Foreshock is just getting started finding ways to apply the telemetry and\nintelligence hooks that come with providing Aldous warm as-a-service for clients. We've got everything from "privacy-respecting Emotional GoPro-like day summarizer" to Social Seismographs on the near-term horizon.
\nBut that means our model is only ever going to grow into our vision of it. How\nmyopic of a vision is that? We can't let that happen. We need people that spot more\nexact pairings between what we can provide and interesting problems that need solved.
\nPartners keep us on the edge of a very important intersection: what we choose to\npursue, and how that pursuit can help people add value to their lives and\nbusinesses.
\nWe want a constant stream of people trying to hammer our research into corners\nthat we never thought it would go, or even existed. And you know what? We're\npretty good friends to have!
\n\nLet's make sure we're clear up-front because our structure might suggest otherwise: we are capitalists. We intend to build a highly sustainable, profitable engine. But how we use that money, and why we are in this game, is drastically different than the standard venture-backed playbook.
\nOur organizational structure is coalescing around a model that hasn't quite been tried before; one designed specifically to prevent the institutional decay that eventually forces platforms to betray their users by 'firing' the vision that blocked them.
\nWe are still finalizing the exact corporate registration details to ensure the legal mold perfectly fits our actual growth, but the core tenet is already locked in: our partners are never treated as mere vendors or extractive revenue streams, not even on a constitutional level.
\nWe are changing the rules of the game so we can win on our own terms, and we are looking for like-minded people who want to build in that same reality.
\nAccess to our engineers through Slack/Discord (not standard public channels)
\nHosting on our infrastructure, as-appropriate
\nCo-marketing and Co-branding opportunities
\nParticipation roles in research
\nRespect, admiration and appreciation
\nPersonalized, detailed responses with our decision, criteria and information\nconsidered, reasons for the decision and next steps (including appeal if\nappropriate). We don't automate it.
\nNon-disclosure or non-compete agreements. If you spill your own beans, it's on\nyou; don't let us stop you. We only publish collaboratively with you.
\nSurprises. Each initiative has its own terms, agreed in advance. If you show\nup, we'll show up. We don't have "Developer program terms" because this isn't\na mass-product, it's a relationship we intend to participate personally in.
\nAn appropriate need for our products. The problem type has to fit. Magnets\nwon't attract wood, for instance, so there's little point in pretending it\nmight work unless you're bringing something new.
\nA willingness to be wrong as well as brilliant. Many iterations fail for many\nreasons, some turn out to be premise-shattering. This isn't the first ever\napplication of this technique, but it's the first attempt at a defined\ninterface for it.
\nA potential for growth. There has to be a business gain in here, somewhere. We\nlook ato many factors when it comes to potential, including how it could add\nto other projects.
\nWe need to know some base information about how you envision us adding value to\nreaching your goals. Fill out this short survey and someone from\nEngineering or our Business execution team will look at your application.
\nYour application is first evaluated by our business end to make sure there's\ncurrently room and resources to support you. It's dynamic, based on our\nbandwidth and number of engineers actively contributing on a set schedule.
\nMany applications will be a great fit at the wrong time, but that doesn't always\nmean it will be politely declined. We may be able to offer you infrastructure\nand some code, or a variation of what you propose. We'll work to say yes.
\nAnd if we can't, we'll politely decline for now, and honestly keep in touch\nwith the goal of trying again when circumstances better align.
\nYou can't violate or conspire to violate US law with our products or services,\nobviously.
\nYour use must withstand ethical scrutiny if we're involved without technical rules-lawyering. This is non-negotiable. We do not discriminate against fields of endeavor but we do draw a line at harm.
\nWe have a code of conduct that relates to how we interact together. We bring\none, but we mostly just care that we use one and work from a few agreed-upon\nexpectations in how we interact.
\nDeference isn't automatic; respect is a phenomenological variable. We invest our attention in actions and outcomes; in our own as well as others. We respect the body of work that credentials and certifications represent, but on missions, it's what you've done that gets the microphone the most because it's what we can actually use.
\nIf you're a good fit for this program, the above bullets should feel more like assurances than obstacles. If that's the case, reach out to us!
\n","tags":["Semantics","Aldous","Product"],"date_published":"2026-07-12T00:00:00.000Z"},{"id":"https://foreshock.io/blog/posts/2026-07-01-splinter-and-scale/","url":"https://foreshock.io/blog/posts/2026-07-01-splinter-and-scale/","title":"Aldous Research Preview Has Been Released!","language":"en","content_html":"I'm pleased to announce that I shared Aldous'\nlaunch announcement:
\n\n\n\nThe article below introduces Aldous, a zero-shot semantic telemetry engine\nbuilt to do exactly that. It operates on Diagonal Emotional Covariance\nEstimation (DECE), calculating the centroid and dimensional variance of\nconcepts in latent space to output distance and similarity floats in constant\ntime.
\nThere is no string matching and no generative reasoning, just open\nobservability. If you are working on the math or governance side of AI, I\nencourage you to read the architecture breakdown. My approach isn't exactly\nnew, but it is pretty novel, if I do say so myself because it's #Semantic,\nboogie woogie, woogie!
\nPlease consider dropping the open-source variance harness into your own CI\nserver to establish your own content baselines, and please subscribe to my new\nnewsletter that tracks the journey of this ongoing, rewarding, and often\nbeguiling research. I won't publish an installment unless there's plenty to\nsay.
\n
I really suggest that you also check out the\narticle that I mention.
\n","tags":["Aldous","Product","Trust & Safety","Releases"],"date_published":"2026-07-01T00:00:00.000Z"}]}