CASE STUDY

Atomic Object leverages AI through TypingMind

Learn how Atomic Object unlocks the full potential of AI with TypingMind, where innovative and seasoned consultants double productivity with cutting-edge tools and features.

80%

members adopt AI(TypingMind is the top-used AI tool)

AI models used

9k+

messages sent

Drew Colthorp - Atomic Object — Drew Colthorp
Software Development Practice Lead at Atomic Object

Overview

In this interview, we have Drew, Software Development Practice Lead of Atomic Object, who will share insights into how his company leverages TypingMind's LLMs to improve their consulting work and software development.

Let's dive into the conversation to learn more about their journey and experiences with TypingMind!

Belle• TypingMind

Welcome, Drew. It's a pleasure to have you with us today. To kick things off, could you give us an overview of what Atomic Object does?

Drew• Atomic Object

Hi Belle, it’s great to be here.

Atomic Object is a full-stack software development consultancy focusing on product development. We help clients craft software solutions from research and validation, through design and development. Formed in 2001 in Grand Rapids, Michigan, USA, we now have more than 100 Atoms across four offices in Grand Rapids; Ann Arbor, MI; Raleigh, North Carolina; and Chicago.

Atomic Object helps companies innovate and grow with custom software that’s beautiful, reliable, and easy to use. We’ll work with you to mitigate risk, make smart trade-offs, and get the maximum value for your budget.

Belle• TypingMind

So, why did you decide to adopt AI for your business, and what made you choose to start with TypingMind?

Drew• Atomic Object

As a consultancy, we’re always looking for ways to help our team to deliver better client results more cost-effectively. Generative AI has real promise to help streamline a number of aspects of how we work, from design to artifact creation, to coding. With models evolving so quickly right now, we wanted a solution that enables experimentation with different models and rapid adoption of the best tool for the job as the landscape shifts.

I gave a talk recently about a Harvard Business School study looking at the impact of GPT-4 use on consulting - here’s a link to the YouTube video.

TypingMind has been a part of our strategy of lowering the barrier to increasing familiarity with using LLMs for consulting work and software development.

Atomic Object is utilizing TypingMind to use multiple LLMs for consulting work and software development

Use Cases for TypingMind

❝ As individual consultants get more experienced and creative in their use of AI to accomplish tasks, a number of capabilities built into TypingMind really start to shine.❞

Drew Colthorp

Software Development Practice Lead at Atomic Object

Belle• TypingMind

I watched your talk on the impact of AI on consulting, and it was fascinating to see the insights and approaches you recommend for effective AI integration.

With this in mind, how did you implement TypingMind within your team?

Drew• Atomic Object

TypingMind has given us a great, consistent UI for LLM use across the various models, including OpenAI, Anthropic, Google, Meta, Mistral, and Perplexity. The tools built into TypingMind including AI Agents, the ability to fork conversations, file upload, and more enable consultants to experiment with different workflows and compare results from across different models.

The various models that the Atomic Object team is using

As individual consultants get more experienced and creative in their use of AI to accomplish tasks, a number of capabilities built into TypingMind really start to shine:

File uploads
Custom System Prompts / AI Agents
The ability to switch models and regenerate a response to compare model outputs.
The ability to easily delete or edit prompts and regenerate responses when you realize a request is too vague or missing important detail.
The ability to fork a conversation, making it easier to create a substantial context for a project that can be reused across a number of conversations/purposes.

Take transcripts from meetings and dictations and generate comprehensive notes

Belle• TypingMind

Using multiple AI models is helpful because each LLM is good at different things. And combining them can help you get better, more accurate results for your use cases. Such a great approach!

Then have you found TypingMind to fulfill what you need?

Drew• Atomic Object

It’s worked really well for our intended use cases. It’s a substantial improvement over some of the early native AI apps many on our team had been using: Way more performant, capable, and nice to use.

It has a lot of nice details, such as being an excellent Progressive Web App that works great as a mobile app when added to your iPhone/Android Home Screen.

❝ It’s worked really well for our intended use cases. It’s a substantial improvement over some of the early native AI apps many on our team had been using: Way more performant, capable, and nice to use.❞

Drew Colthorp

Software Development Practice Lead at Atomic Object

Key Metrics and App Evaluation

Belle• TypingMind

Do you have any key metrics to share regarding the effectiveness of applying TypingMind in Atomic Object?

Drew• Atomic Object

We started using TypingMind in March. Over the past four months, use of the tool has grown to 40% of the company. Our AI use has gone up substantially as many team members have been leveraging AI more and more over the last several months. I increasingly see new names show up at the top of our usage charts, demonstrating awareness and utilization throughout the company.

❝ Over the past four months, use of the tool has grown to 40% of the company. Our AI use has gone up substantially as many team members have been leveraging AI more and more over the last several months.❞

Drew Colthorp

Software Development Practice Lead at Atomic Object

Belle• TypingMind

That’s impressive, Drew! I wonder if you have evaluated other solutions before using TypingMind? What makes TypingMind stand out from the crowd?

Drew• Atomic Object

We had a number of team members using Machato, which was/is an early LLM macOS app that hasn’t kept up with the evolving landscape. We also considered MindMac, however, as an individual Mac app, it makes it harder to manage licenses and API keys across our user base.

With TypingMind, we can centrally manage the available models and API keys, so that individual consultants can just start using the models when they become available.

We’ve also considered the various enterprise licenses from the major providers, such as ChatGPT Teams, Anthropic Claude Teams, Gemini for Google Workspaces, etc. The problem we’ve had with these options is that you’re locked into a single model, and we’re not yet ready to make a commitment there as an organization. Additionally, the tooling with many of these providers lags behind what TypingMind includes. For example, Gemini’s web interface is simplistic and painful to use coming from TypingMind. individual consultants can just start using the models when they become available.

TypingMind gives a good balance of great UI and model flexibility.

❝ TypingMind gives a good balance of great UI and model flexibility. ❞

Drew Colthorp

Software Development Practice Lead at Atomic Object

Belle• TypingMind

Thank you for sharing your insights and experiences with TypingMind. It's been a pleasure talking with you today!

Final Thought

We’ve learned how Atomic Object is leveraging TypingMind Custom to integrate multiple AI models into their consulting and software development workflows.

The strategic adoption of TypingMind has enabled Atomic Object to stay ahead in the rapidly evolving AI market, and ensure they can deliver high-quality results for their clients.

Stay tuned for more stories!

Author

Belle Nguyen

Marketing Team of TypingMind

Posted 12 Aug 2024

This article is written from an interview transcript between Atomic Object and TypingMind. Edited by human with AI assistance. Learn how to harness the power of AI in your business using TypingMind.