Practice · VII — Bot development

Bots that think. Quiet decisions.

Türk Otağ builds production-grade bots — from autonomous trading systems to multi-agent AI. Risk budget, deactivation protocol, observability — all at the architecture layer. Not promises; engineering.

— Philosophy

A bot is not a good decision-maker; it is a good neutral worker. It does not sever decision logic from the human entirely — it takes its limits from the human. Türk Otağ does not guarantee returns; it guarantees risk management, kill-switches, observability. The value of a bot is measured not in what it earns, but in how many quiet nights it allows its owner.

Scope

What's included

  • 01
    Strategy architecture Decision logic, risk budget, capital allocation, halt rules.
  • 02
    Exchange integration Binance, Bybit, OKX, Kraken, FX, MT5, custom REST/WebSocket.
  • 03
    Multi-agent coordination Agents that consult each other and step aside in silence when required.
  • 04
    AI integration LLM calls, RAG, prompt engineering, agent chains.
  • 05
    Risk management Tiered positioning, max drawdown, deactivation protocol.
  • 06
    Backtest & forward test Historical data testing, paper-capital field trials.
  • 07
    Observability Grafana dashboards, alerting, anomaly detection, audit trail.
  • 08
    Launch & operations VPS / cloud setup, continuous monitoring, monthly reports.
Process

Five stops, one discipline.

  1. I —

    Discovery

    We listen for your strategy, assets, risk threshold, target horizon. The brief is ours to write.

    1 week
  2. II —

    Architecture

    Decision diagram, agent composition, risk parameters. Everything clear on paper before production.

    1-2 weeks
  3. III —

    Build

    Modular code, comprehensive testing. Backtest results shared weekly; no surprises.

    3-8 weeks
  4. IV —

    Forward test

    2-4 weeks of field trials with paper capital or small real capital. Patches, calibration.

    2-4 weeks
  5. V —

    Launch

    Production deploy, monitoring panel, runbook. 60 days of continuous atelier presence.

    1 week
Editions

Six editions, one discipline.

Starter

From $800

Single strategy, single exchange, plain risk management. For coding your existing strategy or a simple indicator-based bot.

  • Single strategy
  • Single exchange integration
  • Basic risk management (stop-loss, max position)
  • Backtest report
  • Telegram alerts
  • VPS deploy
  • 30 days of support
Timeline: 3-5 weeks Share a brief →
— Most chosen

Mid

From $2,500

Multi-strategy, multi-asset, real-time monitoring panel. For traders working with serious capital.

  • 3-5 strategies
  • 2+ exchange integrations
  • Advanced risk management
  • Backtest + forward test (4 weeks)
  • Grafana monitoring panel
  • Slack/Telegram alerts
  • Monthly performance report
  • 60 days of support
Timeline: 6-10 weeks Share a brief →

Professional

From $6,000

ML-based, multi-agent system. Production scale, anomaly detection, automated calibration. Quant-fund grade.

  • ML / multi-agent decisions
  • Unlimited strategies & assets
  • Multiple CEX + DEX
  • Advanced portfolio management
  • Anomaly detection & auto-halt
  • Full DevOps + monitoring
  • Audit trail
  • Monthly improvement cycle
  • 90 days of support
Timeline: 10-16 weeks Share a brief →

Starter

From $400

Single-platform Telegram or Discord bot. Plain commands, notification flows, small community management.

  • Telegram or Discord
  • Command-based
  • Webhook notifications
  • Light admin panel
  • VPS deploy
  • 14 days of support
Timeline: 1-2 weeks Share a brief →
— Most chosen

Mid

From $1,300

Multi-channel bot with AI replies. Customer support, community management, content moderation.

  • Multi-channel (Telegram + Discord + Web)
  • LLM-based natural language replies
  • RAG (your own documents)
  • Conversation history
  • Full admin panel
  • Analytics
  • 30 days of support
Timeline: 3-5 weeks Share a brief →

Professional

From $4,000

Multi-agent system — agents that consult, hand off and step aside in silence. Production-scale customer operation.

  • Multi-agent coordination
  • Task handover & specialisation
  • CRM / e-commerce / calendar integration
  • Voice & vision mode (optional)
  • Observability (LangSmith etc.)
  • A/B testing infrastructure
  • Continuous prompt engineering
  • 60 days of support
Timeline: 6-10 weeks Share a brief →

— Prices are not absolute; every bot carries its own complexity and risk. The figures below are starting points. A precise quote follows the first conversation.

— For special requests

Atelier Bespoke

From $8,000+

Custom bot architectures for defence, fintech and production scale. Continuous partnership, SLA, custom security & compliance. We draw the scope together and walk together for a long time.

  • Defence / fintech-grade architecture
  • Multi-agent, tiered trust
  • Custom security & compliance
  • SLA & 24/7 monitoring
  • Backup architecture & deactivation
  • Custom ML model training
  • Monthly audit & improvement
  • Continuous engineering partnership
Open a conversation →
Frequently asked

Answered already

  • 01 Do you guarantee returns?

    No, no one can. Türk Otağ's guarantee is engineering: risk management, kill-switch, observability, audit trail. Strategy comes from you or is built together — outcomes are your responsibility.

  • 02 Do you provide the strategy?

    Two ways: (1) we code and harden your existing strategy, or (2) we design the architecture together and write the indicators and decision rules with you. We do not sell winning strategies; we sell architecture and code.

  • 03 Which exchanges?

    Most often: Binance, Bybit, OKX, Kraken (CEX); Hyperliquid, dYdX (perp DEX); Uniswap, Jupiter (spot DEX); MT4/MT5 (FX/CFD); custom REST/WebSocket. The list can be extended.

  • 04 Will I lose money if the bot crashes?

    A deactivation protocol is designed from the start: when the bot stops, open positions auto-close or fall to stops. The kill-switch works automatically (anomaly) and manually (Telegram command).

  • 05 Will you work on my existing bot?

    Yes. We start with a code audit (~1 week), report bottlenecks and risks. Then a separate quote for the new piece on top.

  • 06 Which languages?

    Usually Python (NumPy/Pandas/CCXT) — the quant ecosystem lives there. For high frequency, Rust or Go. For AI agents, TypeScript with LangChain/LangGraph is common.

  • 07 Legal compliance?

    TR (SPK), EU (MiCA), US (SEC) frameworks differ. Legal structuring is the client's responsibility; we support compliance requirements at the technical layer (KYC integration, audit log, data retention).

Next step

A coffee-length conversation.

A few lines on your strategy or your need; two business days to a reply, by hand. NDA first if needed.

Open a conversation
Open a conversation