Teams & Orchestration - Neotask Documentation | Neotask

Teams aur Multi-Agent Orchestration

Teams Kya Hain?

Teams Neotask ka multi-agent orchestration system hai jo complex goals ko roles aur tasks mein decompose karta hai. Jab aap team banate hain, ek AI planning phase aapke objective ka analysis karta hai aur optimal team structure automatically build karta hai. Phir ek lead agent sabhi subagents ko coordinate karta hai aur completion tak progress monitor karta hai.

Teams Enterprise plan par uplabdh hain, jo 300+ agents ek saath chalane ka support karta hai. Yeh Enterprise ko powerful banane ka core hai: platform dwara coordinated massive parallel orchestration.

Kaise Kaam Karta Hai

  • Apna objective describe karein. High-level goal provide karein (jaise "Mere startup ke liye landing page banao").
  • AI analysis. System complexity analyze karta hai, required roles identify karta hai aur task breakdown determine karta hai.
  • Agent assignment. Agents har role ke liye create aur assign hote hain (team mein 5 roles tak).
  • Task decomposition. Aapka objective discrete tasks mein break hota hai (project mein 8 tasks tak).
  • Orchestration. Ek lead orchestrator agent sabhi subagents par sab kuch coordinate karta hai.
  • Real-time updates. Har task complete hone par status updates stream hote hain.

    • Orchestration Modes

    | Mode | Description | | --- | --- | | Parallel | Sabhi agents independent tasks par ek saath kaam karte hain. Un objectives ke liye best jahan tasks ki ek doosre par koi dependency nahi hai. | | Sequential | Tasks strict order mein complete hote hain. Har task ka output agle mein feed hota hai. Un objectives ke liye best jahan har step pichle par build karta hai. | | Hybrid | Parallel aur sequential execution ka mix: tasks parallel mein fan out hote hain, phir synchronization points par results merge hote hain. Complex objectives ke liye best jismein independent aur dependent kaam ka combination ho. |

    Team Banana

    Neotask mein team banane ke liye:

  • Name aur description. Apni team ko descriptive naam aur uske purpose ka summary dein.
  • Agents select karein. Team mein kaunse agents shamil karne hain, choose karein.
  • Orchestration mode choose karein. Apni workflow needs ke aadhar par Parallel, Sequential ya Hybrid choose karein.
  • Objective define karein. Woh prompt ya objective likhein jis par team kaam karegi.
  • Execution start karein. Team launch karein aur real time mein progress monitor karein.

    • Task Types aur Model Routing

    Neotask smart task-type routing use karta hai har task ke liye automatically best model assign karne ke liye:

    | Task Type | Model | Tier | | --- | --- | --- | | Architecture | Claude Sonnet 4 | Heavy | | Complex coding | GPT-5.1 | Heavy | | Standard coding | Claude Haiku 4.5 | Medium | | Research | Gemini 3 Flash (1M context) | Medium | | Documentation | DeepSeek V3.2 | Medium | | Data extraction | Gemini 2.5 Flash Lite | Light | | Simple tasks | GPT-4.1 Nano | Light |

    Yeh routing har decomposed task ke nature ke aadhar par automatically hoti hai. Aapko manually models select karne ki zaroorat nahi hai.

    Budget-Aware Routing

    Jab budget constraints active hain, Neotask intelligently model selection downgrade karta hai aapke limits mein rehne ke liye:

  • Heavy tier tasks Medium tier models mein downgrade hote hain.
  • Medium tier tasks Light tier models mein downgrade hote hain.

    • System specific model unavailable ya over budget hone par provider fallback ordering bhi apply karta hai:

    Fallback chain: Claude Sonnet 4 → GPT-5.1 → Gemini 3 Pro

    Yeh sunishtchit karta hai ki aapki team runs hamesha complete hoti hain, tight budget constraints mein bhi, gracefully more cost-effective models par shift hokar.

    Session Isolation

    Team mein har agent apni isolated session mein operate karta hai:

  • Koi cross-contamination nahi. Agents ek doosre ke session data access ya interfere nahi kar sakte.
  • Goal context sharing. Lead agent har subagent ke saath relevant goal context share karta hai, lekin isse zyada kuch nahi.
  • Completion reporting. Har subagent se real-time completion reports lead agent ko flow karti hain, orchestration track par rakhte hue.

    • Yeh isolation sunishtchit karta hai ki agents independently aur reliably kaam karein bina same team mein anya agents ke side effects ke.

    Team Runs

    Neotask har team execution ko run ke roop mein track karta hai. Team runs view se, aap:

  • Run history dekhein. Status, duration aur results ke saath past runs dekhein.
  • Runs start aur stop karein. Naye executions launch karein ya progress mein runs rok dein.
  • Individual agent progress monitor karein. Track karein ki har agent kya kaam kar raha hai aur kitna aage hai.
  • Task cascade. Ek task complete hone par automatically koi bhi dependent tasks trigger hote hain, bina manual intervention ke workflow chalte rakhte hue.

    • Enterprise Ke Saath Scaling

    Enterprise plans un teams ke liye design kiye gaye hain jinhe complex, multi-agent workflows scale par chalane ki zaroorat hai:

  • 300+ concurrent agents. Alag teams aur objectives par parallel mein kaam karte sainkdon agents deploy karein.
  • Usage-based pricing. Aap orchestration infrastructure ke liye platform subscription pay karte hain aur usage fees sabhi agents ke actual token consumption ke aadhar par billed hoti hain. Iska matlab aap sirf us compute ke liye pay karte hain jo aapke agents actually use karte hain.
  • Per-agent budgets. Scale par bhi cost control maintain karne ke liye har agent par individual spend limits set karein.
  • Smart routing. Automatic model selection sunishtchit karta hai ki har task most cost-effective model use kare jo use handle karne mein capable ho.

    • View full documentation