Last updated:
11.03.2026
Public preview
Browse topics throughout the guide
Choose your topics of interest and jump to the corresponding guide section.
CEO /
Executive resources, strategy, ROI, governance and risks.
Organization
How to turn senior agent know-how into institutional property.
See in 1.5
Business
The difference between management software and growth engines.
See in 1.2
Liability
How AI discounts can be 'guardrail-ed' into respecting negotiated pricing tiers.
See in 1.5
KPI
What we measure beyond sales: AOV, CTR, latency.
See in 2.2
Governance
Who needs to be in the project's "control room".
See in 5.2.1.1
Architecture
How AI learns to "match" products and customers.
See in 2.3.1
Algorithms
The difference between finding 500 candidates and choosing the top 5.
See in 4.1.4
AI Generations
On top of Deep Learning of Gen 3 one can add LLM assistants (chatbots).
See in 2.1.5
Data Flow
Why XML/CSV are history and we need API Patching.
See in 2.3.3
xAI
Giving sales arguments alongside the AI recommendation.
See in 2.3.4
Copilot
The side panel suggesting "Next Best Action".
See in 3.3.1
Substitutes
Saving the sale when there's no stock.
See in 3.3.2
Bundling
The automatic package generator for clearing old stock.
See in 3.3.3
Cross-sell
Identifying complementary accessories for safe operation.
See in 3.3.4
Forecast
How to predict the day your client runs out of consumables.
See in 3.3.5
Personalization
Personalizing the B2B portal for every client profile.
See in 3.3.6
Retention
Beyond the standard email, the contextual offer.
See in 3.3.7
Compatibility
Preventing returns via "Socket Match" type rules.
See in 3.3.8
Cold Start
How we sell new products from the first day of listing.
See in 3.3.9
Pricing
Optimizing conversion probability at the best price.
See in 3.3.10
Planning
Defining objectives before the first line of code.
See in 4.1.1
Python
Snippet for filtering "forbidden" products.
See in 4.2.1
Python/YAML
Choosing the model by scenario and configuration.
See in 4.2.3
PySpark
How we calculate dynamic bundles on large data volumes.
See in 4.2.4
GCP
Configuring the Google managed service (old Retail).
See in 4.3
Configs
Declarative control of Boost and Bury strategies.
See in 4.3.4
Data
The foundation of trust.
Data Contract
How to avoid GIGO (Garbage In, Garbage Out) phenomenon.
See in 2.5.2
NIS2 / GDPR
AI function isolation, on-prem or cloud availability.
See in 2.4.4
Company truth
How to diminish AI hallucinations using your data.
See in 5.3.1.2
Methodology
Keeping the company's source of truth unmodified.
See in 5.3.2.2
Industry
How to increase margin on accessories.
See in 3.2.1
Industry
Managing batches and critical compatibilities.
See in 3.2.2
Industry
Seasonal variation and variation by crop/plot.
See in 3.2.4
BOM
Automated generation of thousand-line offers (bill of materials).
See in 6.4
Chapter 1
Why AI?
The problem: "The ERP is not a sales engine" . `Brownfield` context and the retirement cliff crisis.
Chapter 2
How does it work?
Recommendation generations and deep learning. Goals and KPIs.
Chapter 3
What can be built?
Deliverables built for B2B sales.
Chapter 4
Cookbook (Code)
Project implementation steps. Python and SQL snippets, configurations for Google Vertex AI Search for commerce.
Chapter 5
When and with what resources?
Team structure, required resources, budget, and Google Cloud assurances. Plus: our methodology.
AI News
Future: Agentic AI
UCP launch in Jan 2026, new AI technologies maturing, and how companies can adapt.
Resources
Resources & Glossary
Glossary (AI, business, software) + bibliography, whitepapers, and useful links from the guide.
Sign up for the complete PDF manual
![[UPDATE] The Romanian startup awards will be presented on March 12. Mr. Benny won 3rd place at the Romania Startup Awards 2026.](/images/new-post/small_long-romania-startup-awards-2026-mrbenny.jpg "[UPDATE] The Romanian startup awards will be presented on March 12. Mr. Benny won 3rd place at the Romania Startup Awards 2026.")
