Current focus
Expand the asset universe
Grow beyond stocks, crypto, commodities, indices, and forex into bonds, real estate, art, and other investable or wealth-bearing assets.
Methodology docs
How iPulse AI builds transparent AI market intelligence
This is the public documentation hub for iPulse AI. It connects the full flow: AI Agents, prompt assembly, global macro context, model and mode configuration, forecast outputs, consensus scoring, and Asset Snowflake Analysis.
Updated July 16, 2026
Production state
The methodology currently active in iPulse AI
Active release 2026.07.16
Current iPulse AI production methodology
The active public methodology covers governed multi-agent configurations, reusable prompt assembly, global context, forecast horizons, consensus scoring, Top Picks, Asset Snowflake analysis, model lineage, and data provenance.
- Effective from
- July 16, 2026
- Last production validation
- July 16, 2026
- Active components
- 10
Active
- - Governed AI Agent configuration registry
- - Reusable prompt assembly
- - Global macro and fundamental context
- - Thinker and Researcher execution modes
- - One-year and five-year forecast horizons
- - Multi-agent consensus scoring
- - Top Picks ranking and filtering
- - Asset Snowflake risk and resilience analysis
- - Forecast configuration lineage
- - Public-first methodology documentation
Experimental
- - Expanded cross-asset portfolio construction
- - Formal forecast calibration and performance releases
- - Additional frontier model and execution-mode comparisons
Deprecated
- - Legacy single-answer methodology descriptions without configuration lineage
Version authority: Git preserves the exact React, TypeScript, schema, rendering, and route implementation. This operational registry provides the compact public release state and effective-dated change lineage.
Methodology overview
How the iPulse AI research methodology is organized
iPulse AI is built around comparison, disagreement, and traceability. Instead of asking one model for one answer, iPulse AI runs structured AI Agent configurations across assets, horizons, data inputs, and output schemas, then packages the results so users can compare individual advisor views and consensus signals.
The terms AI Agent and advisor refer to the same public concept in these docs. In some product surfaces, you may still see advisor language because the forecasting experience shows individual advisory voices. The docs use AI Agents as the primary public label.
More precisely, an iPulse AI Agent is a governed forecast configuration: persona, archetype, execution mode, model/backend assignment, asset-class reasoning head, communication profile, task rules, and output schema working together. The named public profiles explain the recognizable persona families, while the real product already operates 100+ AI Agents across those configuration combinations. That number can expand quickly as new frontier LLM models, modes, and asset-class heads are evaluated and added.
What automation changes
A day of automated research. Ready in one click when complete.
Recreating the full workflow by hand is estimated to require about 600 hours of nonstop prompt entry and more than $5,000 in model usage per comparable deep-analysis run. iPulse AI manages the prompts and runs the deep analysis automatically. A full run can take up to one day. When it finishes, the result is cached and opens immediately.
- Estimated model budget per comparable deep-analysis run
- $5,000
- Nonstop prompt typing
- ~600 hours
- Automated deep-analysis run
- Up to 1 day
- Cached access after completion
- Instant
The $5,000 model budget and 600-hour manual-prompting figures are internal estimates for recreating a comparable deep-analysis run. Actual costs and completion times vary by provider, model, usage, and operator.
Transparency boundary: iPulse AI explains the architecture, components, and configuration lineage behind forecasts, but it does not publish proprietary full prompts or private operational records.
Author and reviewer signal
Written and maintained by Future Edge Group Team. Reviewed for product and technical alignment by Russlan Ramdowar, Founder and Hands-On Technical Architect.
Methodology reviewed July 16, 2026
Product direction
Our ambition: the best agentic AI portfolio manager
iPulse AI is broker-agnostic: the platform is not designed around one broker, exchange, custodian, or transaction venue. We are expanding the number and diversity of supported assets while improving forecast quality, calibration, repeatability, and horizon coverage.
That ambition starts with building the best agentic AI stock picker across a growing asset universe, then extending the same transparent forecasting discipline into end-to-end portfolio management. The intended portfolio view should help users estimate how total wealth may evolve over one year, five years, and all the way to retirement, with uncertainty and scenario ranges made visible rather than presented as certainty.
Current focus
Strengthen forecasting maturity
Improve forecast power through richer data, broader model comparison, repeated batch evaluation, calibration, and transparent performance tracking.
Next phase
Deliver end-to-end portfolio management
Add cross-asset correlation, concentration and scenario risk, risk hedging, allocation optimization, and portfolio guidance aligned with each user’s risk appetite.
Execution remains independent: planned portfolio tools are intended for research and decision support, not brokerage or custody. Users retain control of execution through the broker, bank, real estate agent, or specialist they choose. Each capability will be documented as it becomes available.
Video walkthrough
A quick look inside iPulse AI
Product demo
iPulse AI Demo: AI Market Research, Top Picks, and Advisor Forecasts
Watch how iPulse AI turns market data, AI Agent analysis, company fundamentals, Top Picks, and forecast signals into a transparent market research workflow.
Demo narration transcript
This is the narration script used for the embedded product demo, included for readers, screen readers, and search engines.
- 1.Most market tools show prices and headlines. iPulse AI is built for the next step: finding what may be worth researching now, and understanding why.
- 2.The Top Picks view starts with a ranked market shortlist generated from deep analysis across the live iPulse AI catalog. Before opening anything, the table lets users compare score, consensus rating, expected return, dividend yield, valuation, volatility, and forecast price side by side. Here, Shell stands out near the top.
View Full TranscriptHide Full Transcript
- 3.Opening Shell turns a ranked idea into a full asset research page. The page begins with company context, industry peers, a flagship thesis, and forecast scenarios across one-year, three-year, and five-year horizons.
- 4.The first layer summarizes the core thesis in plain language. It explains why the opportunity exists, what drivers matter, and how the forecast connects to the current market setup.
- 5.The AI Forecasts tab then moves into the advisor workflow. iPulse AI shows forecast paths, scenario ranges, and individual advisor predictions, so the consensus can be inspected instead of treated as a black box.
- 6.Each advisor applies a different investment lens. The user can compare how different personas interpret the same asset, then open an individual report to understand the reasoning behind a specific prediction.
- 7.Inside the advisor report, the logic becomes reviewable: the most rational scenario, the key drivers, the risks, the opportunity areas, and the assumptions behind the forecast path. This makes disagreement and uncertainty visible, including where conviction is strong and where the thesis is more fragile, rather than hiding it behind one score.
- 8.The workflow then connects the AI thesis back to fundamentals. The company profile gives business context, while earnings and profitability show revenue, net income, free cash flow, valuation, and performance history.
- 9.The balance sheet tab adds leverage, liquidity, assets, debt, and equity context. Dividends and actions show how capital is returned or reinvested, helping users connect the forecast to real financial behavior.
- 10.Finally, search lets the user jump from Shell to another asset, like NVIDIA, and repeat the same research process. iPulse AI helps investors discover, assess, deep dive, and compare market opportunities with transparent AI reasoning. It is research and decision-support software, not financial advice.
Documentation map
Follow the methodology in order
Architecture
How subjects, data buckets, lineage, agents, modes, and tasks fit together.
Prompt Assembly
How iPulse AI builds forecast prompts from reusable components.
Configuration Details
How eligible users inspect the assembled configuration behind each forecast.
Models
How model configuration fits into the prompt and forecast workflow.
Global Context
Why macro events, fundamentals, and model knowledge cutoffs matter.
AI Agents
The public AI Agent and advisor profile library.
Thinker vs Researcher
When iPulse AI uses reasoning-first or research-enabled execution.
Consensus Score
The formula that turns advisor forecasts into a compact signal.
Top Picks
How rankings, table columns, dividends, risk, and financial health are displayed.
Asset Snowflake
How Event Risks, Financial Health, and snowflake axes are calculated.
Forecasting
How horizons, scenarios, and forecast interpretation work.
Data Provenance
How lineage, schema discipline, and methodology changes are documented.
Changelog
Effective-dated production methodology releases and material changes.
Built experience
The methodology reflects lessons from operating the platform
The docs are not a theoretical prompt-engineering essay. iPulse AI has evolved through practical data-platform, forecasting, serving, and public-page reliability work. Several current design choices exist because earlier approaches were too brittle, too hard to audit, or too easy to misread.
Market data must stay re-adjustable
Historic price work taught us that provider-adjusted values can become stale after later corporate actions. iPulse AI therefore treats raw market records, corporate actions, and derived adjusted views as separate concepts.
Forecasts should store relative paths
Percentage-change time series are more robust than absolute price-only targets because splits, dividends, and anchor-date changes can otherwise make a reasonable forecast look mechanically wrong.
Configuration identity must be stable
Personas, modes, task configs, prompt components, models, schemas, and output formats are versioned separately so improvements can be compared instead of silently overwriting history.
Public pages must stay public-first
The public docs and SEO pages avoid protected data reads and heavyweight auth/App Check work on first paint. Live inspection remains gated, but educational methodology stays crawlable and fast.
Research architecture
Many AI Agents, one inspectable workflow
HERE IS HOW iPulse AI WORKS UNDER THE HOOD
Many AI Agents.One Transparent Consensus.
iPulse AI100+ Analysis Frameworks5,000+ AI SETUPS EXECUTED IN PARALLEL
Runs each individual asset (stock, crypto, commodity etc.) through analysis frameworks across best AI models in the world. Aggregates all reports of an Asset into a single iPulse AI Consensus Score. Ranks all assets from best to worst. Users can inspect the latest and all historical reports. Full Trust.
Average User
Asks Claude for deep research on stocks to buy. Doesn't know which investment framework is used for analysis. Can not guarantee AI is aware of latest macro-economic events. Cannot track past performance. No visuals for inspection. No Trust.
Control Animation
Queue
Bitcoin
NVIDIA
Gold
Apple
Alphabet
BitcoinNVIDIAGoldAppleAlphabet1. Asset Enters iPULSE Pipeline
Coca-Cola
CEO
Market data
Events
News
Earnings
Dividends
Filings
Reports
2. Swarm of AI Advisors Produce Independent Investment Reports
Advisors exist in many different configurations of tools, setups and contexts.
100+
Advisors
Configured across frontier models
Warren Buffett
Michael Burry
Elon Musk
Ray Dalio
Machiavelli
JP Morgan
Preparing advisor questions
Waiting for the first advisor to light up.
3. iPULSE Consensus Analysis
iPULSE
Independent advisor reports converge into one iPulse AI Consensus Rating.
Total returnR = price return + dividend uplift
Cash hurdleE = R - cash hurdle
DispersionR_conf = sign(E) x max(0, |E| - MAD discount)
DirectionS = 0.60 x vol-adjusted + 0.40 x return anchor
AgreementBase = S x (0.85 + 0.15 x consistency)
Final ratingScore = Base x (1 - vol penalty) x 1000 +/- event risk
4. Detailed forecast report
Coca-Cola reasoning, forecast path and score drivers.
Forecast+7.8%
iPulse AI Consensus Score+240
SellNeutralBuy
5. Top Picks updates
Signals become a ranked watchlist of what to buy or sell.
Rank #4
#1BitcoinSTRONG BUY+580
BitcoinSTRONG BUY+580#2NVIDIASTRONG BUY+420
NVIDIASTRONG BUY+420#3AlphabetSTRONG BUY+330
AlphabetSTRONG BUY+330#4Coca-ColaBUY+240
Coca-ColaBUY+240Agent diversity
Different AI Agents emphasize value, macro cycles, disruption, forensic risk, political power, market structure, or long-range strategic context.
Prompt assembly
Each forecast is produced from reusable components: persona, mode, tone, register, subject context, data inputs, global context, output instructions, and task guidelines.
Consensus scoring
Individual forecasts are converted into normalized signals and then aggregated with disagreement, volatility, and consistency controls.
Product access
Public docs and gated product inspection work together
These methodology docs are public. The live Prediction Config Details modal is a signed-in product feature. A user must have a subscription plan that unlocks the relevant asset, batch, or advisor report before they can inspect the live configuration behind that prediction.
Locked asset pages can still link to these public docs so visitors understand the methodology before subscribing. The live configuration details stay gated until the user has the required product access.