Frequently asked questions

CraftAgent Builder is a no-code workspace for creating AI agents yourself: you connect knowledge sources, wire up integrations, set guardrails, and test agent behavior before anything goes live. It is currently in early access. Request access through the contact page and we will follow up with timing for your use case.

Customer service agents, sales qualification agents, internal knowledge assistants, workflow automation bots, data analysis assistants, and more. Agents fit almost any business function where information processing, decision making, or communication would benefit from automation.

No. The builder is designed so operators and founders can assemble agents without writing code. Technical users get more room to customize: prompts, routing logic, and integrations can all be tuned by hand if you want to go deeper.

CraftAgent connects with over 1,000 business tools, including CRM systems such as Salesforce and HubSpot, communication tools such as Slack and Microsoft Teams, productivity suites such as Google Workspace and Microsoft 365, and automation platforms such as n8n, Make, and Zapier. Custom API connectors are available for proprietary systems.

There is no public product pricing yet. The builder is in early access, and we open seats in small batches so we can work closely with each team. Early-access teams see pricing before it is published and help shape the plans.

Typical implementation timelines run from 2 to 8 weeks depending on the project. Simple agents with standard integrations can be live in about 2 weeks, while enterprise solutions with custom integrations and large knowledge bases usually take 6 to 8 weeks. We give you a specific estimate during the initial consultation.

Three phases. Discovery: we analyze your processes, document requirements, and create a project roadmap. Development: we engineer the agent architecture, build integrations, implement security measures, and test iteratively. Deployment: we ship to production, train your team, and monitor performance. A dedicated implementation manager keeps you updated throughout.

Key stakeholders join the discovery sessions (typically 2 to 3 meetings), review designs, and take part in testing. We recommend naming a project owner on your side as our main contact. Most implementations need about 4 to 8 hours per week from your team in the first few weeks, then less as the project progresses.

Yes, training is included with every implementation. We deliver role-specific training for administrators, end users, and technical staff through live sessions, documentation, and recorded videos. Enterprise clients can add on-site training and custom training materials.

Plans are published: Standard at $2,999 per month, Professional at $7,499 per month, and custom pricing for enterprise deployments. Every plan includes implementation, training, and ongoing support, and annual billing earns a 15% discount. There are no hidden fees.

We track concrete metrics: hours saved through automation, operational cost reduction, conversion rate changes, response times, customer satisfaction scores, and error rates. The analytics dashboard gives you visibility into these numbers, and we review them with you regularly to keep improving results.

All data is encrypted in transit and at rest. Access is restricted to authorized personnel on a need-to-know basis, and your data is never used to train AI models for other clients. Data residency options can keep information within specific geographic boundaries, and we provide data processing agreements that spell out exactly how your information is handled.

Yes. For services and enterprise engagements we support on-premise deployment, private cloud, and hybrid setups that keep sensitive data inside your perimeter while using cloud resources where it makes sense. Talk to us about your compliance requirements and we will scope the right deployment.

Every agent runs on a truth-first architecture: responses are checked against your approved knowledge sources before delivery, answers carry source attribution, and confidence scoring routes uncertain cases to a human instead of guessing. You decide where the escalation thresholds sit.