Sample project overview for retrieval-augmented AI chatbot

Sample Statement of Work: AI-Powered Chatbot for Executive Coaching

Overview:

This custom chatbot will include the content from a series of relevant strategy books, plus customization for the executive’s company in general and the executive in particular. The chatbot will provide the executive and her team with access to all relevant executive coaching and L&D content--in an interactive dialogical format. This will provide and always available, interactive method for the executive and her team to interact with their ongoing coaching, strategy, and L&D content.

Objectives:

  1. Create an interactive text chat-based format for the client to engage with their L&D content, coaching guidance and strategic research.
  2. Build the chatbot in such a way that new information can be easily added to it in the future.
  3. Deliver the text chat-based chatbot in a neat and easy-to-use browser-based frontend.

Scope of Work:

Phase 1: Prototype creation (approximately 2 weeks)

  1. Collect information including:
    1. Books from coaches and advisors from the executive and team’s sessions, transcripts of relevant conversations, and strategic research.
  2. Process the materials for use in the chatbot.
  3. Deliver a working version of the chatbot with basic functionality.

Phase 2: Refinement (4-6 weeks, depending on rate of client feedback)

  1. Initial user testing with team members who are familiar with the source materials (the books and custom forecasts) for accuracy and relevancy of responses.
  2. Secondary normal-usage testing with friendly testers who are not familiar with the source material for relevancy of responses.
  3. Red-teaming with friendly users to prevent undesirable responses.
  4. Polishing frontend with neat, clean, usable design.

Phase 3: End-client delivery

  1. Presentation to end-client on functionality of bot, expectations, and limitations.
  2. Iteration with the end-client team to iron out any last bugs, glitches, or unexpected behavior.

Phase 4: Support (for up to 1 year after handoff of final deliverable)

  1. This project includes 1 year of support for project as-delivered.
    1. In plain language, this means if something we made breaks, we will fix it without additional charge.
    2. This does not include adding additional functionality, or making stylistic changes. These would be scoped separately.

Deliverables:

  1. The chatbot will be delivered in a browser-accessible URL. The chatbot will be primarily built for use on a laptop screen, but will also be usable on mobile.
  2. The code for the chatbot will be made accessible in a Github repo. An employee of the company will be given admin rights on the repo.

Pricing:

Between $20,000 and $150,000, depending on exact specifications.

About Handshake:

Handshake is a resilient transformation consultancy. We identify, envision, design and train at the edge of emerging technology and societal shifts. We help our clients and partners thrive in a volatile world and build adaptive solutions to big problems. We’ve worked with leaders and teams in Fortune 500s, governments, NGOs, and think-tanks who have a mandate to look beyond the day-to-day and bring something new to their organization. Past projects have included AI tools for a management consultancy, AI generative code tools for an org within the US Airforce, technical development and support for EdTech in Northeast Nigeria.

The Handshake dev team has led machine learning and data science projects in both large organizations (e.g., GAP, Uber) and startup environments (telepath.io, calltime.ai, underline.com); and cross-disciplinary technology research and prototyping at universities, corps and others including UCSF, Johns Hopkins, Consensys, Institute for the Future and others. We’ve worked with large teams providing high-level advisement, and led data science projects with startups building data processing tools and ML systems from day 0.