In the pursuit of optimizing operational efficiency, internal chatbots have emerged as formidable allies, reshaping the way organizations communicate and collaborate. These conversational applications, tailored to assist employees, are pivotal in addressing FAQs, completing tasks, and facilitating seamless access to organization-specific information.

The Merits of AI-Driven Chatbots

Integrating an AI chatbot, finely tuned to internal documents, unfolds a myriad of advantages:

• Efficiency Amplification: Rapid responses to common questions and access to organization-specific information expedite decision-making and trim down the time spent searching for essential data.
• Streamlined Workflows: A byproduct of efficient internal communication, these chatbots contribute to streamlined workflows, enhancing overall productivity.
• Customer Service Excellence: Beyond internal use, chatbots find applicability in customer service settings, ensuring immediate responses to customer queries, and thereby boosting satisfaction and loyalty.
• Data Fortress: With an innate focus on security, internal chatbots fortify data protection, ensuring that sensitive information is accessible only to authorized personnel, averting risks associated with unauthorized access.

Selecting the Optimal Path

When navigating the landscape of internal chatbots, organizations are presented with two principal pathways:

Option 1: Leveraging 3rd-Party Tools

As you explore the landscape of internal chatbots, let’s shed light on some third-party tools that have garnered our attention at Capria. Each offers unique features, catering to diverse organizational needs:

1. Dropbox Dash: Venture beyond conventional file searches with Dropbox Dash, an AI-powered universal search tool. This innovative tool has a built-in chatbot feature, allowing you to pose questions about your files. The chatbot’s proficiency in answering queries is still in its early stages. We are closely monitoring its development to gauge improvements to this nascent feature.
2. Zapier: You can create a no-code chatbot using Zapier that is built off of internal file sets. This requires a little bit of a learning curve to be able to effectively use Zapier but it is still a great no-code application with other automation functions Don’t miss our Tool Spotlight for an in-depth exploration of this powerful tool!
3. Glean: Glean is a notable option for businesses seeking enhanced productivity through data intelligence. This platform excels in accessing, querying, and activating first-party data, offering insights derived from various sources like Slack, Asana, G-Suite products, and more. However, it’s essential to note that Glean leans towards higher costs, making it a recommendable choice primarily for larger organizations with expansive budgets.

Pros:

• User-Friendly Solutions: No-code interfaces that demand minimal technical expertise.
• Seamless Integration: Effortless integration with popular SaaS tools like Slack, Google Drive, or Dropbox.

Cons:

• Static Document Limitation: Restrained to static documents manually uploaded, lacking synchronization with live files.
• Cost Challenges: Elevated ongoing costs associated with subscription or usage fees.

Option 2: In-House Customization

For organizations equipped with internal technical proficiency, the choice to construct a bespoke chatbot using language models such as GPT-4 or Claude opens doors to unparalleled customization. This pathway enables the integration of specific features meticulously tailored to the organization’s unique needs, providing control and flexibility that aligns seamlessly with internal requirements.

Pros:

• Tailored Customization: Unmatched flexibility and customization, catering to the organization’s specific needs.
• Cost-Effectiveness: A potentially more cost-effective option for organizations with in-house technical expertise.

Cons:

• Technical Proficiency Required: Development and maintenance of a custom chatbot demand technical capabilities.

Capria’s Decision

We decided to work with our GenAI Fellow to build a custom GenAI chatbot built on GPT-4 and synced with our Google Drive files. We affectionately named it GUS (short for “Get Us Smarter”). The decision was driven by the desire to learn state-of-the-art technology using a vector database and an LLM in a “RAG” architecture, the ability to hand the reference implementation to our portfolio company partners so they could adapt for their own needs, the cost (overall cost of building off GPT and hosting on AWS was relatively cheaper than any of the 3rd party tools) and by accuracy (the chatbots we tested didn’t produce as accurate responses and with the internal chatbot we were able to make tweaks such as syncing to our live fileset).  However, we’re continuously comparing GUS against new or improved 3rd party tools, which are becoming higher quality at lower costs.