{"sections":[{"title":"OdeDevlog01Section1","heading":"01-Selling an engine when they order a car","verbatim":false,"ast":[{"t":"para","c":[{"t":"text","s":"Cloudnine is a lean start-up with some killer IP: An associative data paradigm that procedurally and automatically, rather than manually or via inference workloads, creates contextual webs from the content and taxonomy of ingested data. Part of my responsibilities with them was to bring this data paradigm \"associative data\" to the market: it had been used in bespoke enterprise soilutions, but hadn't yet reached wider market adoption."}]},{"t":"para","c":[{"t":"text","s":"Up until that point, it had been used inside closed systems for intelligence, monitoring, auditing and asset tracking. With a lean startup team, delivering on consumer-grade software at scale that captured all of the bespoke features as a full stand-alone solution was a big ask: although there was feature overlap between a generalized version, so much of the output featuring was wrapped up in proprietary vendor-specific configurations."}]},{"t":"para","c":[{"t":"text","s":"Many customers are pretty locked in to their data and intelligence stacks and don't want to change their vendors or tools. Adding complexity via new solutions to their operations, especially at enterprise-scale, was a daunting prospect for large customers, even if their problems were painfully expensive and resource consuming to leave unaddressed.. "}]},{"t":"para","c":[{"t":"text","s":"Additionally existing vendors had great solutions in isolation. Matching punch-for-punch per vertical was a pretty ambitious ask for a team of our scale."}]},{"t":"para","c":[{"t":"text","s":"So rather than focusing on tackling one or two verticals well, I wondered if we could instead change the axis and value prop. Rather than verticals we'd go horizontal: data, context, applications: virtual associative context engine. "}]},{"t":"para","c":[{"t":"text","s":"We could serve developers with a strong dev tool kit, MCP server, and flexible API that brought associative to the users most likely to take advantage of it. Help solve big problems AI agents are having, and make a unified data-layer that associative self-manages so that tools, apps, and more could be built reliably and easily."}]}]},{"title":"OdeDevlog01Section2","heading":"02-When your audience includes humans, ","verbatim":false,"ast":[{"t":"para","c":[{"t":"text","s":"I worked closely with our CTO and the engineering team to thread the needle and focus on the killer features needed to demonstrate the value of associative paradigm for builders as quickly as possible: trust, flexibility, and capability. "}]},{"t":"para","c":[{"t":"text","s":"The core technology allows for some pretty neat data operations: blind data-discovery, analysis and operations over fully anonymized data, really interesting provenance toolings, and the auto-generating contextual links. Packaging what kind data operations and how they get employed took some work and feedback from early testers."}]},{"t":"para","c":[{"t":"text","s":"We knew that the data consumers were going blended: agents, apps, policies, and people. All needing different entry points to succeed. We wanted the API to be total for developers, giving them the maximum flexibility and performance while keeping complexity at a minimum. Then it was time for the agents."}]},{"t":"para","c":[{"t":"text","s":"We already had a technical demo we had built for enterprise customers with a UI and preset data to showcase some of the value-add of associative. Builders however, want to test by getting their hands dirty. At this time also it was the genesis of the first batch of AI agents and harnesses. We knew that if we wanted to have builders onboard, we'd have to have interoperability between them and their agents. Within a few days we had our prototype tested and live."}]},{"t":"para","c":[{"t":"link","label":"Next Devlog","target":"ode-devlog-02","kind":"node"}]}]}],"dialogue":null,"embed":null}