1 
2AI-ahenty pereishly vid doslidnytskykh demo do vyrobnychykh system. Ochikuietsia, shcho ponad 60% korporatyvnykh AI-zastosunkiv mistytymut ahentni komponenty do 2026 roku. Ale stvorennya ahentiv z nulya  -  keruvannya tsyklamy instrumentiv, stanom, pamiattiu, obrobka pomylok ta koordynatsiia bahatokh ahentiv  -  tse skladna zadacha. Same dlya tsoho isnuiut freimvorky.
3 
4U 2026 rotsi domynuiut chotyry freimvorky: **LangGraph**, **CrewAI**, **OpenAI Agents SDK** ta **Claude Agent SDK**. Kozhen z nykh zastosoviue pryntsypovo inshyi pidkhid do tiiiei samoii problemy: nadaty LLM zdatnist mirkuvaty, planuvaty, vykorystovuvaty instrumenty ta spivpratsiuvaty.
5 
6## Korotkyi ohliad
7 
8| Aspekt | LangGraph | CrewAI | OpenAI Agents SDK | Claude Agent SDK |
9|--------|-----------|--------|-------------------|-----------------|
10| **Rozrobnyk** | LangChain | CrewAI Inc. | OpenAI | Anthropic |
11| **Arkhitektura** | Na osnovi hrafiv | Na osnovi rolei | Na osnovi peredach | Avtonomnyi tsykl |
12| **Filosofiia** | Maksymalnyi kontrol | Komandna spivpratsia | Minimalna abstraktsiia | Dai ahentu kompiuter |
13| **Movy** | Python, TypeScript | Python | Python | Python, TypeScript |
14| **Pidtrymka modelei** | Bud-yaki (OpenAI, Claude, lokalni) | Bud-yaki | Bud-yaki (nezvazhaiuchy na nazvu) | Tilky Claude |
15| **Zirky na GitHub** | ~29k | ~40k | ~21k | ~6k |
16| **Naykrashche dlya** | Skladnykh workflow zi stanom | Spetsializatsiii bahatokh ahentiv | Routynhu ta triazhu | Koduvannya ta zavdan z failamy |
17 
18## LangGraph: Budivnyk hrafiv
19 
20LangGraph modeliuie workflow ahentiv yak **oriientovani tsyklichni hrafy**. Vy vyznachaiete vuzly (funktsiii, shcho vykonuiut robotu) ta rebra (perekhody mizh nymy, za bazhannyam umovni). Stan protikaiie cherez hraf i zberihaietsia cherez checkpointing.
21 
22Tse naibilsh yavnyi ta kontrolovanyi freimvork  -  vy sami pidkliuchaiete kozhen krok.
23 
24```mermaid
25graph LR
26    Start --> Router[Router Node]
27    Router -->|needs research| Research[Research Node]
28    Router -->|needs code| Code[Code Node]
29    Research --> Synthesize[Synthesize Node]
30    Code --> Synthesize
31    Synthesize --> End
32```
33 
34### Kliuchovi kontseptsiii
35 
36- **StateGraph**: vyznachennya hrafu z typizovanym stanom
37- **Nodes**: funktsiii Python, shcho peretvoruiut stan
38- **Edges**: zviazky mizh vuzlamy, mozhut buty umovnymy
39- **Checkpointing**: vbudovana postiinist dlya tryvalo pratsiuiuchykh workflow
40 
41### Pryklad kodu
42 
43```python
44from langgraph.graph import StateGraph, MessagesState, START, END
45from langchain_openai import ChatOpenAI
46 
47llm = ChatOpenAI(model="gpt-4o")
48 
49def call_agent(state: MessagesState):
50    response = llm.invoke(state["messages"])
51    return {"messages": [response]}
52 
53def should_continue(state: MessagesState):
54    last = state["messages"][-1]
55    if last.tool_calls:
56        return "tools"
57    return END
58 
59def call_tools(state: MessagesState):
60    # Execute tool calls and return results
61    results = []
62    for tool_call in state["messages"][-1].tool_calls:
63        result = execute_tool(tool_call)
64        results.append(result)
65    return {"messages": results}
66 
67graph = StateGraph(MessagesState)
68graph.add_node("agent", call_agent)
69graph.add_node("tools", call_tools)
70graph.add_edge(START, "agent")
71graph.add_conditional_edges("agent", should_continue, {"tools": "tools", END: END})
72graph.add_edge("tools", "agent")
73 
74app = graph.compile()
75result = app.invoke({"messages": [{"role": "user", "content": "What's the weather?"}]})
76```
77 
78### Sylni storony
79 
80- Detalnyy kontrol nad kozhnym krokom ta perekhodom
81- Vbudovanyi checkpointing ta human-in-the-loop
82- Povna parnist z TypeScript
83- Pratsiuie z bud-yakym provayderom LLM
84- Naykrashche dlya skladnykh workflow z umovnym haluzhennyam ta tsyklamy
85 
86### Slabki storony
87 
88- Kruta kryva navchannya  -  potribno rozumity kontseptsiii teoriii hrafiv
89- Bahato kodu dlya prostykh vypadkiv  -  bazovyi ahent potrebuie bilshe boilerplate, nizh inshi freimvorky
90- Vidladka potokiv hrafiv mozhe buty skladnoiu bez LangSmith
91 
92### Tsiny
93 
94Open-source (MIT). LangSmith (kerovana platforma sposterezhuvanosti) maie platni plany dlya vyrobnochoho monitorynhu.
95 
96## CrewAI: Tvorets komand
97 
98CrewAI vykorystoviuie liudsku metaforu: vy zbyrayte **komandu** spetsializovanykh ahentiv, kozhen z **rolliu**, **metoiu** ta **peredistoriieiu**. Ahenty spivpratsiuiut nad **zavdannyamy** za dopomohoiu **instrumentiv**, koordynovani **protsesom** (poslidovnym, iierarkhichnym abo konsensusnym).
99 
100Uiavit tse yak naim komandy, de kozhen uchasnyk maie vyznachenu posadu ta spetsializatsiiu.
101 
102```mermaid
103graph TD
104    Crew[Crew Manager] --> R[Researcher\nRole: Find data\nTools: WebSearch]
105    Crew --> W[Writer\nRole: Write content\nTools: FileWrite]
106    Crew --> E[Editor\nRole: Review quality\nTools: FileRead]
107    R --> Task1[Research Task]
108    W --> Task2[Writing Task]
109    E --> Task3[Review Task]
110    Task1 --> Task2 --> Task3
111```
112 
113### Kliuchovi kontseptsiii
114 
115- **Agent**: persona z rolliu, metoiu, peredistoriieiu ta instrumentamy
116- **Task**: zavdannya z opysom, ochikuvanym rezultatom ta pryznachenym ahentom
117- **Crew**: hrupa ahentiv, shcho pratsiuiut razom
118- **Process**: stratehiia vykonannya (poslidovna, iierarkhichna, konsensusna)
119- **Flow**: riven orkestraciii na osnovi podii dlya ziednannya kilkokh komand
120 
121### Pryklad kodu
122 
123```python
124from crewai import Agent, Task, Crew, Process
125 
126researcher = Agent(
127    role="Senior Research Analyst",
128    goal="Find comprehensive data about the given topic",
129    backstory="You have 10 years of experience in technology research. "
130              "You are thorough and always verify facts from multiple sources.",
131    tools=[web_search_tool],
132    verbose=True,
133)
134 
135writer = Agent(
136    role="Technical Writer",
137    goal="Create clear, engaging technical content",
138    backstory="You write for a developer audience. "
139              "Your articles are practical and include code examples.",
140    tools=[file_tool],
141    verbose=True,
142)
143 
144research_task = Task(
145    description="Research the latest developments in WebAssembly in 2026. "
146                "Focus on WASI, Component Model, and production use cases.",
147    expected_output="A structured research document with key findings and sources.",
148    agent=researcher,
149)
150 
151writing_task = Task(
152    description="Write a blog post based on the research. "
153                "Include code examples and Mermaid diagrams.",
154    expected_output="A complete blog post in Markdown format.",
155    agent=writer,
156    context=[research_task],  # Writer receives researcher's output
157)
158 
159crew = Crew(
160    agents=[researcher, writer],
161    tasks=[research_task, writing_task],
162    process=Process.sequential,
163    verbose=True,
164)
165 
166result = crew.kickoff()
167print(result.raw)
168```
169 
170### Sylni storony
171 
172- Intuitivna abstraktsiia na osnovi rolei  -  lehko zrozumity
173- Ponad 100 vbudovanykh intehratsii instrumentiv
174- Spilna pamiat mizh ahentamy (korotkostrokova, dovhostrokova, sutnnostei)
175- Naibilsha spilnota (~40k zirok na GitHub)
176- Iierarkhichnyi protses z ahentom-"menedzherom", yakyi delehuie ta validuie
177 
178### Slabki storony
179 
180- Mensh detalnyy kontrol, nizh u LangGraph  -  vy vyznachaiete roli, a ne tochni shlyakhy vykonannya
181- Iierarkhichnyi protses mozhe buty neperedbachuvanym, koly ahenty ne pohodzhuiutsia
182- Vidladka rozmov bahatokh ahentiv skladnisha, nizh potokiv odnoho ahenta
183 
184### Tsiny
185 
186Open-source yadro (bezkoshtovno). CrewAI Platform: 99 $/misiats (Teams) do 120 tys. $/rik (Enterprise). Tsiny zalezhat vid aktyvnykh komand ta shchomisiachnykh vykonan.
187 
188## OpenAI Agents SDK: Router
189 
190OpenAI Agents SDK (dukhovnyi naslidnyk Swarm) foksuietsia na **peredachakh**  -  ahenty peredaiut rozmovy inshym spetsializovanym ahentam. Tse naibilsh minimalistychnyi freimvork: ahenty, instrumenty, peredachi ta guardrails. Ot i vse.
191 
192```mermaid
193graph LR
194    User --> Triage[Triage Agent]
195    Triage -->|billing question| Billing[Billing Agent]
196    Triage -->|refund request| Refund[Refund Agent]
197    Triage -->|technical issue| Support[Support Agent]
198    Billing --> Response[Response]
199    Refund --> Response
200    Support --> Response
201```
202 
203### Kliuchovi kontseptsiii
204 
205- **Agent**: model + instruktsiii + instrumenty + peredachi
206- **Handoff**: peredacha inshomu ahentu (modeliuietsia yak instrument, yakyi LLM mozhe vyklykaty)
207- **Guardrail**: validatsiia vkhodu/vykhodu, shcho zapuskaiietsia parallelno z ahentom
208- **Runner**: vykonuie tsykl ahenta
209- **Tracing**: vbudovana sposterezhuvanist dlya vsikh vyklykiv LLM, vykorystannya instrumentiv ta peredach
210 
211### Pryklad kodu
212 
213```python
214from agents import Agent, Runner, handoff, InputGuardrail, GuardrailFunctionOutput
215from pydantic import BaseModel
216 
217class SafetyCheck(BaseModel):
218    is_safe: bool
219    reason: str
220 
221async def content_safety(ctx, agent, input_text):
222    result = await Runner.run(
223        Agent(name="Safety", instructions="Check if input is safe. No PII."),
224        input_text,
225        context=ctx,
226    )
227    output = SafetyCheck.model_validate_json(result.final_output)
228    return GuardrailFunctionOutput(
229        output_info=output, tripwire_triggered=not output.is_safe
230    )
231 
232billing_agent = Agent(
233    name="Billing Agent",
234    instructions="You handle billing inquiries. Be precise with numbers.",
235    tools=[lookup_invoice, process_payment],
236)
237 
238refund_agent = Agent(
239    name="Refund Agent",
240    instructions="You process refund requests. Always verify the order first.",
241    tools=[lookup_order, issue_refund],
242)
243 
244triage_agent = Agent(
245    name="Triage Agent",
246    instructions="Route the customer to the right specialist. "
247                 "Ask clarifying questions if needed.",
248    handoffs=[billing_agent, refund_agent],
249    input_guardrails=[InputGuardrail(guardrail_function=content_safety)],
250)
251 
252result = await Runner.run(triage_agent, "I need a refund for order #4521")
253print(result.final_output)
254# The triage agent routes to refund_agent, which processes the refund
255```
256 
257### Sylni storony
258 
259- Chystyi patttern peredach  -  pryrodnyii dlya workflow routynhu/triazhu
260- Guardrails pratsiuiut parallelno z vykonannyam (fail-fast, ne blokuiuchi)
261- Vbudovana panel tracing dlya vidladky
262- Nezvazhaiuchy na nazvu, pidtrymuie modeli ne vid OpenAI
263- Minimalna abstraktsiia  -  lehko zrozumity ta rozshyryty
264 
265### Slabki storony
266 
267- Mensh zrile keruvannya stanom, nizh u LangGraph
268- Nemaie vbudovanoii postiiinosti chy checkpointynhu
269- Ekosystema instrumentiv tretiikh storin mensha
270- Dyzain, oriientovanyi na peredachi, mozhe ne pidkhodty do kozhnoi arkhitektury
271 
272### Tsiny
273 
274Open-source (MIT). Vy platyte za token za bud-yaku model, yaku vykorystovuiete.
275 
276## Claude Agent SDK: Rozrobnyk
277 
278Claude Agent SDK zastosoviuie inshyi pidkhid: zamist vyznachennya workflow chy rolei vy daiete ahentu **nabir instrumentiv i dozvolaiete yomu samomu z'yasuvaty, yak vykonaty zavdannya**. Vin vykorystoviuie toi samyi avtonomnyi tsykl, shcho napovniuie Claude Code  -  chytai, dii, pereviryai, iterui.
279 
280```mermaid
281graph TD
282    Prompt[User Prompt] --> Loop[Autonomous Agent Loop]
283    Loop --> Reason[Reason about next step]
284    Reason --> Act[Execute tool]
285    Act --> Verify[Check result]
286    Verify -->|not done| Loop
287    Verify -->|done| Output[Final output]
288```
289 
290### Kliuchovi kontseptsiii
291 
292- **query()**: holovna tochka vkhodu, shcho zapuskaie tsykl ahenta
293- **Vbudovani instrumenty**: Read, Write, Edit, Bash, Glob, Grep, WebSearch, WebFetch
294- **Korstuvanski instrumenty cherez MCP**: vyznachennya instrumentiv yak in-process MCP serveriv
295- **Sub-agents**: spetsializovani ahenty, yakym batkivskyi mozhe delehuvaty
296- **Sessions**: pidtrymannya kontekstu mizh kilkoma vzaiemodiiamy
297 
298### Pryklad kodu
299 
300```typescript
301import { tool, createSdkMcpServer, query } from "@anthropic-ai/claude-agent-sdk";
302import { z } from "zod";
303 
304const searchDocs = tool(
305  "search_docs",
306  "Search the internal documentation for relevant information",
307  { query: z.string().describe("Search query") },
308  async ({ query }) => {
309    const results = await vectorStore.similaritySearch(query, 5);
310    return {
311      content: [{ type: "text", text: results.map(r => r.pageContent).join("\n\n") }],
312    };
313  }
314);
315 
316const docsServer = createSdkMcpServer({
317  name: "docs",
318  version: "1.0.0",
319  tools: [searchDocs],
320});
321 
322for await (const message of query({
323  prompt: "Find how authentication works in our system and write a summary",
324  options: {
325    mcpServers: { docs: docsServer },
326    allowedTools: ["Read", "Glob", "Grep", "mcp__docs__search_docs"],
327  },
328})) {
329  if (message.type === "result" && message.subtype === "success") {
330    console.log(message.result);
331  }
332}
333```
334 
335### Sylni storony
336 
337- Pershokklasna intehratsiia MCP  -  pidkliuchaytesia do bud-yakoi ekosystemy MCP serveriv
338- Vbudovani instrumenty dlya failovykh operatsii, terminalu ta dostupu do merezhi
339- Avtomatychne stysnennya kontekstu dlya velykykh kodovykh baz
340- Paralelni sub-agents dlya skladnykh zavdan
341- Toi samyi rushtii, shcho y Claude Code  -  perevirrenyi na realnykh workflow rozrobky
342 
343### Slabki storony
344 
345- Tilky modeli Claude  -  nemaie pidtrymky bahatokh provaideriv
346- Novishyi freimvork z menshoiu spilnotoiu
347- Potrebuie seredovyshche Node.js navit dlya Python SDK
348- Mensh yavnyi kontrol workflow porivnyano z LangGraph
349 
350### Tsiny
351 
352Open-source. Standartni tarify za tokeny Claude API. Managed Agents (khostynhova versiia): 0,08 $ za hodynu sesiii plius vartist tokeniv.
353 
354## Koly obraty yakyi
355 
356```mermaid
357graph TD
358    Start{What's your priority?}
359    Start -->|Full control over workflow| LG[LangGraph]
360    Start -->|Multi-agent collaboration| CA[CrewAI]
361    Start -->|Routing and triage| OA[OpenAI Agents SDK]
362    Start -->|Coding and file automation| CS[Claude Agent SDK]
363 
364    LG --> LGU[Complex stateful workflows\nConditional branching\nHuman-in-the-loop]
365    CA --> CAU[Team of specialized agents\nResearch + writing pipelines\nContent generation]
366    OA --> OAU[Customer service routing\nMulti-step handoffs\nInput validation]
367    CS --> CSU[Code generation and review\nFile-heavy automation\nMCP tool ecosystem]
368```
369 
370### Oberit LangGraph, yakshcho:
371- Vam potriben tochnyy kontrol nad kozhnym krokom workflow
372- Vash vypadok vykorystannya vkliuchaie skladnu umovnu lohiku ta tsykly
373- Vy khochete vbudovanu postiinist ta checkpointy human-in-the-loop
374- Vam potribno vykorystovuvaty kilka provaideriv LLM v odnomu workflow
375 
376### Oberit CrewAI, yakshcho:
377- Vy khochete intuityvnu abstraktsiiu na osnovi rolei
378- Vashe zavdannya vkliuchaie kilkokh ahentiv z riznymy spetsializatsiiamy
379- Vam potribno, shchob ahenty spivpratsiuvaly ta peredavaly odyn odnomu kontekst
380- Vy tsiniuiete naibilshu spilnotu ta naibilshu kilkist vbudovanykh intehratsii
381 
382### Oberit OpenAI Agents SDK, yakshcho:
383- Vash osnovnyi pattern  -  marshrutyzatsiia rozmov do spetsialistiv
384- Vam potribni guardrails, shcho validuiut vkhid/vykhid parallelno
385- Vy khochete naiprostishu mozhlyvu abstraktsiiu z minimalnym boilerplate
386- Vbudovanyi tracing ta sposterezhuvanist vazhlyvi
387 
388### Oberit Claude Agent SDK, yakshcho:
389- Vashi ahenty povynni chytaty, pysaty ta vykonuvaty kod
390- Vy khochete pershokklasnu intehratsiiu z MCP serveramy
391- Vam potribni avtonomni ahenty, shcho iteruiut ta samokorehluiutsia
392- Vy vzhe vykorystovuiete Claude i khochete nayhlyibshu intehratsiiu
393 
394## Chy mozhna kombinuvaty freimvorky?
395 
396Tak. Poshyrenyi pattern  -  vykorystannya odnoho freimvorku dlya orkestraciii ta inshoho dlya okremykh ahentiv:
397 
398- **LangGraph** dlya zahalnoho hrafu workflow
399- **CrewAI** dlya konkretnoho vuzla, shcho potrebuie spivpratsi bahatokh ahentiv
400- **Claude Agent SDK** dlya pidzavdan, poviazanykh z koduvanniam, cherez MCP
401- **OpenAI Agents SDK** dlya triazhu ta routynhu na bsotsi kliienta
402 
403Freimvorky ne vykliuchaiut odyn odnoho. Vykorystovuite te, shcho pidkhodyt do kozhnoii chastyny vashoii systemy.
404 
405## Vysnovok
406 
407Kozhen freimvork robytt chitkyi vybir:
408 
409- **LangGraph** optymizuie kontrol  -  vy vyrishuiete kozhen perekhid
410- **CrewAI** optymizuie spivpratsiiu  -  ahenty pratsiuiut yak komanda
411- **OpenAI Agents SDK** optymizuie prostotu  -  minimalna abstraktsiia, chysti peredachi
412- **Claude Agent SDK** optymizuie avtonomiiu  -  daite yomu instrumenty i dozvolte pratsiuvaty
413 
414Pravylnyi vybir zalezhyt vid vashoho workflow, vashoii komandy ta vashoho isnuiuchoho steku. Oberit toi, shcho vidpovidaie vashomu osnovnomu vypadku vykorystannya, dobre vyvchit yoho ta pidkliuchit inshi, koly natrapyte na yikhniu sylnu storonu.
415