Founder @useliaison
San Francisco / San Juan
Joined April 2022
- Tweets1.6K
- Following1.2K
- Followers400
- Likes848
A lot of the software a business runs on has no API, and no amount of prompting fixes that. A machine has to sit in front of it like a person would.
The longer an agent runs without anyone watching, the more it matters that a person can read what it did afterward.
Wait till people start to understand what adjusted EBITDA looks like - 1 time liabilities. Financial engineering perceived as generosity. Don’t get me wrong , great for those who are laid off but not detrimental from a IS standpoint
Half the wrong answers we've fixed came from the agent reading the last message and nothing above it.
Calling BS on this unless they have a model without the distillation and model training guardrails as the rest of us peasants. Coordinated paid ad.
CTO of Legora has Fable one-shot direct write a 100k param LLM.
Note that unlike a trained/grown model, the designed on only uses 4k params, zeroing the rest.
This indicates there’s a readable/writable logical structure that can be reasoned about for this task.
And Fable knows it. Which means that if you had access to the reasoning traces you could see how Fable assembled this logical machine.
Every answer Liaison gives depends on the data underneath it.
We're hiring a Database Engineer. Hybrid, Salt Lake City. Production Postgres, ingestion pipelines from outside APIs, and the record matching that keeps one company one record.
useliaison.com/careers/datab…
Tokens are the number on the invoice. The number I actually want is how much work it takes to get one answer right.
You cannot fundamentally sell proprietary llm intelligence - and simultaneously sell outcomes.
The big 4 will fail at selling outcomes - or if they do it is most likely not in the best interest of their customers because they have no incentive to lower costs, and 2 their models might not be the best at solving that outcome.
When the "big 4" of AI are no longer willing to operate at massive deficits, your access to their agent platforms will be reduced (or become mind-numbingly expensive).
We're built different. With Liaison's novel harness engine, we're able to maintop top level accuracy, performance and experience without the massive token bloat that they carry.
Cost sustainability has been a fundamental aspect of our infrastructure, making sure we can provide our users with an experience that will not "fall-back" or degrade like you're seeing across the board here.
100% only the best
If you need training,
Do not join an early stage start-up
If you need help setting up your computer,
Do not join an early stage start-up
If you want to spend your first month ‘getting up to speed’,
Do not join an early stage start-up
If you want max out your time off,
Do not join an early stage start-up
If you can’t ship in week one,
Do not join an early stage start-up
If you can’t sell without a brand behind you,
Do not join an early stage start-up
If you need kudos for just doing the basics,
Do not join an early stage start-up
We should call it SaaS, what do you think @KenTheRogers ?
Calling it now.
AI is going to usher in a whole new era of software development. I’m calling it Software as a Service.
People will want to use these tools but they’ll realize they don’t want to spend the time to maintain and update them over time.
This will create a whole new category of businesses and entrepreneurs that can take on that task for businesses that would rather work on their core products.
AI is changing everything. This is a whole new paradigm of thinking about the way we interact with software.
Never click install update for at least the first two weeks. Many of these updates / vulnerabilities come quick and fast
We identified a security incident affecting a modified Claude Code VS Code extension accessed through our Cursor IDE extension marketplace, impacting at most 5,000 users. This malicious version of the extension was not developed by Anthropic. It does not impact other extensions downloaded from the cursor.com/marketplace, Grok Bot, Grok or our other SpaceXSI products.
In response, we took down our extension marketplace this afternoon, worked with relevant providers to neutralize the malware, and have reinstated the marketplace after confirming there is no further risk.
If affected, please review instructions in this blog post to understand how to find the malicious package and establish a path to mitigate any potential impact. We are asking all customers update to Claude Code for VS Code version 2.1.314: cursor.com/blog/claude-code-…
Wow
Massively underreported: an official Claude Code extension was compromised and installed backdoors on developer machines.
Could it be behind the Binance hack?
Readers added context they thought people might want to know
The post misstates the facts: a tampered copy of the Claude Code extension was published briefly to Cursor's marketplace only and was not developed by Anthropic; official channels and distributions were unaffected.
cursor.com/blog/claude-co…
x.com/cursor_ai/stat…
x.com/felixrieseberg…
Most AI products die on the data, not the model.
We're hiring a Senior Database Engineer to own ours. Ingestion from outside APIs, Postgres at scale, tenant isolation, identity resolution.
Hybrid, Salt Lake City.
useliaison.com/careers/senio…
CEO Alex Karp: AI labs supplement the cost of tokens so they can steal your alpha "The reason tokens are discounted is not so they can grow the market.”
“They're discounted so that they can get access to your IP so that their models get better. And there was functionally no honest conversation about this or disclosure."
"Our partners are livid. The kind of safety they're most upset about is what they believe is the theft of their alpha."
we know where the puck is going. you wont want an ai assistant to get you there. you'll want .. well, YOU. we launched Liaison to do just that. Digital version of you. In that role at work, personal life, whatever it is. Its you (digital you).
AI agents mostly suck. They're good at doing one specific thing, when you ask it...but 90% of the time you're tinkering or fixing or yelling at it in ALL CAPS.
The goal of an agent was always, "how do I make myself more efficient and effective", and all the agent's I've seen haven't done that yet.
@iamcadec and I had enough, and launched @useliaison
The TL;DR is: AI Agents that are actually trying to become a second version of you. They know your context, they know when to pick up your work and when to ask you about it, and they learn with every one of your actions.
No more ad-hoc agent requests to gmail to try and find your last message. No more "you're absolutely right, I should have searched that". The agent lives in the data, and makes the decisions that you would have made on the information that you didn't have time to read.
Check it out at useliaison.com or shoot me a DM. I'd love to get everyone up and using the platform on a trial.
