We have heavily invested in building capabilities in LLMs to deliver immediate business value to our clients; we’ve built POCs to vet models, stacks and tools, created a one hour crash course to help demystify the space, and delivered real, working LLM-based software to unlock business value for our clients.
Based on this real world experience, we believe LLMs will transform the way most companies do business, and the way everyone builds software.
This will likely impact software development at its core - the iron triangle of cost, scope and quality will shift.
Finally, we believe an AI fluency gap is holding back a tremendous amount of innovation inside companies, who simply need time and experimentation to identify high value opportunities to incorporate LLMs into their workflows. Stride is here to help – the first steps are the hardest, and we have helped clients jumpstart their learning process with Lightning Talks, workshops and custom LLM sandboxes.
We’ve been busy developing this practical and pragmatic approach to using LLMs to generate immediate business value for our clients. Based on our learnings from our client work, I want to share a bit more about how we’re thinking about the business value LLMs can bring, and what’s next. In case you missed it, we delivered an LLM webinar last week which describes our learnings and approach in a lot of detail.
To unlock this broad transformation, we believe we first need to invest in education.
To that end, we’ve published a free, one hour LLM Crash Course and are delivering custom Lightning Talks and AMAs with our AI/LLM experts to help narrow the knowledge gap.
Beyond education, we believe there are two main things we can do to accelerate adoption:
1) help get LLMs into as many hands as possible in a safe and secure manner, and;
2) harness LLMs to accelerate software delivery.
Here are two great examples of custom LLM-powered software we’ve delivered lately:
VPC-hosted secure LLaMa 2 For biometrics and identification company
Our client had a bottleneck problem. The company was scaling and critical services were getting bombarded with requests for information. They needed a way to scale the knowledge in this critical time, while scaling the organization and meeting growing customer demand. The custom LLM had to be secure and fully owned.
Stride built an LLM around the customer’s knowledge base, code repos, technical documentation and business collateral as a first line of defense so that other teams and new hires could get answers to work-blocking questions in real time, and, without interfering with the work of the critical path team.
Stride selected and delivered the initial version of the LLM design for enhanced user experience and a road map for refinement in under 6 weeks. This was just in time for an upcoming hackathon they were attending.
OpenAI-powered search/knowledge engine for Edtech
Our customer has a cutting edge, integrated STEM-based approach to learning that is quite different from the incumbent Common Core models. This creates a unique challenge: how to help a growing education staff - trained on the old model - build and deliver lessons to students with this new methodology?
Stride built an LLM-based search tool that allows educators to search for key concepts in the traditional Common Core model and receive lesson plans and collateral for a new way of working.
Leveraging Open-AI’s Embedding API and Pinecone, each search deliver’s results for Common Core questions in the new models vernacular, thereby reducing the maintenance and upkeep required for a traditional syntactical search database.
Conclusion
We built these LLM-powered apps to deliver business value, and to learn by doing – and to share our learnings with our client partners, and help their teams develop fluency in AI. We believe this fluency gap is holding back a tremendous amount of innovation inside companies, who simply need time and experimentation to identify high value opportunities to incorporate LLMs into their workflows.
Stride has always cared about the craft, about teaching and coaching. We strive to improve the conditions for building great software, leading to happier and healthier teams that deliver better outcomes.
Francisco Martin
