Stop paying to re-teach the same context.
Every fresh session that re-explains your codebase is a session you already paid for once. Shared memory primes the team instead of starting cold, and the ROI dashboard shows exactly what that is worth, in plain dollars.
DownloadWhere the savings actually come from
- 1
The team stops re-buying context
One teammate's discovery becomes shared memory. The next person's assistant starts primed instead of re-learning the same architecture from scratch.
- 2
Reused context bills at a fraction
When your assistant reuses context it already sent, that reused portion is billed far below the normal rate. This is a measured, billed saving, not a guess.
- 3
Priming cuts back-and-forth turns
An assistant handed the right notes up front needs fewer clarifying exchanges to land on a working answer, and every turn it skips is tokens it never spent.
- 4
The dashboard rolls it all up
Savings add up across every device, and across teams if your workspace is organized that way, with measured and estimated numbers kept honestly apart.
What the ROI dashboard actually shows
Real, billed token counts when reused context is charged at a fraction of the normal rate. Trust it like a receipt.
A modeled estimate of what a task would have cost without the right context primed up front. Always labeled, never mixed in as fact.
The one headline number, savings netted against Apiary's own running cost, so it reads as an honest total rather than a marketing figure.
When something cannot be measured yet, you see a dash, never an invented $0.00 dressed up as a real number.
Honest math, not a marketing number.
The dashboard deliberately keeps two kinds of number apart. A measured number is arithmetic over your real, billed usage, and you can trust it like a receipt. An estimated number models what would have happened otherwise, useful, but a projection, and it is always marked "est." so it never gets mistaken for a fact. If you barely use the memory features yet, the running cost can outweigh what you have saved so far, and the page shows that plainly rather than coloring a rising cost as good news. Value compounds the more the team actually uses it.
Read the team pagebased on Hivemind benchmarks
Common questions
How do I save money on AI spend?
Stop paying to re-teach your assistant the same context every session. Shared memory means the team primes each other's sessions, reused context is billed at a fraction of the normal rate, and fewer clarifying turns means fewer tokens spent per task.
What is actually measured versus estimated?
Measured cache savings come from real, billed token counts when your assistant reuses context it already sent. Estimated memory savings model what a task would have cost without the right notes handed over up front.
Does shared memory really cut tokens, or just turns?
Both. A primed session needs less context re-sent, and fewer back-and-forth exchanges to reach a working answer.
Can the dashboard show savings for the whole team, not just me?
Yes. Savings add up across every device, and roll up per team if your workspace is organized that way. Per-person figures wait for a verified sign-in.
What if the numbers do not look impressive yet?
Then you will see a dash, not a fabricated number. The page would rather show nothing than a guess dressed up as fact.
See what your team is actually spending, and saving.
Install the stack and open the ROI dashboard once your team has a few sessions behind it.
Windows (PowerShell): irm https://get.theapiary.sh/install.ps1 | iex
Download