What is Pocket Mentor?
Pocket Mentor is an enterprise AI leadership and performance mentoring platform created by founder Daniel Pratt, co-founder Matthew Spangher, and CEO Kim Keedle. Designed to democratize high-performance coaching, Pocket Mentor delivers personalized, context-aware leadership guidance directly to users through an intelligent mobile interface. After completing an initial concept phase with a startup accelerator, the founding team engaged Techne HQ to execute the full technical design and engineering build. Techne HQ collaborated directly with Daniel Pratt, Matthew Spangher, and Kim Keedle to overhaul all prompt engineering and expand initial wireframes into a 50 to 70-page UX/UI system. While the initial v1.0.0 prototype utilized OpenAI as the backbone LLM, User Acceptance Testing (UAT) revealed that standard out-of-the-box model outputs fell short of their rigorous coaching standards. Techne HQ upgraded the platform by architecting a custom Retrieval-Augmented Generation (RAG) pipeline, integrating AWS Bedrock, and building a domain-specific knowledgebase on AWS infrastructure. When incoming CTO Craig Sheen joined the team, Techne HQ executed a flawless technical handover, enabling Pocket Mentor's internal team to take over immediately with minimal ongoing support required.
Upgrading Initial LLM Prototypes to Enterprise-Grade RAG Architecture
Delivering high-stakes executive coaching through AI requires absolute precision, strict domain boundaries, and zero hallucination. To make Pocket Mentor a market-ready commercial product, Techne HQ worked directly with Daniel Pratt, Matthew Spangher, and Kim Keedle across three core technical hurdles:
Redesigning Prompts & Expanding UX/UI Workflows: Executing a 4-week design sprint to author PRDs, conduct API feasibility studies, and build 50 to 70 pages of missing UX/UI wireframes in Figma.
Architecting a Custom RAG Pipeline on AWS Bedrock: Replacing basic OpenAI API calls with a RAG architecture that indexes custom leadership knowledgebases, utilizing AWS Bedrock for context-aware response generation during UAT.
Building for Clean Internal CTO Handover: Maintaining pristine GitHub commit hygiene, modular "Lego block" architecture, and complete technical documentation so an incoming internal engineering leader could take full ownership instantly.
Enterprise AI Engineering & Clean In-House Handover
01 · 4-Week Design Sprint, Prompt Architecture & UX/UI Expansion
Techne HQ ran structured strategy workshops with Daniel Pratt, Matthew Spangher, and Kim Keedle. We designed 50 to 70 custom UX/UI screens in Figma, conducted thorough API feasibility studies, and defined system logic across all mentee and admin interfaces prior to development.
02 · Upgrading from OpenAI LLM to Custom RAG & AWS Bedrock
When initial UAT testing demonstrated that basic LLM responses lacked the depth required for professional mentoring, Techne HQ upgraded the entire system. We engineered a custom Retrieval-Augmented Generation (RAG) pipeline backed by vector search and custom knowledgebases, deploying model orchestration through AWS Bedrock to deliver highly tailored, authoritative coaching outputs.
03 · Secure AWS Cloud Deployment & Modular Systems
We deployed Pocket Mentor's backend on AWS Cloud Infrastructure utilizing clean service separation, role-based access control, and AWS Secrets Manager for enterprise API credential security. This ensured low-latency response times and bank-level data isolation for user conversations.
04 · Frictionless CTO Handover to Craig Sheen
From the first day of engagement, the leadership team was transparent about bringing engineering leadership in-house. Techne HQ maintained strict git hygiene, well-commented repositories, and detailed PRDs throughout. When CTO Craig Sheen joined Pocket Mentor, Techne HQ executed a seamless code and repository handover. Due to the clean architectural standards, Craig and his internal team took over the codebase instantly without friction.
4-Week Intensive Design, Prompt Architecture & 50-70 Page UX Expansion
Custom RAG (Retrieval-Augmented Generation) Pipeline Built with Custom Knowledgebases
AWS Bedrock Infrastructure Integration to Overcome OpenAI Output Constraints during UAT
Modular Cloud Infrastructure Deployed on AWS for High Accuracy & Low Latency
100% Clean CTO Handover Executed for Incoming CTO Craig Sheen with Immediate Codebase Adoption
Product Features
Talk to a Technical Partner Before You Build.
Great products start with the right decisions early.
Discuss your idea with Gordon, understand the technical risks, and create a clearer path from concept to market.
Tech stack
Pocket Mentor's AI architecture combines AWS Bedrock, custom Retrieval-Augmented Generation (RAG) pipelines, vector search, and domain-specific knowledgebases to deliver accurate, context-aware leadership mentoring. Built on AWS Cloud Infrastructure with modular backend services, role-based access control, and AWS Secrets Manager, the platform supports secure, low-latency AI interactions and seamless in-house engineering handover.
MORE Work
Evaluating an Accelerator or Looking to Bring an Idea to Life Fast?
Whether you're considering a startup accelerator or starting with a raw concept, you don't need to spend months collecting static design files or sitting through generic workshops. Techne HQ combines senior product strategists and full-stack software architects under one roof. We work directly with founders to take you from Ideation, Strategy, Validation, and Prototyping, all the way through to Full-Stack Engineering, Market Launch, and Ongoing Scale.
By eliminating middle management and agency bloat, we take you from an initial concept to a live, market-ready product faster than traditional agencies or slow-moving accelerator programs, ensuring your startup launches with real code, battle-tested logic, and scalable infrastructure from day one.
Talk to a Technical Partner Before You Build
FAQ
Got specific questions?
Why was a custom RAG (Retrieval-Augmented Generation) pipeline necessary for Pocket Mentor?
Out-of-the-box LLM APIs like generic OpenAI models often produce generalized or unverified responses. Techne HQ engineered a custom RAG pipeline using AWS Bedrock and specialized knowledgebases to ground AI responses in verified leadership methodologies, eliminating hallucinations and elevating output depth.
How does Techne HQ prepare AI applications for a future internal CTO handover?
We build with long-term handover in mind from day one. By maintaining clean GitHub commit histories, modular "Lego block" code structures, and comprehensive PRD documentation, incoming CTOs (such as Craig Sheen at Pocket Mentor) can adopt the codebase effortlessly without inheriting technical debt.
What technologies power Pocket Mentor's AI architecture?
Pocket Mentor utilizes AWS Bedrock for model orchestration, custom vector databases for RAG retrieval, AWS Cloud Infrastructure for backend hosting, and modular frontend mobile interfaces.
How quickly can Techne HQ transform accelerator wireframes into an AI-powered platform?
Techne HQ typically executes intensive 4 to 8-week design and architecture sprints (such as the 4-week sprint completed for Pocket Mentor) to expand incomplete wireframes into 50+ production UX/UI screens, author PRDs, conduct API feasibility studies, and deploy live AI pipelines.