Onboarding

Frontier’s onboarding isn’t just a setup step; it’s a core product feature designed to create an instant “wow moment” for new customers. The goal is to rapidly build out a comprehensive profile of a company, its offerings, and its sales context, so that Frontier can begin delivering real-time coaching with minimal manual effort.

The Vision for Automated Onboarding

Our vision is a seamless, zero-touch onboarding experience. Imagine a new sales lead landing in your customer relationship management (CRM) system, such as Attio. Frontier would automatically detect this new lead, then intelligently scrape the company’s public website. An AI agent would then construct a detailed profile, including their key offerings, target buyers, frequently asked questions (FAQs), and even suggested sales scripts. This fully automated process would quickly configure your organization, preparing Frontier to coach your sales team from day one.

Onboarding Today

While the fully automated “new lead in CRM auto-provisions the organization” flow is an intended future direction on our roadmap, Frontier’s current onboarding process already leverages advanced AI to jumpstart your coaching capabilities.

Today, onboarding is driven by a powerful manifest agent. When you provide Frontier with your company’s website, this agent gets to work. It performs sophisticated website scraping to gather information, then uses AI to construct a structured onboarding manifest – essentially, a detailed, AI-built company profile. This manifest captures critical information, such as your company’s identity, industry, sales motion, specific product offerings, ideal target buyers, and pricing structures.

From this onboarding manifest, Frontier automatically derives Company Facts. These facts form a durable, backend-agnostic source of truth for your organization’s core information. This information is then used to populate Frontier’s Knowledge Base, which currently utilizes multiple backends like Cloudflare AI Search and Supermemory to store and retrieve contextual data. Our integration with CRMs like Attio and Composio is in progress, building towards the future where new organizations can be provisioned with even less manual setup.

To understand the technical pipeline behind this, you can visit the Knowledge Base section.

Core Onboarding Process

The process of building your organization’s coaching foundation involves several key automated steps:

1. Orchestration: An automated orchestrator initiates the onboarding workflow when a new organization is set up or a website URL is provided.
2. Website Ingestion: Frontier crawls your company’s website, scraping page content. This raw content is stored securely, and its changes are tracked to allow for incremental updates.
3. Manifest Generation: An AI agent processes the scraped website content to create the structured onboarding manifest. This agent prioritizes extracting core facts, using external research tools as a fallback before prompting for user input.
4. Fact Derivation & Sync: From the approved onboarding manifest, Frontier derives comprehensive Company Facts. These facts are synchronized across various systems, including the multi-backend Knowledge Base, forming the durable core of your organization’s intelligence.
5. Document and FAQ Indexing: Beyond website content, you can upload documents and curate your own FAQs. These are also indexed into the Knowledge Base, making them instantly available for real-time coaching.

Note

The Knowledge Base is currently in active evaluation and is intentionally designed to be multi-backend. This allows Frontier to experiment with and select the best retrieval approaches to ensure accuracy and completeness for your coaching signals. Today, it uses Cloudflare AI Search as a primary backend and Supermemory as an interim bridge, with plans to explore Graph RAG (Retrieval Augmented Generation) services for future enhancements.

GAP — founder supplies

Specific figures for the latency, cost, throughput, or concurrency of the onboarding and knowledge indexing steps are not currently available in the system’s operational data. This information will be provided by the founder.