maciejpi/nordabiz
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

docs: Aktualizacja dokumentacji do Gemini 3 Flash

Browse Source 
Zmiana domyślnego modelu w dokumentacji i kodzie:
- gemini-2.5-flash → gemini-3-flash-preview
- gemini-2.5-pro → gemini-3-pro-preview

Zaktualizowane pliki:
- README.md - opis technologii
- docs/architecture/*.md - diagramy i przepływy
- nordabiz_chat.py - fallback model name
- zopk_news_service.py - model dla AI evaluation
- templates/admin/zopk_dashboard.html - wyświetlany model

Zachowano mapowania legacy modeli dla kompatybilności wstecznej.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Maciej Pienczyn 2026-01-29 14:19:05 +01:00
parent 2e67c233f7
commit 23493f0b61
11 changed files with 52 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
README.md
View File

@ -70,7 +70,7 @@ Norda Biznes Partner is a **Flask-powered web platform** built with PostgreSQL,
**Key Capabilities:** **Key Capabilities:**
- **Company Directory** - Complete catalog of 80 member companies with verified data - **Company Directory** - Complete catalog of 80 member companies with verified data
- **AI Chat Assistant** - Google Gemini 2.5 Flash-powered conversational search - **AI Chat Assistant** - Google Gemini 3 Flash-powered conversational search
- **Advanced Search** - Multi-mode search with FTS, fuzzy matching, and synonym expansion - **Advanced Search** - Multi-mode search with FTS, fuzzy matching, and synonym expansion
- **Admin Panels** - News moderation, SEO audit, social media tracking, GBP/IT audits - **Admin Panels** - News moderation, SEO audit, social media tracking, GBP/IT audits
- **User Authentication** - Secure login with email confirmation and role-based access - **User Authentication** - Secure login with email confirmation and role-based access
@ -92,7 +92,7 @@ Norda Biznes Partner is a **Flask-powered web platform** built with PostgreSQL,
### AI & Machine Learning ### AI & Machine Learning
- **Google Gemini AI** (google-generativeai 0.3.2) - **Google Gemini AI** (google-generativeai 0.3.2)
- Models: gemini-2.5-flash (default), gemini-2.5-flash-lite, gemini-2.5-pro - Models: gemini-3-flash-preview (default), gemini-3-pro-preview (advanced)
- Features: Multi-turn conversations, context-aware recommendations, AI-powered search - Features: Multi-turn conversations, context-aware recommendations, AI-powered search
- Limits: Free tier (200 requests/day) - Limits: Free tier (200 requests/day)
@ -160,7 +160,7 @@ Norda Biznes Partner is a **Flask-powered web platform** built with PostgreSQL,
- **Weighted scoring:** Prioritized results by relevance - **Weighted scoring:** Prioritized results by relevance
#### 💬 AI Chat Assistant #### 💬 AI Chat Assistant
- **Conversational AI** powered by Google Gemini 2.5 Flash - **Conversational AI** powered by Google Gemini 3 Flash
- **Context-aware recommendations** for company discovery - **Context-aware recommendations** for company discovery
- **Multi-turn conversations** with full history tracking - **Multi-turn conversations** with full history tracking
- **Answer questions** about member companies and services - **Answer questions** about member companies and services

2
docs/architecture/01-system-context.md
View File

@ -139,7 +139,7 @@ graph TB
### 🤖 Google Gemini AI API ### 🤖 Google Gemini AI API
**Provider:** Google AI Studio **Provider:** Google AI Studio
**Purpose:** Generative AI for chat and content analysis **Purpose:** Generative AI for chat and content analysis
**Model:** gemini-2.5-flash (primary), gemini-2.5-pro (advanced) **Model:** gemini-3-flash-preview (primary), gemini-3-pro-preview (advanced)
**Integration Points:** **Integration Points:**
- AI chat responses with company context - AI chat responses with company context
- Image analysis for logo/photo descriptions - Image analysis for logo/photo descriptions

6
docs/architecture/02-container-diagram.md
View File

@ -57,7 +57,7 @@ graph TB
%% External systems %% External systems
subgraph "External APIs & Services" subgraph "External APIs & Services"
Gemini["🤖 Google Gemini API<br/>gemini-2.5-flash<br/>REST API (HTTPS)"] Gemini["🤖 Google Gemini API<br/>gemini-3-flash-preview<br/>REST API (HTTPS)"]
BraveAPI["🔍 Brave Search API<br/>News & Social Discovery<br/>REST API (HTTPS)"] BraveAPI["🔍 Brave Search API<br/>News & Social Discovery<br/>REST API (HTTPS)"]
PageSpeed["📊 Google PageSpeed API<br/>SEO & Performance<br/>REST API (HTTPS)"] PageSpeed["📊 Google PageSpeed API<br/>SEO & Performance<br/>REST API (HTTPS)"]
Places["📍 Google Places API<br/>Business Profiles<br/>REST API (HTTPS)"] Places["📍 Google Places API<br/>Business Profiles<br/>REST API (HTTPS)"]
@ -376,7 +376,7 @@ results = search_companies(db, "strony www", limit=10)
### 🤖 Gemini Service (`gemini_service.py`) ### 🤖 Gemini Service (`gemini_service.py`)
**Purpose:** Interface to Google Gemini AI API **Purpose:** Interface to Google Gemini AI API
**Models:** `gemini-2.5-flash` (default), `gemini-2.5-pro` (advanced) **Models:** `gemini-3-flash-preview` (default), `gemini-3-pro-preview` (advanced)
**Authentication:** API key from `.env` **Authentication:** API key from `.env`
**Capabilities:** **Capabilities:**
@ -387,7 +387,7 @@ results = search_companies(db, "strony www", limit=10)
**API Wrapper Functions:** **API Wrapper Functions:**
```python ```python
generate_text(prompt, model="gemini-2.5-flash") generate_text(prompt, model="gemini-3-flash-preview")
generate_chat_response(messages, context, stream=True) generate_chat_response(messages, context, stream=True)
analyze_image(image_bytes, prompt) analyze_image(image_bytes, prompt)
score_content_relevance(content, company_name) score_content_relevance(content, company_name)

9
docs/architecture/04-flask-components.md
View File

@ -559,15 +559,14 @@ track_usage(session, input_tokens, output_tokens)
#### 🤖 GeminiService (`gemini_service.py` - ~500 lines) #### 🤖 GeminiService (`gemini_service.py` - ~500 lines)
**Purpose:** Google Gemini AI API integration **Purpose:** Google Gemini AI API integration
**Primary Functions:** **Primary Functions:**
- `generate_text(prompt, model="gemini-2.5-flash")` - `generate_text(prompt, model="gemini-3-flash-preview")`
- `generate_streaming(prompt, model="gemini-2.5-flash")` - `generate_streaming(prompt, model="gemini-3-flash-preview")`
- `analyze_image(image_path, prompt)` - `analyze_image(image_path, prompt)`
- `moderate_content(text)` - `moderate_content(text)`
**Supported Models:** **Supported Models:**
- **gemini-2.5-flash** - Fast, cheap ($0.075/1M tokens) - **gemini-3-flash-preview** - Default, 7x better reasoning (free tier)
- **gemini-2.5-pro** - High quality ($1.25/1M tokens) - **gemini-3-pro-preview** - Advanced reasoning, 2M context ($2.00/1M tokens)
- **gemini-3-flash-preview** - Experimental (free)
**Key Features:** **Key Features:**
1. **Text Generation** - Chat, content creation, analysis 1. **Text Generation** - Chat, content creation, analysis

33
docs/architecture/06-external-integrations.md
View File

@ -51,7 +51,7 @@ graph TB
%% External integrations %% External integrations
subgraph "AI & ML Services" subgraph "AI & ML Services"
Gemini["🤖 Google Gemini API<br/>gemini-2.5-flash<br/><br/>Free tier: 200 req/day<br/>Auth: API Key<br/>Cost: $0.075-$5.00/1M tokens"] Gemini["🤖 Google Gemini API<br/>gemini-3-flash-preview<br/><br/>Free tier: unlimited<br/>Auth: API Key<br/>Cost: Free (preview)"]
end end
subgraph "SEO & Analytics" subgraph "SEO & Analytics"
@ -149,17 +149,19 @@ graph TB
| **Endpoint** | https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent | | **Endpoint** | https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent |
| **Authentication** | API Key | | **Authentication** | API Key |
| **Environment Variable** | `GOOGLE_GEMINI_API_KEY` | | **Environment Variable** | `GOOGLE_GEMINI_API_KEY` |
| **Default Model** | gemini-2.5-flash | | **Default Model** | gemini-3-flash-preview |
| **Timeout** | None (default) | | **Timeout** | None (default) |
#### Available Models #### Available Models
```python ```python
GEMINI_MODELS = { GEMINI_MODELS = {
'flash': 'gemini-2.5-flash', # Best for general use '3-flash': 'gemini-3-flash-preview', # Default - 7x better reasoning, thinking mode
'flash-lite': 'gemini-2.5-flash-lite', # Ultra cheap '3-pro': 'gemini-3-pro-preview', # Advanced - best reasoning, 2M context
'pro': 'gemini-2.5-pro', # High quality 'flash': 'gemini-2.5-flash', # Legacy - balanced cost/quality
'flash-2.0': 'gemini-2.0-flash', # 1M context window 'flash-lite': 'gemini-2.5-flash-lite', # Legacy - ultra cheap
'pro': 'gemini-2.5-pro', # Legacy - high quality
'flash-2.0': 'gemini-2.0-flash', # Legacy - 1M context (wycofywany 31.03.2026)
} }
``` ```
@ -167,14 +169,16 @@ GEMINI_MODELS = {
| Model | Input Cost | Output Cost | | Model | Input Cost | Output Cost |
|-------|-----------|-------------| |-------|-----------|-------------|
| gemini-2.5-flash | $0.075 | $0.30 | | gemini-3-flash-preview | Free | Free |
| gemini-3-pro-preview | $2.00 | $12.00 |
| gemini-2.5-flash | $0.30 | $2.50 |
| gemini-2.5-flash-lite | $0.10 | $0.40 | | gemini-2.5-flash-lite | $0.10 | $0.40 |
| gemini-2.5-pro | $1.25 | $5.00 | | gemini-2.5-pro | $1.25 | $10.00 |
| gemini-2.0-flash | $0.075 | $0.30 | | gemini-2.0-flash | $0.10 | $0.40 |
#### Rate Limits #### Rate Limits
- **Free Tier:** 200 requests/day, 50 requests/hour - **Free Tier (Gemini 3 Flash Preview):** Unlimited requests
- **Token Limits:** Model-dependent (1M for flash-2.0) - **Token Limits:** Model-dependent (1M for flash-2.0)
#### Integration Points #### Integration Points
@ -914,10 +918,9 @@ All APIs use HTTPS:
- **News Monitoring:** Schedule daily/weekly runs - **News Monitoring:** Schedule daily/weekly runs
### 3. Model Selection ### 3. Model Selection
- **Gemini:** Use cheaper models where appropriate - **Gemini:** Use appropriate models for task complexity
- `gemini-2.5-flash-lite` for simple tasks - `gemini-3-flash-preview` for general use (default, free)
- `gemini-2.5-flash` for general use - `gemini-3-pro-preview` for complex reasoning (paid)
- `gemini-2.5-pro` only for complex reasoning
### 4. Result Reuse ### 4. Result Reuse
- Don't re-analyze unchanged content - Don't re-analyze unchanged content
@ -938,7 +941,7 @@ All APIs use HTTPS:
**Test External API Connectivity:** **Test External API Connectivity:**
```bash ```bash
# Gemini API # Gemini API
curl -X POST "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent?key=${GOOGLE_GEMINI_API_KEY}" \ curl -X POST "https://generativelanguage.googleapis.com/v1beta/models/gemini-3-flash-preview:generateContent?key=${GOOGLE_GEMINI_API_KEY}" \
-H 'Content-Type: application/json' \ -H 'Content-Type: application/json' \
-d '{"contents":[{"parts":[{"text":"Hello"}]}]}' -d '{"contents":[{"parts":[{"text":"Hello"}]}]}'

18
docs/architecture/10-api-endpoints.md
View File

@ -1309,7 +1309,7 @@ POST /api/chat/<conversation_id>/message
``` ```
**AI Chat Configuration:** **AI Chat Configuration:**
- Model: gemini-2.5-flash - Model: gemini-3-flash-preview
- Max companies in context: 8 - Max companies in context: 8
- History: Last 10 messages - History: Last 10 messages
- Average latency: 250-350ms - Average latency: 250-350ms
@ -1694,21 +1694,21 @@ GET /api/model-info
"models": [ "models": [
{ {
"id": "flash", "id": "flash",
"name": "gemini-2.5-flash", "name": "gemini-3-flash-preview",
"description": "Fast, cost-effective model for most tasks", "description": "Default model - 7x better reasoning, thinking mode",
"pricing": { "pricing": {
"input_per_1m_tokens": 0.075, "input_per_1m_tokens": 0.00,
"output_per_1m_tokens": 0.30 "output_per_1m_tokens": 0.00
}, },
"default": true "default": true
}, },
{ {
"id": "pro", "id": "pro",
"name": "gemini-2.5-pro", "name": "gemini-3-pro-preview",
"description": "High-quality model for complex tasks", "description": "Advanced model for complex reasoning",
"pricing": { "pricing": {
"input_per_1m_tokens": 1.25, "input_per_1m_tokens": 2.00,
"output_per_1m_tokens": 5.00 "output_per_1m_tokens": 12.00
}, },
"default": false "default": false
} }

2
docs/architecture/flows/02-search-flow.md
View File

@ -575,7 +575,7 @@ sequenceDiagram
Note over ChatSvc: Compress company data<br/>(name, desc, services, competencies, etc) Note over ChatSvc: Compress company data<br/>(name, desc, services, competencies, etc)
ChatSvc->>Gemini: POST /generateContent<br/>System prompt + context + user message ChatSvc->>Gemini: POST /generateContent<br/>System prompt + context + user message
Note over Gemini: Model: gemini-2.5-flash<br/>Max tokens: 2048 Note over Gemini: Model: gemini-3-flash-preview<br/>Max tokens: 2048
Gemini->>ChatSvc: AI response text Gemini->>ChatSvc: AI response text

20
docs/architecture/flows/03-ai-chat-flow.md
View File

@ -19,7 +19,7 @@ This document describes the **complete AI chat flow** for the Norda Biznes Partn
- **Search Integration** for company discovery - **Search Integration** for company discovery
**Key Technology:** **Key Technology:**
- **AI Model:** Google Gemini 2.5 Flash (gemini-2.5-flash) - **AI Model:** Google Gemini 3 Flash (gemini-3-flash-preview)
- **Chat Engine:** NordaBizChatEngine (nordabiz_chat.py) - **Chat Engine:** NordaBizChatEngine (nordabiz_chat.py)
- **Gemini Service:** Centralized GeminiService (gemini_service.py) - **Gemini Service:** Centralized GeminiService (gemini_service.py)
- **Search Integration:** Unified SearchService (search_service.py) - **Search Integration:** Unified SearchService (search_service.py)
@ -34,7 +34,7 @@ This document describes the **complete AI chat flow** for the Norda Biznes Partn
- Compact data format to minimize token usage - Compact data format to minimize token usage
**Cost & Performance:** **Cost & Performance:**
- **Model:** Gemini 2.5 Flash - **Model:** Gemini 3 Flash
- **Pricing:** $0.075/$0.30 per 1M tokens (input/output) - **Pricing:** $0.075/$0.30 per 1M tokens (input/output)
- **Free Tier:** 1,500 requests/day, unlimited tokens - **Free Tier:** 1,500 requests/day, unlimited tokens
- **Typical Response:** 200-400ms latency, 5,000-15,000 tokens - **Typical Response:** 200-400ms latency, 5,000-15,000 tokens
@ -79,7 +79,7 @@ flowchart TD
ContextBuilder -->|19. Compact format| Context[Full Context<br/>JSON] ContextBuilder -->|19. Compact format| Context[Full Context<br/>JSON]
Context -->|20. Query AI| GeminiService[Gemini Service<br/>gemini_service.py] Context -->|20. Query AI| GeminiService[Gemini Service<br/>gemini_service.py]
GeminiService -->|21. API call| GeminiAPI[Google Gemini API<br/>gemini-2.5-flash] GeminiService -->|21. API call| GeminiAPI[Google Gemini API<br/>gemini-3-flash-preview]
GeminiAPI -->|22. AI response| GeminiService GeminiAPI -->|22. AI response| GeminiService
GeminiService -->|23. Track cost| CostDB[(ai_api_costs)] GeminiService -->|23. Track cost| CostDB[(ai_api_costs)]
@ -144,7 +144,7 @@ INSERT INTO ai_chat_conversations (
is_active, message_count, model_name, created_at is_active, message_count, model_name, created_at
) VALUES ( ) VALUES (
?, NOW(), 'general', ?, ?, NOW(), 'general', ?,
TRUE, 0, 'gemini-2.5-flash', NOW() TRUE, 0, 'gemini-3-flash-preview', NOW()
); );
``` ```
@ -208,7 +208,7 @@ sequenceDiagram
Note over Engine: 3. Query AI with full context Note over Engine: 3. Query AI with full context
Engine->>Gemini: generate_text(<br/> prompt=system_prompt + context + history,<br/> feature='ai_chat',<br/> user_id=current_user.id,<br/> temperature=0.7<br/>) Engine->>Gemini: generate_text(<br/> prompt=system_prompt + context + history,<br/> feature='ai_chat',<br/> user_id=current_user.id,<br/> temperature=0.7<br/>)
Gemini->>API: POST /v1/models/gemini-2.5-flash:generateContent Gemini->>API: POST /v1/models/gemini-3-flash-preview:generateContent
API->>Gemini: AI response text API->>Gemini: AI response text
Note over Gemini: Track API cost to database Note over Gemini: Track API cost to database
@ -277,7 +277,7 @@ WHERE id = ?;
"message_id": 456, "message_id": 456,
"created_at": "2026-01-10T10:35:22.123456", "created_at": "2026-01-10T10:35:22.123456",
"tech_info": { "tech_info": {
"model": "gemini-2.5-flash", "model": "gemini-3-flash-preview",
"data_source": "PostgreSQL (80 firm Norda Biznes)", "data_source": "PostgreSQL (80 firm Norda Biznes)",
"architecture": "Full DB Context (wszystkie firmy w kontekście AI)", "architecture": "Full DB Context (wszystkie firmy w kontekście AI)",
"tokens_input": 8543, "tokens_input": 8543,
@ -408,7 +408,7 @@ flowchart TD
GeminiSvc --> AutoCost[Automatic cost tracking<br/>to ai_api_costs table] GeminiSvc --> AutoCost[Automatic cost tracking<br/>to ai_api_costs table]
DirectAPI --> NoCost[No cost tracking] DirectAPI --> NoCost[No cost tracking]
AutoCost --> APICall[Gemini API Call<br/>gemini-2.5-flash] AutoCost --> APICall[Gemini API Call<br/>gemini-3-flash-preview]
NoCost --> APICall NoCost --> APICall
APICall --> Response[AI Response<br/>~200-400 tokens] APICall --> Response[AI Response<br/>~200-400 tokens]
@ -524,7 +524,7 @@ sequenceDiagram
Gemini->>Gemini: total_cost = input + output Gemini->>Gemini: total_cost = input + output
Note over Gemini: Global cost tracking Note over Gemini: Global cost tracking
Gemini->>GlobalDB: INSERT INTO ai_api_costs<br/>(api_provider='gemini',<br/> model='gemini-2.5-flash',<br/> feature='ai_chat',<br/> user_id=123,<br/> tokens, cost, latency) Gemini->>GlobalDB: INSERT INTO ai_api_costs<br/>(api_provider='gemini',<br/> model='gemini-3-flash-preview',<br/> feature='ai_chat',<br/> user_id=123,<br/> tokens, cost, latency)
Gemini->>Engine: Return response text Gemini->>Engine: Return response text
@ -538,7 +538,7 @@ sequenceDiagram
### 6.3 Cost Calculation ### 6.3 Cost Calculation
**Gemini 2.5 Flash Pricing:** **Gemini 3 Flash Pricing:**
- **Input:** $0.075 per 1M tokens - **Input:** $0.075 per 1M tokens
- **Output:** $0.30 per 1M tokens - **Output:** $0.30 per 1M tokens
- **Free Tier:** 1,500 requests/day (unlimited tokens) - **Free Tier:** 1,500 requests/day (unlimited tokens)
@ -720,7 +720,7 @@ CREATE TABLE ai_api_costs (
id SERIAL PRIMARY KEY, id SERIAL PRIMARY KEY,
timestamp TIMESTAMP NOT NULL DEFAULT NOW(), timestamp TIMESTAMP NOT NULL DEFAULT NOW(),
api_provider VARCHAR(50) NOT NULL, -- 'gemini' api_provider VARCHAR(50) NOT NULL, -- 'gemini'
model_name VARCHAR(100), -- 'gemini-2.5-flash' model_name VARCHAR(100), -- 'gemini-3-flash-preview'
feature VARCHAR(100), -- 'ai_chat', 'image_analysis', etc. feature VARCHAR(100), -- 'ai_chat', 'image_analysis', etc.
user_id INTEGER REFERENCES users(id), user_id INTEGER REFERENCES users(id),
input_tokens INTEGER, input_tokens INTEGER,

2
nordabiz_chat.py
View File

@ -125,7 +125,7 @@ class NordaBizChatEngine:
raise ValueError("GOOGLE_GEMINI_API_KEY not found in environment") raise ValueError("GOOGLE_GEMINI_API_KEY not found in environment")
genai.configure(api_key=api_key) genai.configure(api_key=api_key)
self.model_name = "gemini-2.5-flash" self.model_name = "gemini-3-flash-preview"
self.model = genai.GenerativeModel(self.model_name) self.model = genai.GenerativeModel(self.model_name)
self.tokenizer = self.model self.tokenizer = self.model
self.gemini_service = None self.gemini_service = None

2
templates/admin/zopk_dashboard.html
View File

@ -1030,7 +1030,7 @@
<div class="ai-model-info" style="background: linear-gradient(135deg, #e0f2fe 0%, #bae6fd 100%); border: 1px solid #7dd3fc; border-radius: var(--radius); padding: var(--spacing-sm) var(--spacing-md); margin-bottom: var(--spacing-md); font-size: var(--font-size-sm);"> <div class="ai-model-info" style="background: linear-gradient(135deg, #e0f2fe 0%, #bae6fd 100%); border: 1px solid #7dd3fc; border-radius: var(--radius); padding: var(--spacing-sm) var(--spacing-md); margin-bottom: var(--spacing-md); font-size: var(--font-size-sm);">
<div style="display: flex; justify-content: space-between; align-items: center; flex-wrap: wrap; gap: var(--spacing-xs);"> <div style="display: flex; justify-content: space-between; align-items: center; flex-wrap: wrap; gap: var(--spacing-xs);">
<div> <div>
<strong>🤖 Model:</strong> <code style="background: #fff; padding: 2px 6px; border-radius: 4px;">gemini-2.5-flash-lite</code> <strong>🤖 Model:</strong> <code style="background: #fff; padding: 2px 6px; border-radius: 4px;">gemini-3-flash-preview</code>
</div> </div>
<div> <div>
<strong>📅 Prompt v2:</strong> 2026-01-15 <strong>📅 Prompt v2:</strong> 2026-01-15

2
zopk_news_service.py
View File

@ -1523,7 +1523,7 @@ def reevaluate_low_score_news(db_session, limit: int = 50, user_id: int = None)
news.ai_relevance_score = new_score news.ai_relevance_score = new_score
news.ai_evaluation_reason = result['reason'] news.ai_evaluation_reason = result['reason']
news.ai_evaluated_at = datetime.now() news.ai_evaluated_at = datetime.now()
news.ai_model = 'gemini-2.5-flash-lite' # Updated model name news.ai_model = 'gemini-3-flash-preview' # Gemini 3 Flash
# Track change # Track change
stats['total_evaluated'] += 1 stats['total_evaluated'] += 1