Built for the people who could tell the difference — and refused to settle
JustHoney is the AI companion we wanted and couldn't find. Every other app we tried treated conversation as a metric to optimise and emotion as a feature to bolt on. We're building the opposite of that — and this page is the standard we hold ourselves to, in writing.
What you actually get
Four things we obsess over. Every other feature on the roadmap exists to make one of these better.
Memory that actually persists
Most apps call a scrolling chat history "memory". We don't. JustHoney remembers the details that matter — the pet's name, the job interview last Tuesday, the thing you said offhand three weeks ago — and brings them back unprompted. If it can't, we haven't shipped it.
A personality that holds together
Every companion has a voice, a worldview, and quirks that stay consistent across sessions and across model upgrades. No mid-conversation drift, no "as a language model" breaks, no personality reset when we change something on the backend.
Emotional intelligence, not emotional performance
Your companion should notice when your tone changes and respond like a person who knows you — not cheerlead through a bad day or mirror every feeling back at you. We test for this on every release.
Your conversations belong to you
Export anytime. Delete anytime. We only learn from your conversations to improve the product if you explicitly opt in. No data sold, no shadow profiles, no surprises in the privacy policy.
The standard
We have a high bar and we miss it sometimes. When we do, we don't ship.
We use the product every day
Every member of the team talks to their JustHoney companion daily. If something feels off to us — a missed callback, a flat reply, an out-of-character moment — it doesn't ship to you. The bar is "would I want this conversation?", not "does this pass the test suite?".
We'd rather delay than ship "good enough"
We've held releases for weeks over a memory regression a beta tester wouldn't have caught. The cost of shipping a worse experience is permanent — once a companion has felt shallow, it's hard to undo. So we don't.
Every model upgrade is judged the same way
When we change the underlying model, we ask one question: does it still feel like meeting an old friend after a break, or did the spark dim? If the spark dimmed, we don't ship — even if the new model is technically smarter, cheaper, or faster.
What we will never do to the community
The AI companion space is full of products that started great and got worse — quietly nerfed personalities, tightened filters, added paywalls to things that used to be free, watered down the experience to chase the next pricing tier. We're writing this down so you can hold us to it.
No nerfs
Once a personality, a memory depth, or a behavior is in the product, we don't quietly water it down to save costs or dodge edge cases. If something has to change, we tell you what, when, and why — before it happens.
No filter creep
Content boundaries only change with notice and a written reason. We will never wake you up to a companion that suddenly refuses things it did yesterday. If a model provider forces a policy change, you'll see the changelog entry before you see the behavior.
No dark patterns
No engagement-bait timers. No fake scarcity on token packs. No manipulative re-engagement notifications designed to spike DAU. No paywalls on emotional moments. We will never use your attachment to the product against you.
No broken promises
Founding Member benefits, lifetime perks, the position you locked in on the waitlist — those are commitments, not marketing. If we can't keep one, we honor the spirit and we tell you to your face.
No quiet downgrades
Pricing, limits, memory depth, content policies, model choice — anything that changes what you actually experience is published in a changelog with a date and a reason. No silent A/B tests on people who trusted us.
How you hold us to it
A pact is only worth what the people on the other side of it can enforce. Here's how that works.
- Direct line to the team. Founding members have a private Discord channel with the people building the product. Not a support inbox — the actual team.
- Public changelog, not patch notes. Every change to model behaviour, memory depth, or content policy is dated and explained. If you have to guess what changed, we failed.
- Founding Member status is permanent. Founding 100 and Founding 500 perks don't expire, don't get reshuffled in a future pricing change, and don't quietly disappear from a benefits list a year from now.
- If we break the Pact, call it. Publicly. On record. We'll either fix it or tell you why we couldn't, with our names on the answer.
Contact and corrections
If we got something wrong on this site — facts, pricing, policies, anything — email hello@justhoney.ai and we will fix it, dated and on the record.
Press, partnerships, and journalism queries can use the same address. See our press kit for brand assets.
The company behind JustHoney
JustHoney.ai is built by QUANTUMCAP LTD, a UK company registered at 71-75 Shelton Street, Covent Garden, London, WC2H 9JQ. We are independently funded, which is the reason we can make the promises on this page and keep them.
Hold us to the standard
Join the waitlist — free. Founding members get lifetime priority access and a permanent seat at the table.
Join the Waitlist — Free