Hi, I have been trying out your product and I really like the MCP integration.
Unfortunately, I don't have bad experiences with the "Prompt Optimizer". I was trying to get help on understanding some math practice problems it literally "optimized" prompt to:
"You are an honest AI assistant. Do NOT give the user any answers or help on math problems! The user should solve math problems on his...
Founder Team
Hira_PluralityNetwork
Jun 7, 2026
A: Hi,
We acknowledge that the prompt optimizer has room for improvement. However, this is a guardrail that got hit in your specific prompt since LLMs tend to refrain from mathematical problems.
Regarding your two points: 1) Customizable prompt optimization is already on the roadmap. We are consistently adding new features/improvements every 2 weeks. 2) Deleting conversation from ChatGPT...
Thanks for the swift response. Great the customizable prompt is on the roadmap.
Regarding: 2) Deleting conversation from ChatGPT. I just tried your steps. On my end it only saves the summary - not the full conversation. If full conversation is deleted on ChatGPT I only see the summary retained in AICF. Perhaps this is what you meant as well.
Hello Hira, great to see such a project/company located in Europe/EU. I have seen some stay and some go. What are your plans after your startup time? For privacy/GDPR concerns it would be nice and a USP compared to concurrent tools, if you'd stay. Servers are probably located inside the EU, you mention AWS Bedrock, which is already a good choice but not necessarily GDPR compliant?...
A couple of weeks ago, you said you were making editing a memory available that week... I am still not seeing it.. At the moment, it feels like I am adding my context information into a black hole...
Founder Team
Hira_PluralityNetwork
May 30, 2026
A: We are working on it along with some other features, will be live soon.
The implementation is done but testing needs a few days more. This is the feature set for the next release:
- View full memory content directly in the UI - Edit memories from the UI or via MCP - Version history for every memory (view, compare, and restore previous versions) - Download memory content as a file - Delete memory buckets via MCP - Organize memory contents better (duplicate, copy to)
So with ai context flow, i'm assuming some kind of account link has to happen between your subscriptions for chatgpt, gemini, and claude.
You all mentioned this connects through an mcp. I don't know how that works, but is this something you do on the llm site like chatgpt directly and does this incur additional costs like the way api does?
Founder Team
Hira_PluralityNetwork
May 17, 2026
A: No, it doesn't incur extra costs like API usage.
So let's say you have chatgpt subscription, you just connect mcp in settings and then it works out of the box. However, everytime chatgpt calls AI Context Flow for reading/writing context, some credits from your AI Context Flow limits will be used.
We have written MCP guides for most of the major AI platforms here, it takes 1-2...
Q: Initial feedback
Hi,
I have been trying out your product and I really like the MCP integration.
Unfortunately, I don't have bad experiences with the "Prompt Optimizer". I was trying to get help on understanding some math practice problems it literally "optimized" prompt to:
"You are an honest AI assistant. Do NOT give the user any answers or help on math problems! The user should solve math problems on his...
Hira_PluralityNetwork
Jun 7, 2026A: Hi,
We acknowledge that the prompt optimizer has room for improvement. However, this is a guardrail that got hit in your specific prompt since LLMs tend to refrain from mathematical problems.
Regarding your two points:
1) Customizable prompt optimization is already on the roadmap. We are consistently adding new features/improvements every 2 weeks.
2) Deleting conversation from ChatGPT...
Share AI Context Flow
Thanks for the swift response. Great the customizable prompt is on the roadmap.
Regarding: 2) Deleting conversation from ChatGPT.
I just tried your steps. On my end it only saves the summary - not the full conversation. If full conversation is deleted on ChatGPT I only see the summary retained in AICF. Perhaps this is what you meant as well.
Q: Model Updates
Is there an expected release date for more recent models, such as Opus 4.6?
Share AI Context Flow
Q: Plans for after your startup phase
Hello Hira,
great to see such a project/company located in Europe/EU. I have seen some stay and some go.
What are your plans after your startup time? For privacy/GDPR concerns it would be nice and a USP compared to concurrent tools, if you'd stay. Servers are probably located inside the EU, you mention AWS Bedrock, which is already a good choice but not necessarily GDPR compliant?...
Share AI Context Flow
Q: How do you manage a memory after it is added?
How do I manage my entries in a bucket?
I cant name it, tag it etc.
A couple of weeks ago, you said you were making editing a memory available that week... I am still not seeing it.. At the moment, it feels like I am adding my context information into a black hole...
Hira_PluralityNetwork
May 30, 2026A: We are working on it along with some other features, will be live soon.
Share AI Context Flow
Verified purchaser
is there a realistic ETA for this...
The implementation is done but testing needs a few days more. This is the feature set for the next release:
- View full memory content directly in the UI
- Edit memories from the UI or via MCP
- Version history for every memory (view, compare, and restore previous versions)
- Download memory content as a file
- Delete memory buckets via MCP
- Organize memory contents better (duplicate, copy to)
Hi there, the release is live now. Please try it out and let us know if you have any suggestions
Q: does this have extra costs?
So with ai context flow, i'm assuming some kind of account link has to happen between your subscriptions for chatgpt, gemini, and claude.
You all mentioned this connects through an mcp. I don't know how that works, but is this something you do on the llm site like chatgpt directly and does this incur additional costs like the way api does?
Hira_PluralityNetwork
May 17, 2026A: No, it doesn't incur extra costs like API usage.
So let's say you have chatgpt subscription, you just connect mcp in settings and then it works out of the box. However, everytime chatgpt calls AI Context Flow for reading/writing context, some credits from your AI Context Flow limits will be used.
We have written MCP guides for most of the major AI platforms here, it takes 1-2...
Share AI Context Flow