First production agent
$10k + $1k/mo
Typical starting point: onboarding with the first agent included, private runtime, and one live agent instance.
Pricing
Pricing for production agents, not software seats.
Buy the governed runtime, build the agents that do real work, and expand only when observed demand justifies more capacity.
User seats
$0
Pricing follows governed agents and runtime capacity, not employee headcount.
Pricing Model
Clear up-front pricing. No seat math.
The model is intentionally direct: onboard the platform with the first agent included, run it privately, and pay for each deployed agent instance.
Platform onboarding
$10,000 one-time
Customer onboarding, governance setup, admin alignment, and the first production agent build.
Agent instance
$250 / agent / month
Recurring fee for each dedicated agent operating in production with its own role and governed access.
Additional agent build
$5,000 one-time
Bespoke workflow mapping, prompts, tools, review gates, integrations, testing, and production handoff for each added agent.
Local inference-capable AI computer
$750/mo fully managed
On-prem runtimeManaged hardwareLocal inference
On-prem runtime appliance managed by Aech AI, including hardware lifecycle, updates, monitoring, and local inference-capable compute.
Agent build scope includes workflow mapping, required integrations, and review gates
No per-user fees
No separate SaaS deployment surcharge
One managed local runtime can support many agents before more capacity is needed
Nominal one-agent monthly cost is $1,000/month: $750/month for the fully managed local inference-capable AI computer plus $250/month for one deployed agent.
How It Scales
Adding agents is cost effective after the local runtime is in place.
The managed local runtime is the base monthly cost. It can run many agents, so growth is driven mostly by one-time bespoke build services and a small monthly cost per deployed agent.
1 agent
$10,000 onboarding includes the first agent. Monthly is $750 for the managed local inference-capable AI computer plus $250 for the agent.
One-time
$10,000
Monthly
$1,000/mo
2 agents
Adds one $5,000 bespoke agent build and one $250/month agent instance on the same managed local inference-capable AI computer.
One-time
$15,000
Monthly
$1,250/mo
5 agents
Four additional agent builds plus five $250/month agent instances, assuming the first local runtime has enough capacity.
One-time
$30,000
Monthly
$1,750/mo
Cost Structure
The model separates runtime from bespoke implementation.
The managed local inference-capable AI computer is the base runtime cost.
The first agent is included in new-customer onboarding.
Additional agents add implementation cost once, then a small monthly operating cost.
No per-user pricing and no separate capability packages.
FAQs
Pricing questions worth answering up front.
What is the nominal base monthly cost for one agent?
The base monthly cost for one deployed agent is $1,000/month: $750/month for the fully managed local inference-capable AI computer plus $250/month for the agent.
What does the $10,000 new-customer fee include?
It includes customer onboarding and the first production agent build. After that, additional bespoke agent builds are $5,000 one-time each.
What happens when we add more agents?
Each additional agent adds a $5,000 one-time build service and $250/month for the deployed agent instance. The hardware can run many agents, so adding agents is cost effective once the local runtime is in place.
Do you charge per user or sell separate capability packages?
No. Pricing is based on the local runtime, deployed agents, one-time bespoke agent builds, and the new-customer onboarding fee.
Why is there a one-time build service for each agent?
Agent Aech agents are customer-specific implementations. The one-time build fee captures workflow mapping, prompts, tools, review gates, integration work, testing, and production handoff for that agent.
Next Step
Bring one workflow and we will map the first agent.
The first conversation should validate the workflow, deployment fit, and whether one managed local runtime is enough for the first set of agents.