A broad technical catalog and customers who need precise answers.
A1 Security Supply is a North American company with more than 35 years of experience in the wholesale distribution of low-voltage equipment and components for contractors, installers, and businesses.
Its offering brings together video surveillance, access control, networking, communications, and fire-safety solutions from multiple manufacturers. The commercial process combines direct purchasing, special quote requests, inquiries, and support from account specialists.
In this environment, preparing a commercial proposal involves more than locating a product reference. The team must understand the project, identify missing information, consult authorized sources, and review whether products are appropriate for the intended context. A1’s primary asset is its decision-making ability: accumulated knowledge, technical judgment, and professional experience for projects that do not allow generic answers.
Information arrived with very different levels of detail.
Some requests included specific manufacturers and product references. Others described a general need or required clarification before a response could be prepared.
The specialist had to rebuild the context by consulting the catalog, product pages, manufacturer documentation, and the team’s accumulated knowledge. When a project could be subject to procurement or security requirements, the eligibility of proposed references also had to be verified.
The company had the information, but it did not reach the specialist in a shared structure that made it possible to review the request, its sources, and open issues quickly.
This way of working could produce inconsistent responses, late questions, and excessive dependence on individual memory.
Useful tools, but no shared commercial file.
A1 already had the assets needed to handle requests:
- Digital catalog and product pages.
- Quote and contact forms.
- Manufacturer and product-reference documentation.
- Specialists with technical and commercial knowledge.
- Information for professional customers and authorized resellers.
The challenge was to connect these elements through an understandable, reviewable way of working without replacing existing tools or delegating sensitive decisions to an automated system.
Define what should be automated today and what should remain under A1’s judgment.
E2M analyzed the path of a request from intake through proposal validation. The diagnostic identified which information was required, where it resided, and which decisions required human accountability.
The decision in this phase was to implement a bounded workflow that prepares the commercial work. A1 did not need to delegate the entire quote: its specialists already handled it effectively through their experience and business knowledge. AI would provide an organized file with context, sources, and open questions without displacing their responsibility.
AI organizes, retrieves, and prepares. The specialist checks compatibility, availability, terms, price, delivery, and regulatory suitability before approving any response.
An assisted workflow that turns each request into a reviewable proposal.
Interpret the request
Organize information by customer, manufacturer, product reference, product family, intended use, and stated need.
Detect missing information
Identify missing data and prepare specific questions to complete the context before quoting.
Retrieve approved knowledge
Consult the catalog, documentation, and authorized sources while keeping the provenance of the information visible.
Review special requirements
Flag projects that may require additional regulatory, government-procurement, or security checks.
Prepare the draft
Generate a response structure with references, items to confirm, and a recommended next action.
Validate before responding
The specialist retains final approval and documents exceptions, corrections, and decisions.
A1 could evolve toward an agentic workflow, but it does not need to delegate its judgment today.
An agentic AI architecture connected through limited permissions to the catalog, inventory, pricing, CRM, manufacturer documentation, and approved rules could coordinate more stages: interpret the request, consult systems, detect and request missing data, compare alternatives, gather evidence, prepare the quote, record activity, and route each approval.
That would not make the agent a technical or legal authority. Compatibility, price, margin, regulatorand theigibility, delivery, and commercial commitments would still require verifiable rules, traceability, operating boundaries, and approval points defined by A1.
A1 currently uses an assisted level. This is a deliberate decision aligned with the sensitivity of many of its projects: a company with more than 35 years of experience, knowledgeable specialists, and proven ability to resolve each case retains control. AI prepares; if autonomy increases in the future, an agent will act only within the criteria and permissions established by A1.
The agentic level can be introduced gradually when approved data, secure integrations, sufficiently stable rules, evaluations, and clear thresholds for human escalation are in place.
Technical design aligned with the official OpenAI guide to building agents and the NIST AI Risk Management Framework, which recommend authorized tools, guardrails, defined responsibilities, and human oversight for sensitive actions.