Most small businesses can run a working AI agent for $0 to $150 a month using an off-the-shelf platform; custom-built agents start around $1,500 to build and $300 to $800 a month to run. The number that matters isn't the sticker price on the homepage — it's usage-based fees, which can multiply the headline number if the agent runs on every message or every ticket.
Vendors advertise the low end. The real number depends on volume, how many tools the agent touches, and whether you're paying per seat, per action, or per resolved task.
The three pricing tiers
Off-the-shelf, no-code: Platforms like Lindy, Relevance AI, and n8n's AI nodes typically run $0–$150 per month for a small team. This covers most single-workflow use cases: one agent, one job, a handful of tool connections.
Usage-based add-ons: Several of the above bill by action or credit rather than flat seat price. A workflow that fires a thousand times a month costs meaningfully more than one that fires fifty times, even on paper the "same" plan.
Custom-built agents: A single-purpose custom agent (a lead qualifier, a support-deflection bot built on your own data) usually costs $1,500–$5,000 to build, plus $300–$800 a month in hosting and API costs. Full custom systems for larger workflows run well into five figures — not the right starting point for most small businesses.
Gumloop sits in the middle: no-code like Lindy, but priced around usage, which makes it easier to estimate cost before rolling an agent out to a whole team.
The hidden costs vendors don't put on the pricing page
Budget for more than the subscription line. Integration time, retraining the agent when your process changes, and the review time for a human to check outputs in the first month all cost real hours, even if no invoice shows up for them. A rough rule: budget 1.5x the advertised platform price for the first few months, then it should settle closer to the sticker number once the workflow is stable.
Want a cost estimate specific to your business instead of a generic range? Take the free AI Readiness Audit → — it maps your workflows to the tools that fit and gives you a 30-day roadmap, no signup required.
Quick comparison
| Tier | Typical cost | Best fit |
|---|---|---|
| Off-the-shelf (Lindy, n8n) | $0–$150/mo | One team, one or two workflows |
| Usage-based (Gumloop, Relevance AI) | Scales with volume | Workflows with variable, growing volume |
| Custom-built | $1,500–$5,000 to build, $300–$800/mo to run | A single high-value workflow at scale |
Which one should you use?
Start with an off-the-shelf platform: Lindy if you want to describe the job in plain English, Zapier if the task is simple enough for trigger-based rules instead of a full agent, or Gumloop if you expect volume to grow and want usage-based pricing from day one. Reach for a custom build only once you've proven the workflow is worth automating and the off-the-shelf cost has actually become the bottleneck.
FAQ
Is there a free way to try an AI agent first? Yes. Most of the no-code platforms, including n8n and Lindy, have free tiers good enough to test a single workflow before paying for anything.
Why did my AI agent bill jump after the first month? Usually volume. A plan advertised at a flat rate often has action or credit limits that a busier-than-expected month can exceed.
Is a custom-built agent ever worth it for a small business? Sometimes, but only after an off-the-shelf tool has proven the workflow works and the usage-based cost has become the limiting factor, not before.
Related guides
*Ratings and pricing reviewed monthly. Last updated July 2026.*
Bogdex · Founder & editor, woska
Bogdex builds and curates woska, testing AI tools against real workflows to judge which ones actually save time rather than which have the longest feature list.