Usage you can predict before you buy
memmoo meters two things — workflow actions and AI credits. Both are counted per co-worker run. Pick a co-worker, set your volumes, and see the month against any real plan allowance.
Workflow action
One discrete step a co-worker takes: read an email, extract a field, call an API, validate against your ERP, write a row, send a message, display a result. Counted per co-worker run.
Deterministic stepssend, write, display, API call — these cost an action and nothing else.
AI credit
Consumed only when a step calls an AI model. A credit reflects the weight of that model call, not a flat per-request fee.
Weighted so a one-line classification does not cost the same as extracting 40 fields from a scanned invoice.
One-time setup actions are on us
Connecting a device, verifying the Python runtime, cloning the co-worker, installing dependencies and the first-time vector-database build all run once when a co-worker is deployed — never on every run. memmoo does not meter them, so your allowance is spent entirely on business work. Typical one-time setup: 3–8 actions, depending on whether the co-worker runs on your device or in your cloud.
AI credit weighting
| Step type | What it does | AI credits |
|---|---|---|
| Deterministic | Trigger, read, write, display, send, API call | 0 |
| Classify / route | Email triage, intent detection, filtering | 1 |
| Retrieve (RAG) | Vector search + grounded answer | 2 |
| Transcribe | Audio to text | 2 |
| Generate | Draft a reply, summary, formatted response, brief | 2 |
| Extract | Pull structured fields from a PDF, invoice, claim or document | 3 |
Every step also counts as one workflow action. Credits stack on top only where a model is involved.
One RCM claims run, start to finish
Two fixed steps bracket the run; six steps repeat for every claim inside it.
Fixed — once per run
- Trigger — pull the claim batch from the worklist1 action0 credits
- Post summary — + notify the biller1 action2 credits · generate
Per item — repeats for every claim in the run
- Extract patient, payer and encounter data1 action3 credits · extract
- Eligibility check against the payer API1 action0 credits
- Coding validation1 action2 credits · generate
- Scrub against payer-specific edit rules1 action2 credits · generate
- Submit to the clearinghouse1 action0 credits
- Denial / remittance triage1 action2 credits · generate
memmoo bills per co-worker run, not per claim. Batch more claims into one run and your effective cost per claim falls — the estimator below shows exactly how much.
Model your own scenario
Everything updates live. Nothing is stored — reload and it resets.
12 fixed actions + 0 per run · 20 fixed credits + 0 per run · 4 free setup actions
Volumes
This co-worker processes its whole batch in a single pass, so item count does not change its action cost.
Launch — pooled across your account: 2,500 workflow actions and 5,000 AI credits per month.
Plan fit
Launch · 2,500 actions / 5,000 creditsMaximum runs per month on this plan: 208 — that’s 9.5 runs per working day.
The actions allowance alone would allow 208 runs; the credits allowance alone 250 runs. Workflow actions set the limit.
What this costs
Launch lists at $299 per month with an effective cost of $0.120 per workflow action at its included allowance. Overage is $40 per 1,000 actions.
Cost per item at these settings
1 vs 5 vs 10 vs 25 items per run
Typical custom co-worker (benchmark) on Launch (2,500 actions / 5,000 credits), billed per co-worker run.
| Runs / run | Actions per run executed | Actions per run billed | Max runs / mo actions allowance | Max runs / mo binding meter | Effective actions per item |
|---|---|---|---|---|---|
| 1 | 12.0 | 12.0 | 208 | 208 | 12.0 |
| 5 | 12.0 | 12.0 | 208 | 208 | 2.4 |
| 10 | 12.0 | 12.0 | 208 | 208 | 1.2 |
| 25 | 12.0 | 12.0 | 208 | 208 | 0.5 |
Metering FAQ
Sized your scenario? Let’s build the co-worker.
The discovery session turns this estimate into your real per-run profile — before you commit to a plan.