Real planning numbers for AI agents, AI receptionists, automation, and consulting, from a team that builds and runs these systems for named clients.
By Shawn Kercher, Founder & CTO, HummingAgent · Updated August 2026
The Short Answer
Subscription AI software runs roughly $50 to $2,000 per month depending on the tool and volume. Custom-built AI that you own runs roughly $500 to $10,000 in one-time setup for a single workflow, $5,000 to $25,000 for a Private GPT or internal knowledge assistant, and $25,000 to $100,000+ for enterprise multi-team systems, plus optional scoped monthly support. The ranges below are planning numbers; the real number comes from your workflow volume, integrations, and ownership requirements.
Planning ranges based on 2026 market pricing and our own scoped builds, not a quote. Final pricing depends on scope, integrations, data sources, and support needs.
Want the deep dive?
The full guide breaks these ranges down by solution type, with ROI math, cost-per-conversation examples, and when automation is not worth it.
See the full AI automation pricing guideThe Real Decision
Most pricing confusion comes from comparing these two as if they were the same product. They are not.
What Drives Cost
One AI receptionist costs far less than automating five departments. Scope is the biggest driver of price.
A subscription tool is cheap to start but generic, and you rent it forever. A custom build costs more up front, fits your exact workflows, and you own it.
Connecting your CRM, phone system, QuickBooks, or proprietary tools adds engineering time and moves builds toward the top of each range.
Private or governed AI that touches sensitive data takes more rigor than a public chatbot, which is why Private GPT builds start higher.
Optional monthly management keeps systems tuned as your business and the underlying models evolve. It is scoped, not a mandatory per-seat fee.
Real Builds, Real Clients
Three named, published case studies. Each maps to a row in the pricing table above.
A 24/7 AI chatbot ("Chatty") that handles customer service across 9,085 managed parking spaces.
Chatbot-class build: the first row of the pricing table above.
Read the case studyColorado's premier legal research firm moved from paper subscriptions and manual PDF searching to a fully digital AI search platform.
Workflow-automation-class build with heavy document processing.
Read the case studyA private ChatGPT with memory across clients and projects, using GPT, Claude, and 30+ models while keeping company data private.
Private GPT-class build: governance and data boundaries drive the budget.
Read the case studyFAQ
Subscription AI software typically runs $50 to $2,000 per month depending on the tool and volume. Custom-built AI you own typically runs $500 to $10,000 in one-time setup for a single workflow, $5,000 to $25,000 for a Private GPT or knowledge assistant, and $25,000 to $100,000 or more for enterprise multi-team systems, plus optional monthly support.
Plan on $500 to $5,000 in setup and $200 to $2,000 per month for an AI voice agent or AI receptionist, depending on call volume, phone system, booking rules, and CRM integrations. Basic off-the-shelf tools sit at the bottom of that range; custom-built receptionists scoped around your call flows and owned by you sit toward the top.
Custom AI builds are priced by scope. A single-workflow agent typically starts around $1,000 to $10,000 in setup, a Private GPT or internal knowledge assistant runs $5,000 to $25,000, and multi-system enterprise automation runs $25,000 to $100,000 or more. Unlike a monthly SaaS tool, you own the build, and ongoing support is optional and scoped.
AI consulting cost depends on organization size, stakeholder access, data readiness, and whether the work is strategy-only or includes implementation. A good engagement delivers workflow analysis, business-case assumptions, and a prioritized roadmap before any larger build is scoped, so you spend on strategy once instead of on the wrong software for years.
Often, for repetitive high-volume work. U.S. Bureau of Labor Statistics data puts the median receptionist wage around $34,000 to $37,000 per year before taxes and benefits, while an AI voice agent typically costs $200 to $2,000 per month and answers around the clock. The honest comparison is workflow volume, error cost, integration effort, and the value of never missing a call, not just salary versus software.
Yes, but they are optional and scoped. Custom builds carry model usage costs (public per-token pricing from providers like OpenAI and Anthropic) and, if you want it, a monthly management plan that keeps the system tuned. Off-the-shelf tools, by contrast, charge every month for as long as you use them and you never own anything.
Shawn Kercher is the founder and CTO of HummingAgent, winner of the TMC 2025 AI Agent Product of the Year. A former Comcast Labs engineer, he led computer vision projects across 1,600 data centers and 40,000 technicians before building custom AI systems for businesses. The ranges on this page come from scoping and running real client builds, not vendor marketing.
Tell us what you want to automate and we will give you an honest estimate, including whether an off-the-shelf tool would serve you better than a custom build.
Get a scoped estimate