end IPO intelligence for Hong Kong listings: heuristic and machine-learning deal scoring, exchange-data backfills, and a pipeline view from first filing to debut.

Workflow Automation

Listenr Meeting Intelligence

Records, transcribes, translates, and summarises webcasts, meetings, and podcasts, pushing structured, searchable notes straight into the firm's document store.

04. Research Hub

Automation, explained for long-short equity desks.

Practical essays and demos on AI automation for hedge fund research desks, written from real, working builds, not theory.

Articles

Generative AI

The Future of Agentic Workflows in Banking

Why the next wave of automation is agents that own outcomes, not scripts that follow steps, and what that means for compliance.

June 2026 · 5 min read

Read Article →
Machine Learning

RAG vs. Fine-Tuning for Financial Data

A practical decision framework: when retrieval wins, when fine-tuning pays off, and the hybrid pattern we deploy most often.

May 2026 · 5 min read

Read Article →
FinOps

How I'd Build a Compliance AI in 30 Days

A practitioner's blueprint: the stack, the guardrails, and the audit-trail design I'd use to ship a regulator-ready compliance assistant in one month.

April 2026 · 6 min read

Read Article →
Deep Learning

Demystifying Transformer Architecture for Traders

Attention, embeddings, and context windows explained through order books and price series, no PhD required.

March 2026 · 6 min read

Read Article →

Reading about automation is the start. Watching it run on your portfolio is the point.

See Case Studies Talk to Us
05. Values

What we optimize for.

Five principles sit under every platform we ship and every engagement we take.

01Trust FirstWhen institutional clients hand us the workflows their decisions run on, we earn that with NDA-first engagements, transparent builds, and audit trails on everything the system reads, writes, or recommends. Your data never trains public models.
02Deterministic Over ProbabilisticWhere a number feeds a decision, we prefer rules and reproducible pipelines over model guesses. AI drafts; deterministic systems verify. Same input, same output, every time.
03Accuracy Before SpeedA fast wrong answer costs more than a slow right one. Every claim our systems make carries a citation back to the source page, so an analyst can check the work in seconds.
04Efficiency That Shows Up TwiceAutomation has to pay for itself in both currencies: money saved and time recovered. If a workflow does not clearly return more than it costs, it does not ship.
05Built on RelationshipsWe plan to co-develop with the teams who use the software — iterating weekly with analysts on the workflows they actually run, not a ticket queue.
06. FAQ

Questions institutional buyers ask first.

Where does our data live? Can it leave our infrastructure?

Deployment is your call: fully on-premises, your private cloud, or our managed environment. For funds with strict mandates we build local-first, documents, models, and outputs never leave your network, and we work under NDA from the first conversation.

How is this priced?

Annual per-firm licence tiered by team size, plus scoped builds for custom platforms. A pilot on your live workflow comes first, so you see measured time savings before committing.

How long until we're live?

Pilots deploy in 2–4 weeks on your real portfolios. Full custom platforms typically ship in 6–10 weeks, iterating weekly with your analysts.

Who is this for?

Mid-market long-short equity hedge funds in Asia, teams that need institutional-grade, auditable AI without enterprise pricing or year-long implementations. If your analysts need to defend every number in an IC meeting, this is built for that specifically, not as a general-purpose research tool.

Hong Kong

Built in Hong Kong. Deployed anywhere.

We work from the middle of Asia's capital markets, in the same time zone as the exchanges we cover, and ship to funds worldwide.

07. Contact

Let's engineer your financial edge.

Tell us about your fund, platform, or workflow. A member of Avanis AI replies within 24 hours.

NDA-first engagements On-prem / local-first available Your data never trains public models Security practices →
Book a Meeting Pick a time that works for you. Added straight to both our calendars.
or send a message

Prefer email? Avanis.AI@outlook.com