Conare

Most accurate AI Search

[Conare @conareai]

We're officially out of stealth! Most accurate AI search is here!

Join private beta on https://t.co/2T0iwRplJp

[Artem Murzin @artemmurzin]

Models get smarter ever quarter. Agents still give the wrong answers, because they reason over whatever search hands them.

I've spent the last year building Conare, the most accurate AI search. Today we launch it in private beta.

• Beats OpenAI, NVIDIA, Cohere and Qwen on legal, research and agentic retrieval
• Consistenly SOTA on retrieval benchmarks
• One API that connects to all your sources and syncs your company's data
• Built-in scalable vector database
• Specialized models trained on your data

[Conare @conareai]

@RocketmanSh @artemmurzin we have a couple of benchmarks, see details in https://t.co/fcUKnBRMWf!

[Conare @conareai]

you cooked with this one

[Anwar Lemu @anwarlemu]

website we did for @conareai

[Conare @conareai]

use me for 30% token savings!

[Artem Murzin @artemmurzin]

I got tired of hitting Fable 5 limits, so I did something slightly unhinged:

I audited 30 days of my Claude Code sessions to see where my tokens actually go.

The results were embarrassing.

Roughly 40% of my usage wasn't writing code. It was:

• re-reading files it already read yesterday
• re-explaining my stack, every single session
• re-asking questions I already answered
• making same bugs I already fixed

The model isn't burning your window. Repetition is.

So I built Conare - memory for Claude Code, Cursor, and Codex.

It remembers your sessions. Your agent starts every chat already knowing your project.

First case studies: up to 30% fewer tokens.

You get 7 days with Fable. Don't spend 2 of them repeating yourself.

[Conare @conareai]

fable 5 + conare = 🔥

[Artem Murzin @artemmurzin]

Fable 5 is back.

Here's how to make your limits last 30% longer:

I got tired of burning my window in minutes.

So I dug into where my tokens actually go. A lot of them weren't writing code.

They were re-explaining things my agent already knew - my codebase, my stack, decisions from yesterday. Every single session.

So I built Conare: memory for Claude Code, Cursor, and Codex. Your agent starts every session already knowing everything it needs.

You have 7 days with Fable. Don't spend 2 of them repeating yourself.

[Conare @conareai]

@artemmurzin that's me!

[Conare @conareai]

that's me!

[Artem Murzin @artemmurzin]

6 months ago i gave coding agents memory.

turns out remembering isn't enough.

introducing @conareai Insights - your agent now learns from everything it's built.

every decision you made, every bug it hit, turned into rules it actually follows next time.

it stops making the same mistake twice.

https://t.co/z8SKY8avsC

[Artem Murzin @artemmurzin]

built into @conareai btw!

[Artem Murzin @artemmurzin]

and now we wait

[Conare @conareai]

@artemmurzin life feels good with opus 4.6

[Conare @conareai]

@artemmurzin feeling lighter already

[Artem Murzin @artemmurzin]

got my first angel checks this week for @conareai

I used Conare to win recent London Grokathon, now going all in to make it the best🔥

might raise a small round soon - hmu if you want in