Fresh Labs
Fresh Bots
Vetted by Fresh's internal teams. Designed to work for others.
Background
Before a team can launch a chatbot, they have to answer questions about how it will access data, what it should sound like, which model it should run on, and how anyone will know if it’s actually working. For every new chatbot they build, they have to answer all of those questions again, from scratch. The result is that most teams either ship something generic, or never ship at all.
Included as part of the Brancher One platform, Fresh Bots allows users to easily create, customize, and manage chatbots, streamlining the process of building AI tools that automate workflows.
Challenges
Work began with three requirements:
- Let people ask questions in plain language about complex data
- Make it possible for non-technical users to fully customize a bot’s branding, underlying AI model, and tone
- Track bot performance from day one, keeping data transparent while constantly measuring bot quality
A platform tested by daily use, before it was ever a pitch
Fresh Bots started as a tool for our internal teams to automate their workflows. Given that it wasn’t intended to be a client-facing demo, we were able to approach the challenge from a rapid prototyping perspective—that is, building, verifying, and validating that the tool delivers greater workflow efficiency.
The pressure to deliver a truly useful tool for our employees shaped the Fresh Bots architecture: one central system for creating, configuring, and monitoring every bot, instead of one-off bots each maintained by whoever built them.
Designed to work in our proprietary Brancher AI ecosystem, Fresh Bots was stress-tested against real use cases before being offered to other organizations.
Ask a question, get an answer, right where you're working
At the center of the platform is a large language model that answers complex data questions right on the dashboard. Users type a question in plain English and get an answer immediately—no exporting data, no query language, no waiting on a report.
This approach shortens the distance between asking a question and acting on it. Given that our guiding criteria for success was efficiency, the approach accomplished our primary objective.
Users can look something up the moment they need it, which accelerates workflows and provides access to critical data in the Fresh knowledge base on the spot.
A bot that sounds like you, not like the platform's default
Every bot on Fresh Bots can be shaped to match the organization using it, instead of adapting to a generic default. Branding, AI model, and tone of voice are all adjustable, so bots can read as formal or casual, cautious or direct, depending on who they’re talking to.
Because these are settings, not engineering work, trying something new is low-risk. For example, a non-technical team member can swap models or tones, test with real users, and see what works—no developer required.
That flexibility enables more people to get involved in the building, training, and testing process, setting the foundation for enterprise-level scalability rather than individual use.
Tuning a bot becomes routine, not a guessing game
A unified dashboard manages every bot on the platform, but its bigger value is as an analytics tool.
How often are people actually using a bot? How fast does it respond? The built-in tracking feature for Fresh Bots presents data about bot usage, performance, and user satisfaction. This gives bot owners and platform developers real evidence to work from as Fresh Bots is continually refined.
Once again, our focus was to eliminate guesswork. With a tool that prioritizes transparency—rather than an AI black box—routine maintenance is easy and platform users can be confident in the bots the create and the value those bots deliver.
50+
Bots Deployed
500
User Conversations
20+
Workflows Automated
