Designing Bots: Creating Conversational Experiences

by Amir Shevat

From Alexa to Siri and from Slack bots to SMS and email bots, we rely more and more on software-driven conversational apps. In this practical guide, author Amir Shevat shows you how to design and build great conversational experiences and delightful bots that makes people s life more fun and productive.

You ll explore several real-world bot examples to understand what works and what doesn t, and learn practical design patterns for your own bot-building toolbox. This book is ideal for beginners and intermediate designers, as well as senior professionals exploring the conversational user experience paradigm. No coding experience or prior knowledge of conversational UI is required.


Learn what bots are, and understand bot types and major components that compose a bot
Explore different use-cases of bots and best practices around these use cases
Examine real-life examples and learn from their experience
Understand the bot anatomy (Onboarding, Notifications, Conversations, Advance UI controls) and their associated design patterns
Prototype your own first bot and experiment with user feedback
"

Genres: Design, Technology, Nonfiction, Technical

348 pages, Paperback

First published May 1, 2017

Reviews

[Yoly · Rating 4 out of 5]

A must-read if you work with bots or want to work with bots.

[Rosie · Rating 4 out of 5]

Already a bit dated in terms of what's available, but the basics on processes are always relevant.

[Sarah · Rating 4 out of 5]

Dated, despite only being 5 years old. That's the nature of the topic, though, and not any fault of the author. Solid beginner's guide!

[Inga Potzblitz · Rating 4 out of 5]

It was a good overview and even understandable for someone with no experience so far.

[Angelique · Rating 4 out of 5]

Helpful if you want to jumpstart your chatbot project. Easy to read and understand, plus the author's tips are very useful.

[Sudnya]

Notes to self:
Bots will probably/augment replacing websites as an interface (conversational) to a service.
web interface -> mobile interface -> conversation interface
Types of bots: personal vs team; super vs. specific (domain/content/purpose); b2b vs b2c (this changes how they interact / engage with users); new vs integration with legacy (salesforce)
Bots platforms: business (slack), consumer (FB); voice (alexa); legacy (email/sms). We need to pick the right platform to ensure success of our bot -- audience? revenue model? features to deliver your service most effectively;

Use cases:
to exploit business flows with bots and enhance productivity/reduce op-ex;
coach bot
notification/reminder bot
router between specialist humans
customer service and FAQ
entertainment

Anatomy: important to focus on how humans communicate & design bots. Features that matter:
personality: depends on audience
logos & icons: to show human like attributes
name: brand
human intervention: for error handling
NLP, conversation, image recognition, prediction (Q&A), sentiments
on boarding conversation is crucial - explicitly state the purpose of the bot
functionality scripting: flow for each function the bot performs, mitigate failure modes
feedback & error handling: crucial for long term user engagement
help & support - with a direct human, expected response latency
links, formatting, emojis, persistent menus are nice to haves.
Context and memory are a big challenge - since humans do it effortlessly in conversation
engagement methods: notifications, asking users to invoke, subscription to periodic notifications

Branding, logo, name matter.
Personality of the bot is a differentiator - think of audience, environment, job it does, social acceptance & values of your company. Keep this personality consistent across the experience.
To avoid personality issues, some human intervention might be required to keep things going smoothly (at least initially) - resolve ambiguity, escalation (speak to your supervisor :P ), supervised training online, humans improve conversation (domain experts)

AI: NLP, conversation management, computer vision, sentiment analysis to steer flow. Using AI is not always necessary for all types of bots.

Bots offer a new interface to an established human interaction - so tailor it to those etiquettes. During on boarding a user - state the purpose of the bot clearly and how to configure it, don't be a pushy salesman but make the value prop clear to engage users.

Task led conversation leads user through a funnel (via stories, intent mapping, entity extraction etc). Best to have least number of steps to complete a task. Topic led conversation is about user engagement so no such restriction but use divergence to bring it back. Add character, persona, gifs, random variation in dialogue. Priming users for formatting, ack to confirm, consistent experience. To revive engagement - try to ask questions and offer reminders etc and proactively show follow up available features.

Error handling is crucial to success - when to ignore vs. human intervention (latency expected), restarting conversation gracefully, refer to other bots.

Buttons are a great to guide the conversation, frame the interaction or steer the users within a set of possibilities (and avoid errors). Buttons are NOT good when there are too many options.

Links should show preview so user knows where they will lead - unfurling.

Bots should keep as much context as possible - e.g.: user intent so it can be used when referred to if user comes back to it later (otherwise face palm! errors), infer from pronouns, just be explicit & ask specific questions to avoid ambiguity in interaction.

Monetization ??? subscription, ads, market research, selling services, referrals, virtual goods.

Design process:
use case definition -- purpose, platform, persona, logo visuals;
conversation scripting -- workflow then feedback +/-, error handling, intent, entity; buttons? validate on users?
design & test - small text

Make sure you look at your data & continuously improve the bot.

[Omar Venturi · Rating 5 out of 5]

Ottima introduzione all'industria dei bot. Storia, situazione attuale e previsioni sul futuro sono le tre classiche parti del libro, dove la parte più corposa è composta dalla situazione attuale.
I bot, come più volte riportato dall'autore, sono solo all'inizio e molte potenzialità restano ancora inespresse.
Il libro si focalizza sui bot "testuali" ed un ottimo testo per la parte vocale, suggerita dallo stesso autore, è "Designing Voice User Interfaces: Principles of Conversational Experiences" di Cathy Pearl.

Il libro affronta tutti gli aspetti relativi al design (NON all'implementazione!) di un bot, dal progetto iniziale, alla prototipazione, fino al testing ed all'analisi dei dati.
Propone una metodologia o comunque un approccio al design che può essere seguito anche solo parzialmente.
Il libro è molto valido ed attualmente (novembre 2017) assolutamente allineato allo stato dell'arte, ma vista la velocità con cui si sta evolvendo questo mercato, presto sarà superato. Consiglio quindi di verificare, prima di acquistarlo, che l'autore non abbia rilasciato un aggiornamento o che non ci sia disponibilità di alternative più recenti

[Ray · Rating 3 out of 5]

It took me about 2 years from the time I read the first page of this book to read the last page. In that time, I discovered that I really don't like this kind of book! That is a subjective preference, and does not take away from how well "Designing Bots" is written - and it is fairly well written. However, something about the fact that the value of the information in the book will decay quickly made me less motivated to read it. Perhaps it should have made me feel like I should finish reading quickly, but it did not.

The thing I feel like "Designing Bots" does well is separate the discipline of crafting conversational interfaces from the mythos surrounding bots. It really is a type of design, yet the general perception of the public probably revolves around artificial intelligence. Many of the design principles hold across different types of products in addition to bots, such as apps and websites. That said, a large portion of the content was specific to (time-based snapshots of) certain platforms. This was the sort of thing that made the book boring for me to read.

[madi · Rating 1 out of 5]

Find it very amusing that the watermelon triangle accounts are all flocking to Bluesky - Bsky.app whose investor is Amir Shevat an Israeli expat now living in Texas.

He is certainly knowledgeable about bots evidenced by the millions that overnight flooded Bluesky prior to the 2024 election. Anonymous accounts provoking rage and political propaganda to drive engagement.

But all of this makes sense being thar Amir used to work with Twitter and is an investor in multiple social media start-ups including Reshuffle, an Israeli company bought out by Twitter.

Sell your Jewish people out to the hateful trolls for money, right Amir?

[Nikhil Bansal · Rating 5 out of 5]

This book is for bot enthusiast who have done all the coding stuff required to build a bot and after that want's to improvise his/her bot for market or for the user. Author explain's each and every aspect which makes bot flow smoother. Book also explain's role of Artificial Intelligence in the bot (chat/voice) very precisely.

[Valentino · Rating 5 out of 5]

That's a new era. The book to read to understand how it works. Too US centric, in Europe we have different tools and visions, but it's stil excellent

[Roberto Chavez · Rating 3 out of 5]

OK. Not very technical, but bot design/development does not have to be deeply technical any more. Not much here to make it compelling. Wait for a second edition.

[Scott · Rating 4 out of 5]

Great primer on designing bots. Heavily weighted to instances on Slack’s platform since the author works there.

[Sweemeng Ng · Rating 3 out of 5]

It covers design on bots, the business case, the pitfall, and how to build a bot. it is ok

[Martijn Reintjes · Rating 3 out of 5]

Designing Bots gives a nice overview of the space, but lacks the depth to be really useful.
It's basically a long ass blogpost repurposed as a book.

[Natalia · Rating 5 out of 5]

Give a very nice overview with interesting practical examples of what you have to consider while building a chatbot

[Sebastian Godoy · Rating 3 out of 5]

I find the content too obvious. The book it's a good overview of how to design a chatbot, but the author never goes deep in a subject

[Cinantya · Rating 4 out of 5]

baca buku buat kerjaan biar kayak orang bener.

[Tim · Rating 1 out of 5]

This book is quite outdated. Glad I borrowed from the library.

[Robert Foley · Rating 4 out of 5]

If your a product manager or designer interested in bots this is a must read. Focused on the experience for the customer and what problems need to be solved. Still discusses the technology involved, but only one chapter instead of an entire book...no programming experience required for reading.

[Héctor Iván Patricio Moreno · Rating 3 out of 5]

Este libro parece que en su tiempo fue una buena introducción para aquellos que no sabían nada de los programas y servicios que llamamos bots, sin embargo, lo siento poco denso para la cantidad de información que aporta.

Creo que es bueno darle una lectura rápida (muy rápida) para aprender aspectos de los servicios que permiten exponer un servicio como un bot y los temas relacionados con su planeación y diseño. Algunas partes se siente como documentación de las plataformas y para esto, ya se quedó desfasado.

Lo recomiendo como una lectura rápida, sobre todo para aquellos que quieren empezar con esos servicios, pero en realidad si ya has hecho bots, no te aportará mucho.

[Lorena · Rating 4 out of 5]

I hadn’t realized how complex bot design can be and how much work goes into creating a performant and likeable chatbot. Overall, I found the information provided in this book interesting and helpful, and recommend it to anyone interested in chatbot design.
full review on my blog

[Edgar Lezama]

Gracias