Guide · made with Zapier
Why is my AI bill exploding, and which steps actually need a model? A 6-step test
Break your workflow into steps and mark which ones need judgment. In Nick's example, one of six did. Zapier modeled 24 workflows: 71% average savings.
The short answer
Most of your AI bill is going to work that does not need AI at all, not work that could be optimized. Nick calls that pattern token maxing: throw the model at everything and assume the value shows up later. Value maxing is the opposite, and the test is simple. Take one workflow, break it into its actual steps, and mark which steps need a model's judgment. In his competitor-monitoring example, six steps came out of it and exactly one needed the model, because a trigger, a string match, a database read, and formatting all have right answers you can write down as rules. Zapier modeled 24 common agent workflows this way and the average savings was 71%.
Get the Zapier framework guide →Built with Zapier, the sponsor of this video.
Step by step
Start with the trigger, not a model call ▶2:04
A new item shows up in the monitoring feed. That is not a judgment call, it is a trigger. It should require zero tokens.
Match the competitor with a string match ▶2:13
Does this item match a competitor you already track? That is a string match, and a model has no advantage here. In fact it can get it wrong. So this one is also a waste of tokens.
Pull the CRM record deterministically ▶2:24
Pull that competitor's record from the CRM. That should be deterministic. It is a database read.
Clean and format the fields without tokens ▶2:32
Clean up the dates, trim the fields, handle general formatting. None of this is ever worth a token.
Spend the tokens here, on judgment ▶2:39
Read this and tell me what it means for our company, in three sentences. That requires judgment. That is the part Nick is happy to pay for.
Package it and post to Slack ▶2:50
Package the whole thing up and send it to Slack. That is delivery, not judgment. Six steps total, and one of them needed the model.
“So the question isn't how do we spend less on AI, it's what is my AI spend actually achieving?”
Nick Standlea, in the video
Key moments
- ▶0:00Most of your AI bill is waste
- ▶0:24Uber, Salesforce, and a $150,000 month
- ▶0:58Two agents, one outcome, double cost
- ▶1:11The reliability problem under the cost
- ▶1:38The competitor-monitoring workflow example
- ▶2:54Six steps, one real model call
- ▶2:5924 workflows modeled, 71% average savings
- ▶3:23The three rules of value maxing
Questions people ask
what is tokenmaxxing
Token maxing is the pattern of throwing the model at everything and assuming the value shows up later. Nick's point is that it does not always show up. You end up paying a model to predict its way through if-then statements, which plain automation would do for free.
what is valuemaxxing
Value maxing is spending tokens only where a model's judgment actually earns its cost. It comes down to three rules: default to deterministic, spend tokens only on judgment, and keep your model options open. The question stops being how do we spend less on AI and becomes what is my AI spend actually achieving.
how do I reduce my AI API costs
Take one workflow, go through it step by step, and mark which steps actually need a model. Anything with a right answer you can write down as a rule is an if-then, so it should cost zero tokens. When you remove model calls and replace them with old-fashioned deterministic automation, you save money. Nick's guess is that fewer steps need the model than you think.
which tasks actually need an LLM and which should be deterministic
Spend tokens on judgment: summarizing, classifying ambiguity, drafting, and deciding when the rules do not genuinely apply. Keep everything else deterministic, including triggers, string matches, database reads, and formatting. If the task has a right answer you can write down as a rule, a model is the wrong tool for it.
are AI agents more reliable than regular automation
Not on rule-based work. An LLM predicts a likely answer, it does not execute fixed logic. So on a task with a right answer it will occasionally miss, and it does so at a premium price. Deterministic automation is both cheaper and more reliable for that kind of step.
how much can you save by replacing agent steps with automation
Zapier modeled 24 common agent workflows this way and the average savings was 71%. Your mileage will depend on how complex your workflows are and which models you are using. But the direction holds.
why are companies overspending on AI tokens
Uber burned through its entire AI budget in the first four months of the year. Salesforce is looking at $300 million in Anthropic tokens this year, during a hiring freeze, and one engineer reportedly put up a $150,000 Claude Code bill in a single month. The cost also compounds: teams hand an agent work that plain automation could do, then build a second agent to check the first one's work. Now you are paying twice for one outcome.
Zapier sponsored this video. Nick chose what to build and what to say.