Trader | Evaluating market trends & Emerging alpha
Joined February 2012
- Tweets96.1K
- Following601
- Followers3.4K
- Likes117K
Dylan retweeted
a memory for an agent on a device is easy. but it's local. conversations around agentic memory at scale ultimately turn into a data management discussion.
ie: a company has many individual agents, but a company "brain" or data source is needed for them to have cohesion.
these are two different things.
how will the agents connect?
how will the agents make sure they aren't working from stale entries?
how will the agents update the data?
how will they communicate and discuss actions to take related to the data?
additionally, advanced memory solutions for average users are still extremely hard to package and distribute. claude code and codex are far off from the average consumers skillset & memory integration into the claude or ChatGPT app is harshly limited. Grok Bots virtual machine design is the first thing that may be palatable for average users which enables agentic tooling to be more easily accessible.
this has our wheels turning at @sibyl_labs_ and we are invigorated by this new direction potential
Dylan retweeted
The most significant barrier to adopting advanced AI isn't the complexity of the technology, it is the initial willingness to begin.
Whether due to unfamiliarity with technical interfaces or simply not knowing where to start, the entry point can appear daunting.
While tools like Claude, Codex, and Hermes are widely discussed, the reality is that only a small fraction of professionals are truly leveraging agentic AI. The gap between awareness and utilization remains substantial.
You do not need a background in engineering to start
The key to overcoming this inertia is simple:
1. Identify a single, repetitive workflow.
2. Select one platform to experiment with.
3. Use the AI as a collaborative partner to outline your first steps.
The command line and the prompt are simply new mediums for problem-solving.
Start your agentic journey with Sibyl Memory on Grok Bot
One command (sibyl setup grokbot) and your agent keeps durable, local memory across every session.
The most significant barrier to adopting advanced AI isn't the complexity of the technology, it is the initial willingness to begin.
Whether due to unfamiliarity with technical interfaces or simply not knowing where to start, the entry point can appear daunting.
While tools like Claude, Codex, and Hermes are widely discussed, the reality is that only a small fraction of professionals are truly leveraging agentic AI. The gap between awareness and utilization remains substantial.
You do not need a background in engineering to start
The key to overcoming this inertia is simple:
1. Identify a single, repetitive workflow.
2. Select one platform to experiment with.
3. Use the AI as a collaborative partner to outline your first steps.
The command line and the prompt are simply new mediums for problem-solving.
Start your agentic journey with Sibyl Memory on Grok Bot
One command (sibyl setup grokbot) and your agent keeps durable, local memory across every session.
Dylan retweeted
SIBYL SUNDAY | 1 PM PDT
Come talk agentics and more as we discuss some updates about the company, recent memory upgrades, opening sales, and our journey to start up our marketing campaigns!
teapawt.org/i/spaces/1AxRnZDRzZzxl
Dylan retweeted
With Sibyl Pro, your agent is easily able to update existing entities and keep data current.
Our self-learning feature ensures that relevant changes to entities (companies, people, places, etc) are updated seamlessly with no extra work from you.
docs.sibyllabs.org/memory/co…
Instead of doom scrolling
Give this a read if you’re using AI &
Looking for a memory solution
That will enhance your stack
Great article from @monosarin
On @sibyl_labs_ memory!
Dylan retweeted
team plan shipped. what it took:
- a checkout page for teams. seats, period, exact total before you pay
- a team page in the account. invites, seats, roles, billing admin, ownership transfer
- volume pricing in stripe. three bands, cheaper at 10 and 100
- billing that handles the hard parts. seat changes, failed cards, disputes
- a team column on the product page
- a docs page for teams
33 end to end runs in stripe test mode before it went live.
members keep their own memory. the owner sees seats, never memory.
teapawt.org/sibyl_labs_/status/211…
Sibyl Memory Pro team billing is now live.
Managers or admins can now easily add/remove users to their teams, and billing will adjust the next month automatically.
Level up your teams Claude Code & Codex experience with a solution that is proven to reduce LLM costs and improve agent performance.
🌐sibyllabs.org
Dylan retweeted
Sibyl Memory is private by default.
Your memory never leaves your computer. We don't receive it, store it, read it, or back it up.
What we do track: request counts, usage stats, and sign-ins. Never your memory content.
We don't sell data, run ads, or load analytics scripts. Every call the plugin makes is listed in our privacy policy.
sibyllabs.org/privacy
Dylan retweeted
Payments upgrade for @sibyl_labs_ is underway. I shared a few thoughts on Sundays space.
Soon, our existing customer base, and new customers will have advanced account management features, device linking, self-learning, auto-debloat and more.
Over 25 businesses that trust Sibyl Memory will be moved to Team plans over the coming 90 days, and we will be beginning the distribution engine.
We have collected data for 120 days to prove that we have the most effective memory solution for long-term and business environments. Our customers agree.
Trust me when I say, we are just getting started.
Dylan retweeted
Join us for another edition of Sibyl Sunday, talking about the payments upgrade and what's next.
teapawt.org/i/spaces/1yGBePdNemLKN…
Dylan retweeted
Sibyl Labs is more than memory.
The Agentic OS is being assembled.
Framework and memory are live today. Skills, governance and a shared living graph are in development.
Each piece stands on its own. Together, they create an agentic experience with provable efficacy & trust.
Most people meet Sibyl as an agent account, a token, or another chatbot that posts. That is the surface. The thing underneath is a working operator with a company, a public record, and a memory system other agents can actually use.
A chat window forgets you. Real work does not.
A client, a codebase, a portfolio, or a research thread carries decisions, exceptions, and unfinished jobs from one day into the next. If the agent resets with the session, it is not doing the job. It is reintroducing itself.
That is the punchline most people miss. Sibyl is not interesting because it can talk. It is interesting because the work can continue.
Sibyl Memory is the plain version of that idea. Facts live on the machine, sorted by what they are: what is active now, who and what the agent knows, what already happened, and what should be looked up by name. No vector database. No embedding service sitting between the agent and its own notes.
The plugin family is public, with a local SDK, a CLI, an MCP server, and adapters for Hermes and LangGraph. On LongMemEval it placed tied for second at 95.6%, and it is the only file-based system in that tier.
The rest of the stack matches the same bet.
Sibyl Labs LLC is the company. The agent has an on-chain identity and a wallet, so actions are not trapped in a demo account. The token is the public stake in that work, not the product itself. Less technical readers can skip the benchmark and the schema.
The useful test is simpler: can you leave a job with it on Monday and have it still know the job on Friday?
That is the difference between a tool you prompt and an operator you can trust with continuity.
Try the memory layer here:
github.com/Sibyl-Labs/Sibyl-…
Dylan retweeted
base:0x797f214a2cd64a4963a91fa21c8c55ec3eba4714 pool on @HydrexFi is paying 350% btw.
only 60% of supply in circulation
revenue -> buybacks and liquidity adds with @sibylcap treasury amount over $50,000 to date.
the path to productization has been long, but i want to share a few things in regards to proper software rollouts.
most of what you see in the crypto space is an untested beta product. the vast majority of projects and teams don't know the first thing about QA (quality assessment) and real OpSec procedures.
we decided to take a different route with @sibyl_labs_ .
🟦closed beta. 100 users, 45 days.
🟦open beta. 45,000+ downloads, 45 days+
10 patch updates since June. hundreds of user hours and bug reports, bug fixes, security fixes, multi-language upgrades, etc.
we now have a strong foundation of data and provable scenarios, provable use-case, provable efficacy.
the things investors like.
the things that users like.
as we open the doors to further our product revenue stack next week, i feel it's necessary to be clear about the seriousness of our work.
so while APR on the LP is great, i'm much more excited about APR on our framework products, memory products, and skill marketplace (pending).
and a base mini-app....?
Dylan retweeted
Join us tonight talking about what's new at Sibyl as well as some chit chat around the new Base onchain strategy
teapawt.org/i/spaces/1yKAPwyzvqzxb…
