The product
One engine.
Two ways to buy it.
coreCerebrum is a shared, permanent memory for a whole company. The Harness trims every request. Cerebrum stores every solved problem and serves it free on reuse. It runs on your keys, across Claude, ChatGPT, and Gemini.
What every seat includes
The brain, the Harness, and the meter.
Trims dead-weight tokens off every request before it leaves. Smaller calls, same answers.
Stores every hard solve once and serves it free on reuse. One person's answer becomes the team's.
Tokens routed, tokens avoided, percent reduction. Verifiable against your own provider console.
Claude, ChatGPT, Gemini. Your keys, your infra, no markup on a single token.
Packages
Two packages. The usage meter stays off.
An active user is any seat that routed a call that month. Idle seats cost $0. There are no metered usage charges, ever.
Billed only for seats that were actually used that month. Idle seats cost $0.
- The brain plus the Harness
- Any model, your keys, your infra
- The meter, in tokens you can verify
- Unlimited usage, no metered usage charges, ever
Everything in Cerebrum, plus design and build anything: sites, apps, dashboards. DNA is simply Cerebrum plus $150.
- Everything in coreCerebrum
- Design and build sites, apps, and dashboards
- Wireframe and prototype free, unlimited
- Pay a go-live fee only when a site reaches production
Go-live · Design DNA only
$500 per launched site
plus $50/mo hosted by us
$1,500 one-time
self-hosted, no monthly
Prepaid packs: 5 launches $2,000 · 10 launches $3,750. Pay only when a site reaches production. Wireframe and prototype free, unlimited.
Enterprise · custom
200+ seats, SSO, or on-prem.
Priced to your scale and governance needs, not to a token meter. Same brain, same Harness, run inside your controls.
Talk to us →Plan calculator
Size it to your team.
Annual billing is 20% off and on by default. Enter your current bill for a private estimate at your own rate. We never author a dollar figure. That input and its estimate are yours.
Your number, your provider rate. Used only for your own estimate.
Your plan
8 active seats by $150/mo, annual 20% off
Total
$960 /moTwo readers, one package
Same engine, whether you pay per token or per plan.
You pay per token
Your bill climbs with usage. The Harness makes each call leaner and the brain removes the repeat calls entirely. The meter reports the drop in tokens, at whatever rate you negotiated. You keep your caching and see only what stacks on top.
You pay a flat plan
On a $20-$200/mo plan the first win is headroom: do 3-5x more inside the plan you already pay for and hit rate limits far less. Over time, as reuse takes over the repeat work, many teams move down to a cheaper tier without running out, the way we did at corePHP. And the memory is the reason you stay.
Point it at your real workload for 30 days.
No card. If tokens-per-task do not drop at least 20% in your first paid month, that month is free.