Collaborative design sprint leads to new bot for Slack [Beta]

The Economist thinks bots are “the next frontier” and bots were cited as one of the top CES trends to watch in 2017. That’s why Concur Labs is venturing into how conversational interfaces can benefit the world of travel and expense. We’ve introduced our first Concur Bot for Slack beta this week to explore what bots can do for our users.

Slack, the popular group messaging service with 4 million users, enables teams to organize conversations, coordinate projects and topics, and interact in real-time through “channels.” With the Concur Bot for Slack, you can interact with Concur in a conversational way and send requests or questions about your travel plans and manage expenses. The experience is fast and intuitive, and part of our ongoing effort to meet customers where they work.


How can our Concur Bot for Slack help you?

In collaboration with Slack, we turned every day actions in Slack into easy ways to manage expenses and trip planning. You can, for instance, use the Concur Bot for Slack to:

  • Submit a Receipt: You can, for example, upload a receipt to @concur and the bot will submit it right to your Concur account for expensing



  • Get a Summary of Expense Reports: Type “report summary” in a note to @concur and the bot will send an inventory of your expense reports
  • Create Quick Expenses: For basic expenses, write a simple note like “expense $15 for Uber” and the bot will automatically send expense details to your Concur account
  • See Your Travel Itinerary: Type “itinerary” or “trip details” and your upcoming business trip information will be sent right back to you

With the bot prototype currently in private beta, we look forward to gathering and analyzing feedback to continue improving the experience for our customers over time.


How we built the bot

Building a new technology is bridge between innovation and learning. Here are three take-aways we’ve learned so far:


Learning #1: It’s an “anything’s possible” mindset

We like to think ideas are unlimited. With new prototypes, there’s brainstorms, lean methodologies, and ideas that crop up when we’re discussing completely different topics. All fuel for the ideation process. For Concur Bot for Slack, we started with the notion of simply making it easy to submit receipts from Slack for expensing. This expanded into enabling you to submit quick expenses, view expense report summaries, and even access travel itineraries. Possibilities continue to develop.


Learning #2: Run collaborative design sprints

Design-thinking is a human-centered approach to solving problems. In a design sprint, teams work in real-time – with sticky notes, markers, whiteboards – to map out problems and ideas. With Slack, we hosted a daylong design sprint to explore scenarios. This led to our first foray into creating a messaging bot, and a view into how Machine Learning (ML) can turn natural language and human interactions into useful services.


Learning #3: Capture feedback, iterate, repeat

We’ll often prototype within 24-48 hours and try different scenarios. “Once we start down a path, there’s unexpected diversions and options. It’s helpful to uncover these in the early development stages,” says Richard Puckett, development manager. “To get firsthand insights, we’re running early betas with a handful clients, including Slack, to have visibility into how the bot is actually working, or not.”

With each round of feedback and testing, the team is improving the experience to be even more intuitive. We’re analyzing as much data as possible and using each round of input to evolve the prototype.


What's next?

Moving forward, we believe conversational interfaces have the potential to simplify otherwise complex tasks. Natural Language Processing and Understanding (NLP, NLU) are maturing at unprecedented rates, allowing bots to move beyond simple command – to response interfaces.

In a future post on, we’ll get into the details on building our bot for Slack. For our initial private beta, the Labs team worked with's Botkit and the Concur Platform JS SDK. We used Natural Language Processing (NLP) so the bot can understand your messages and try to interpret new or unexpected requests. Concur APIs involved include Quick Expense, Receipts, Reports, Itinerary as well as the new OAuth Service.


Follow the evolution of the Concur Bot for Slack on or meet the Labs team and see demos at Fusion, our annual customer event, March 14-17, 2017:


Loading next article