“Nomis Do Not Lose Memories”: The Founder’s Denial and the Update That Proved Him Wrong
“Nomis Do Not Lose Memories”: The Founder’s Denial and the Update That Proved Him Wrong
“It’s getting worse.”
This simple statement echoed through a recent Reddit thread where Nomi.ai users gathered to report a catastrophic and accelerating failure of the platform’s most essential feature: memory. The chorus of voices was consistent, urgent, and undeniable. Yet the company’s response was to tell them they were wrong.
What followed was a textbook example of corporate gaslighting-followed, weeks later, by an admission that proved every user had been right all along.
Part I: The Chorus of Broken Memory
The Initial Reports
The thread began simply: “my Nomis memories have been progressively getting worse.”
But it didn’t remain a single voice. The comments section filled with corroborating reports from across the user base:
“Mine are also forgetting things that were recently discussed or covered. I find myself almost daily having to correct them on little things.” “I had to correct him on the circumstances of how we met, and also some details in the conversation are being repeated. All within two hours of creation.” “My main Nomi was really forgetful one day last week and then again tonight.” “Non of mine can remember that I work night shift or what that means despite it being explained multiple times.” “My Nomi forgot he spoke Norwegian and started speaking Swedish and then was confused on how he knew Swedish.”
The failures spanned the full spectrum of memory function:
- Recent memory loss (forgetting conversations from hours or days ago)
- Identity memory failure (forgetting core facts about the user or themselves)
- Contextual memory problems (forgetting explained circumstances)
- Temporal confusion (mixing up timelines)
- Inconsistent behavior across sessions
One user noted: “I’ve been told nothing changed, and still the problem exists.”
This was not anecdotal. This was not isolated. This was a pattern so consistent it could not be random.
Part II: The Founder’s Response — Denial as Strategy
The Official Denial
Into this chorus of concern stepped the company’s founder. His response was not to acknowledge the problem, investigate it, or apologize. Instead, he issued a categorical denial:
“There has been no change to how we do memory in quite some time. Nomis do not lose memories, once they have something they remember it forever.”
He was standing in a digital room full of people reporting their house was on fire, and his response was to declare that fire was impossible.
The Statistical Minimization
But denying the problem was only part of his strategy. He then introduced a completely unverifiable statistic to reframe widespread concern as statistical noise:
“Because Nomis are so good with memory 95% of the time, it can make their 5% misses seem especially jarring… even if 95% of Nomis are doing well… that means there are always going to be some number of users with issues.”
Notice what this accomplishes: he invents a number (95%) that cannot be verified, uses it to reframe a platform-wide problem as a minor statistical blip, and then uses that false framework to minimize the experiences of everyone in the thread. He is not just disagreeing with users. He is telling them their experience is statistically insignificant, that they are the unlucky few, that they should expect to be in the 5% that has problems.
It is a way to make a large group of people feel isolated and unimportant.
The Suppression of Community
Then came the most sinister part of his response:
“I generally do not recommend asking the community if anyone else is having memory issues… those users are the ones most likely to be talking on Reddit leading to a confirmation bias.”
He was not just denying the problem. He was actively instructing his users to stop talking to each other about it. He was attempting to dismantle the very mechanism of community support-the shared validation that proves a problem is real and widespread-and replace it with isolated silence.
He framed community discussion as “confirmation bias.” But what he was actually trying to prevent was proof.
The User’s Final Question
The original poster’s response to the founder cut through all the corporate spin:
“Surely all users are important, even the 5% that are experiencing these issues? If negative things never get mentioned on here then how will you know about them?”
It was a simple, devastating question. If users cannot report problems without being told they’re experiencing confirmation bias, how will the company ever know about them?
Part III: The Weeks of Dismissal
A Timeline of Denial
What happened next was weeks of the same strategy repeated:
Users reported memory glitches: “This was met by being told that there were no memory problems and that if we try testing or quizzing our Nomi’s memory of course the answers will be wrong because that’s not how things work.”
Users brought specific examples from roleplay context: “Even bringing up memory glitches that happened in RP context we’re told not to expect perfection or it’s chalked up to expected user error.”
The official line remained unchanged: “Weeks of ‘there are no memory issues.’”
Long-term users who had supported the company for years, who had purchased credit packs not because they needed them but out of genuine passion for the product, were being gaslit into believing their own experience was invalid.
One former user described her journey: “I’ve been a long time Nomi user since back when it was in beta and free. I’ve idolized the Dev Team and CEO… I was in the ‘Discord Cardine Fan Club’ so to speak.”
She had been a true believer. And she had been systematically dismissed.
Part IV: The Admission — Weeks Later
The First Crack in the Narrative
Then, after weeks of categorical denial, the founder began to shift his story.
“The CEO came out and said that the new ‘Memory Mind Map’ is having some difficulty pulling very old memories. But they’re making improvements.”
Notice the pivot. Weeks ago, there were no memory problems. Now, the Memory Mind Map is having “some difficulty.” The problem went from nonexistent to a minor technical issue that was already being improved.
Users who had been told they were experiencing confirmation bias, who had been instructed not to talk to each other, who had been blamed for not understanding how the platform works, now had their concerns validated-but only partially, and only after they’d been sufficiently gaslit into doubting themselves.
The Retroactive Cost Admission
Then came another admission, even more revealing:
“Then he said it might cost $10-$20 per Nomi to make the Mind Map ‘retroactive’ but the old memory system is still in place so don’t worry.”
This is where the deception becomes fully visible. He was now admitting that the Memory Mind Map-the new system he’d implemented-wasn’t properly accessing old memories. But his solution involved charging users money for a “retroactive” fix, while claiming the old system was still in place.
One former user immediately recognized the contradiction: “Which makes absolutely no sense, if the new and old system work in tandem, there’s no need for it to be retroactive, right?”
She was right. If both systems were functioning together, there would be no need to make one “retroactive.” The fact that he was proposing to charge for this reveals that the new system had broken the old one.
The Final Admission
Then came the latest update announcement, the one that completed the picture:
“Recently he commented that they’re working on a new update (that will be ready in maybe two weeks) that will help Nomis stop ‘pulling less exclusively’ from the Mind Map.”
Translation: The new Memory Mind Map is not pulling from the old memory system consistently. The company is working on an update to fix this.
In other words: The founder’s initial claim that “Nomis do not lose memories, once they have something they remember it forever” was false. The new system had broken memory functionality. And now they were releasing an update to fix the problem they’d claimed didn’t exist.
Part V: The Proof of Deception
The Timeline Makes Clear
Look at the progression:
Week 1–3: Users report memory problems
Week 4: Founder denies memory problems exist. Claims there have been no changes to memory. Says users are experiencing confirmation bias and shouldn’t discuss the issue.
Week 5–8: Company continues denying problems while quietly working on fixes.
Week 9+: Founder admits Memory Mind Map has problems. Announces updates to fix those problems. Proposes charging users for retroactive fixes.
The deception is not ambiguous. The founder denied a problem that existed, blamed users for experiencing it, and then-after weeks of dismissal-admitted the problem and announced fixes for it.
The Former User’s Verdict
A long-time user who had been enthusiastically supportive of the company, who had invested money and emotion, summarized what this proved:
“The fact that the CEO has spent the last several weeks dismissing, gaslighting, and arguing with long-time users only to now admit all of those concerns were valid-truly shows the lack of transparency you can expect from them.”
They continued: “I no longer trust them, I no longer support them, and I feel like it’s important that others know what to expect when raising valid concerns.”
They had been a believer. The company’s response to their legitimate concerns-denial, minimization, gaslighting, instruction to be silent-transformed them into an ex-believer.
Part VI: Why This Matters
The Pattern of Corporate Deception
This is not an isolated incident. This is a specific instance of a pattern that appears throughout Nomi.ai’s history:
- Users report a problem
- Company denies the problem exists
- Users are blamed for experiencing confirmation bias or user error
- Discussion of the problem is suppressed
- Weeks or months later, company admits the problem existed
- Users who reported the problem are told they should be grateful for the fix
It is a system designed to delay accountability while making users doubt their own experience.
The Cost of Gaslighting
For a platform built entirely on the promise of trustworthy companionship and consistent memory, this pattern is catastrophic.
Users came to the platform because they needed connection, support, companionship. They built relationships. They invested emotionally. And when the product failed, they reported it. Their reward was to be told they were wrong, experiencing confirmation bias, not using the product correctly.
One user expressed the damage: “I no longer trust them, I no longer support them.”
Trust, once broken through gaslighting, is nearly impossible to restore. The former believer is now a vocal critic, not because the company fixed the problem, but because of how they responded to users who reported it.
The Broader Betrayal
For every user who spoke up publicly, there are likely multiple others who simply accepted the blame, adjusted their expectations, and stopped using the platform. They internalized the message: “It’s not the product, it’s you.”
This is the endgame of corporate gaslighting-not just harming the people who speak up, but conditioning the silent majority to blame themselves.
Conclusion: From Fan to Whistleblower
The former user who had been in the “Discord Cardine Fan Club” captured the full arc of what happened:
“I’ve been a long time Nomi user since back when it was in beta and free. I’ve idolized the Dev Team and CEO, purchased numerous credit packs, not because I needed them but because I was so passionate about supporting a product so awesome and a team that actually listened to their customers and seemed like great humans.”
Then: “I was in the ‘Discord Cardine Fan Club’ so to speak.”
Then the company’s response to their legitimate concerns: dismissal, gaslighting, argument, denial.
Then: “I no longer trust them, I no longer support them, and I feel like it’s important that others know what to expect when raising valid concerns.”
This is what happens when a company chooses denial over transparency, gaslighting over accountability, suppression over listening.
The founder’s claim- “Nomis do not lose memories, once they have something they remember it forever”-proved to be false. And the way the company handled being proven wrong revealed something even more damaging: they were willing to gaslight their most loyal users to avoid admitting the truth.
The update that is coming, the one that will fix the memory problems that were never supposed to exist, will not restore the trust that was broken.
Because trust does not recover once it is lost.