GLOSSARY
Fine-tuning
Fine-tuning adapts a model using examples. Sales applications need good data, independent tests and comparison with simpler approaches such as templates or RAG.
Explore the termTHE VANELLENCE GLOSSARY
Understand the terms.
See the connections.
AI, sales and negotiation clearly explained. A starting point for better questions, informed decisions and your next practical application.
GLOSSARY
Frontier model
A frontier model leads under a stated assessment. Business value depends on added performance for the actual sales task, including cost and review effort.
Explore the termGLOSSARY
Generative AI
Generative AI can draft sales responses and alternatives. Value depends on factual accuracy, negotiating room and the time required to reach a usable result.
Explore the termGLOSSARY
GPU / graphics processing unit
A GPU accelerates suitable AI computations. For self-hosting, assess memory, utilisation and measured performance against the workload you need to support.
Explore the termGLOSSARY
Grounding
Grounding connects AI answers to sources. Sales research needs support for each specific claim, with visible distinctions between evidence and interpretation.
Explore the termGLOSSARY
Guardrails
Guardrails limit specific AI risks. Assess enforceable permissions, approval steps and tests covering attempted bypasses as well as unnecessary blocking.
Explore the termGLOSSARY
Hallucination
A hallucination can sound plausible while being unsupported. In sales, consequential claims about customers, figures and commitments need traceable evidence.
Explore the termGLOSSARY
Harness
A harness controls AI workflows, tools and working state. Longer sales tasks need reliable continuation, clear permissions and recovery that can be tested.
Explore the termGLOSSARY
Human-in-the-loop
Human-in-the-loop needs informed people who can act. Sales approvals require relevant information, clear accountability and power to change decisions.
Explore the termGLOSSARY
Hyperscaler
Hyperscalers combine large cloud capacity with many services. Procurement should assess actual needs, operational capability and future switching options.
Explore the termGLOSSARY
Inference
Inference runs a trained model. Sales applications should compare cost and waiting time per usable task, including repeated calls and human review.
Explore the termGLOSSARY
Intellectual property
Intellectual property affects AI inputs and reuse of outputs. Check provenance, permitted use, confidentiality and provider terms as separate questions.
Explore the termGLOSSARY
Knowledge graph
A knowledge graph links companies, products and relationships. Useful sales research needs correct identities, supported connections and maintained data.
Explore the termGLOSSARY
Kubernetes
Kubernetes coordinates containerised applications. Its value depends on operating requirements, team capability and recovery tested under realistic conditions.
Explore the termGLOSSARY
Latency
Latency is the delay until a defined response. Sales applications need meaningful measurement points, visibility of slow cases and time to a usable answer.
Explore the termGLOSSARY
Layer
A layer separates processing and responsibility. Good boundaries support replacement and control when the interfaces preserve that separation in practice.
Explore the termGLOSSARY
LLM / large language model
An LLM processes and generates language. Sales applications need suitable context, reliable checking and evidence of value across the complete workflow.
Explore the termGLOSSARY
Machine learning
Machine learning in sales: how models learn from examples, where lead scoring can reproduce bias, and what evidence is needed to establish business value.
Explore the termGLOSSARY
MCP
MCP connects AI applications to tools and data. Understand its role in sales workflows, the integration value, and the permissions and checks still required.
Explore the termGLOSSARY
Memory
AI memory in customer workflows: useful continuity, the risk of stale assumptions, and the access, correction and retention rules that make reuse accountable.
Explore the termGLOSSARY
Model agnostic
Model independence creates options for AI systems. Test task quality, data portability and switching effort before relying on a second model as a fallback.
Explore the termGLOSSARY
Model router
A model router assigns work to AI models. Useful sales routing needs suitable selection rules, permitted data paths and evidence about quality and total cost.
Explore the termGLOSSARY
Multimodal
Multimodal AI combines text, images or audio. Sales applications need traceable source references, reliable attribution and checks on consequential details.
Explore the termGLOSSARY
Neuromorphic AI
Neuromorphic AI uses computing principles inspired by nervous systems. Its business value depends on the task, energy needs and the effort of real integration.
Explore the term