AI that builds.
One idea in — a product plan,
a trading engine, a customer brain out.
Smritixs Tech Solutions is a four-product AI platform. Our flagship Forge is an AI Development Life Cycle — five agents turn a one-line idea into a vision, data model, architecture, build backlog, QA plan and a costed resource plan in seconds. Alongside it run ARTAI, our advisory-first Indian F&O markets-intelligence engine, a customer-AI suite — Pulse and Helpdesk — and Palette, an AI brand & design studio that turns a brief into a complete, WCAG-accessible design system. Every product ships with a self-contained, no-login live demo.
Four products, one AI platform
One pnpm TypeScript monorepo, one design language, four AI products that solve real work — building software, trading with discipline, understanding customers, and designing brands. Each has its own self-contained, offline, no-login live demo.
Forge — AIDLC Studio
An AI Development Life Cycle. Five agents — Master Orchestrator, Inception, Construction, QA and Resources — take a one-line product idea and produce a vision + user stories, a data model, an architecture, a build backlog, a QA plan with quality gates and a costed resource plan + readiness scorecard, in seconds. Domain-aware: fintech, health, commerce, devtools each get different entities, NFRs (PCI / HIPAA / GDPR), stack and cost.
ARTAI — Markets Intelligence
An automated Indian F&O advisory engine: multi-timeframe candles, opening bias, real strike selection, self-building and scored strategies, hard risk controls + a kill switch, and an advisory → paper → live safety ladder. Broker-agnostic across Zerodha Kite and Angel One.
Pulse — Customer Pulse
Turns raw customer feedback into a pulse score, NPS, the themes that matter, and a clear list of at-risk accounts — so teams see how customers really feel before churn shows up in the revenue.
Helpdesk — Ticket Triage
Reads an incoming support ticket and returns a category, severity, churn risk, the right routing and SLA, plus a suggested reply — turning a noisy inbox into prioritized, ready-to-action work.
Palette — Brand & Design Studio
From a brand brief it generates a complete, WCAG-accessible design system — color roles with 50–900 scales, semantic colors, a type scale, spacing, radii and shadows — plus a contrast/accessibility audit and exports as CSS variables, JSON or Tailwind. Real color science: a pure @artai/design engine with exact WCAG contrast math (33 tests), served by design routes on @artai/services-api.
A retail boom — and a discipline crisis
A closer look at ARTAI, the markets-intelligence product in our suite. Millions of new traders entered India's options market in a few short years. Most are fighting the market with emotion and no system. The data is brutal and well documented.
Why they lose
The market doesn't beat retail traders. Their own behaviour does.
- Emotion over process — revenge trades, FOMO entries, holding losers and cutting winners.
- No risk discipline — no pre-set stop-loss, no position sizing, no daily-loss limit.
- No systematic edge — gut-feel entries with no tested, scored strategy behind them.
- Wrong strikes, late exits — picking contracts and timing by intuition, square-off after the move.
- Opaque tools & hype — "tip" channels and black-box bots promising fake accuracy.
Directional context from SEBI's public studies of individual equity-derivatives traders and NSE/CDSL/NSDL demat and turnover data. Figures cited as directionally representative of the segment, not a precise point estimate.
ARTAI — process, not predictions
ARTAI watches the market in real time, decides with a tested system, and refuses to let emotion or risk run wild. In plain language: it tells you what to trade, which strike, the entry, stop-loss and exit — and never lets a bad day spiral. Real, in-code today
Multi-timeframe candles
15s · 30s · 1m · 2m · 3m · 5m candles built from the live broker tick stream — the same multi-frame picture a disciplined desk trader watches, computed continuously.
Opening bias → CE / PE
Reads how the day opened: gap vs. previous close, the opening range and VWAP, and resolves a directional bias — bullish → CE, bearish → PE, unclear → wait.
Real strike selection
Turns spot price into a concrete, tradable contract — ATM by default, configurable ITM/OTM — resolved from the live instruments dump (e.g. NIFTY 24000 CE), not an abstraction.
Self-building, scored strategies
A library of parameterized strategies that the engine grows itself. An end-of-day evaluator scores each on win-rate, P&L, expectancy and drawdown, and promotes the best for the next session.
Hard risk controls + kill switch
Per-trade stop-loss, position sizing, max concurrent positions, a daily-loss kill switch, a trading-hours guard and EOD square-off. The risk manager is the only authority that can block a trade.
Advisory-first safety ladder
Three rungs — ADVISORY → SIM (paper) → LIVE. It ships in advisory by default and places no orders; LIVE is double-gated. Safety is the product, not a setting.
Pluggable AI sanity-checks
Optional Claude or OpenAI annotates each signal with a non-blocking, explainable second opinion using per-workspace prompts. The AI never places or vetoes orders — it adds clarity, not control.
Broker-agnostic, cross-platform
One engine, two brokers — Zerodha Kite and Angel One SmartAPI — feeding a single Expo app that runs on iOS, Android and web from one codebase.
Config-driven workspaces
Every screen is a JSON workspace composed from generic, reusable widgets. Unlimited workspaces, editable prompts and layouts — the product extends without a rebuild.
Live, explainable alerts
Real-time ENTER / EXIT alerts with entry, stop-loss, target and a written rationale, streamed over WebSocket to the app — every call comes with its "why".
A massive base, perfectly mistimed for discipline
The retail options wave already happened. What's missing is the systematic, safety-first layer on top. The sizing below is illustrative; the market context is directionally real.
Market sizing Illustrative
TAM / SAM / SOM shown as illustrative ranges for a subscription markets-intelligence product. Not a forecast of achieved revenue.
Largest by contract count. India is the world's largest equity-derivatives market measured by number of contracts traded — the deepest possible top-of-funnel for an options product.
Demat accounts and climbing. Well over 100 million demat accounts, with retail participation in options that has surged over recent years.
A clear, urgent pain. SEBI has repeatedly flagged that the large majority of individual F&O traders lose money — the exact discipline-and-risk problem ARTAI is built to fix.
Regulatory tailwind. A market pushed toward safer, more transparent participation rewards an advisory-first, risk-controlled, explainable tool over black-box "tip" bots.
Why this is hard to copy
The defensibility isn't one feature — it's a shared platform that turns a one-line idea into a costed build plan (Forge), an integrated, safety-first trading pipeline where risk, not a model, holds final authority (ARTAI), and a design engine with exact WCAG contrast math that turns a brief into an accessible design system (Palette). All of it engineered to a real bar.
Forge — the AIDLC engine
A one-line idea flows through five agents and comes out as a complete, domain-aware, costed build plan with a readiness scorecard — deterministic by default, with optional Claude enrichment.
A Master Orchestrator runs the four specialist agents and assembles the plan. Domain-aware: fintech, health, commerce and devtools each get different entities, NFRs (PCI / HIPAA / GDPR), stack and cost. Production API @artai/aidlc-studio with SSE streaming, persistence and optional LLM enrichment — try it at /aidlc.
Every recommendation passes through the risk manager before it ever reaches a human — and orders only flow when the safety ladder is explicitly armed.
Forge: idea → costed plan engine
Five coordinated agents turn a one-line idea into a vision, data model, architecture, backlog, QA plan and a costed resource plan + readiness scorecard. Domain-aware (fintech / health / commerce / devtools → different entities, NFRs, stack, cost), deterministic by default with optional LLM enrichment — a reusable AIDLC, not a one-off prompt.
Engineered to a real bar
One pnpm TypeScript monorepo: 190 unit tests passing and a full typecheck clean across 12 projects. Every product ships a self-contained, offline, no-login live demo — proof you can run, not screenshots you have to trust.
Real-time multi-TF engine
Sub-minute candles built forward from live ticks (Kite historical backfill is 1m+; finer frames build from the live stream) — a non-trivial streaming pipeline that's already running.
Self-building strategy loop
The engine generates strategy variants and an EOD evaluator scores and promotes the best for tomorrow — a compounding, data-driven loop that improves with every session.
Risk-as-sole-authority
Architecturally, only the risk manager can block a trade and AI can never place or veto orders. Safety isn't bolted on — it's the structural center of the system.
Config-driven, broker-agnostic
JSON workspaces and a two-broker engine (Kite + Angel One) mean new screens, prompts and integrations ship as config — fast iteration, low switching cost for users.
Advisory-first SaaS, low regulatory surface
Recurring subscription revenue from a tool that, by default, only advises — it places no orders unless the user explicitly arms execution. Pricing below is illustrative; see /pricing for live plans.
Starter
- Clearly-labeled synthetic replay demo
- Candles, bias & advisory walkthrough
- One workspace
Trader
- Live broker feed (Kite / Angel One)
- Advisory + paper (SIM) modes
- Full risk controls & kill switch
- AI sanity-checks & alerts
- Unlimited workspaces
Pro
- Everything in Trader
- Double-gated LIVE execution
- Custom strategies & prompts
- Priority support
Expansion paths: broker partnerships and referral revenue, premium market-data tiers, and a future marketplace of community strategies — each scored by the same EOD evaluator that powers the core engine. All figures above are illustrative targets, not achieved results.
Built and verified — with an honest next list
We ship continuously and we don't fake results. Here's exactly what exists in code today versus what's on the roadmap, clearly labeled.
Shipped & verified
Roadmap · projections
Honesty is the policy. Every product capability marked "Shipped/Verified" exists in the codebase today — across Forge, ARTAI, Pulse, Helpdesk and Palette. Every company-level metric — users, revenue, AUM, raise, pricing, market size — is a clearly-labeled target or illustration, never presented as already achieved. Demos are clearly labeled and self-contained; ARTAI's public demo runs on synthetic data and its numbers are never real trading results.
Engineered, not assembled
Smritixs Tech Solutions builds intelligent software that works in the real world. In a space drowning in hype, our edge is the boring stuff done well: safety, honesty and shipping velocity.
Engineered, not assembled
A real pnpm monorepo — four AI products, decoupled engines and apps, 190 tests, a clean typecheck across 12 projects and a live demo each — not no-code wrappers around someone else's models. Built to extend and to last.
Ships continuously
We push to main and watch it go live. The product moves every week — the pace investors want to see between rounds is already our default operating speed.
Safety + honesty as a moat
Advisory-by-default, risk-as-sole-authority, and a refusal to fake "accuracy" or results. In a market the regulator is cleaning up, trust compounds — and we're built for it.
The ask
A target seed round to take the platform — flagship Forge, ARTAI, the customer-AI suite and Palette — from verified engines to polished, distributed products. Amount and allocation below are an illustrative target, not a closed round.
F&O trading is high-risk and can lose money rapidly. No software guarantees profit or "accuracy." ARTAI is advisory by default and nothing here is financial advice. All company-specific figures are targets / illustrations.