@@ -91,6 +91,45 @@ ACE uses three AI "roles" that work together to help your agent improve:
9191
9292Think of it like a sports team reviewing game footage to get better!
9393
94+ ### The ACE Learning Loop
95+
96+ ``` mermaid
97+ flowchart TD
98+ Start([New Task/Question]) --> Generator
99+
100+ subgraph Playbook ["π Playbook (Evolving Context)"]
101+ Bullets["π Strategy Bullets<br/>β’ Helpful strategies β<br/>β’ Harmful patterns β<br/>β’ Neutral observations β"]
102+ end
103+
104+ Generator["π― Generator<br/>Uses playbook strategies<br/>to produce answer"] --> Output[Answer Output]
105+
106+ Playbook -.->|Provides Context| Generator
107+
108+ Output --> Environment["π Task Environment<br/>Evaluates answer<br/>Provides feedback"]
109+
110+ Environment --> Reflector["π Reflector<br/>Analyzes outcome<br/>Tags bullet contributions:<br/>β’ Which helped?<br/>β’ Which hurt?<br/>β’ What's missing?"]
111+
112+ Reflector --> Curator["π Curator<br/>Emits delta operations"]
113+
114+ Curator --> DeltaOps{{"π Delta Operations<br/>ADD new strategies<br/>UPDATE existing ones<br/>TAG helpful/harmful<br/>REMOVE outdated"}}
115+
116+ DeltaOps -->|Incremental<br/>Updates| Playbook
117+
118+ Environment -->|Ground Truth +<br/>Feedback| Reflector
119+
120+ style Playbook fill:#e1f5fe
121+ style Generator fill:#fff3e0
122+ style Reflector fill:#f3e5f5
123+ style Curator fill:#e8f5e9
124+ style DeltaOps fill:#fff9c4
125+ ```
126+
127+ ** Key Insights:**
128+ - ** Incremental Learning** : The playbook evolves through small delta updates, not complete rewrites
129+ - ** No Context Collapse** : Strategies are preserved and refined, preventing loss of valuable knowledge
130+ - ** Self-Improving** : Each task makes the agent smarter by updating its strategy playbook
131+ - ** Three-Role Architecture** : Separation of concerns - generating, analyzing, and updating are distinct phases
132+
94133## Examples
95134
96135### Simple Q&A Agent
0 commit comments