Start free trial
Trending New Genres Authors
EnglishEnglish
EspañolSpanish
简体中文Chinese
繁體中文Chinese (Traditional)
FrançaisFrench
DeutschGerman
日本語Japanese
PortuguêsPortuguese
ItalianoItalian
한국어Korean
РусскийRussian
NederlandsDutch
العربيةArabic
PolskiPolish
हिन्दीHindi
Tiếng ViệtVietnamese
SvenskaSwedish
ΕλληνικάGreek
TürkçeTurkish
ไทยThai
ČeštinaCzech
RomânăRomanian
MagyarHungarian
УкраїнськаUkrainian
IndonesiaIndonesian
DanskDanish
SuomiFinnish
БългарскиBulgarian
עבריתHebrew
NorskNorwegian
HrvatskiCroatian
CatalàCatalan
SlovenčinaSlovak
LietuviųLithuanian
SlovenščinaSlovenian
СрпскиSerbian
EestiEstonian
LatviešuLatvian
فارسیPersian
മലയാളംMalayalam
தமிழ்Tamil
اردوUrdu
Searching...
SoBrief
The AI Advantage

The AI Advantage

How to Put the Artificial Intelligence Revolution to Work
Make AI useful at work without betting the business on a miracle.
by Thomas H. Davenport 2018 248 pages
3.54
149 ratings
Listen
Immersive
V2.3
Amazon Kindle Audible
Try Full Access for 3 Days
Unlock listening & more!
Continue

Key Takeaways

Treat AI as a long investment, not a near-term miracle

Grand promises can stall. M.D. Anderson’s cancer-advice project with IBM Watson cost $62 million and was never used to treat a patient. The project struggled with management and system integration, despite the excitement around Watson’s game-show victory. DBS Bank also stopped an ambitious AI wealth-adviser pilot when the technology could not reliably interpret the bank’s varied charts and research reports.

Quiet gains still matter. DBS redirected its efforts toward narrower operational problems. Machine learning helped predict when ATMs needed cash, reducing the average interval between cash shortages from three months to fifty-five years and cutting replenishment trips by more than 10 percent. Amazon’s founder similarly described much of the company’s machine learning as invisible improvements to everyday operations.

Analysis

The contrast between a stalled moon shot and a quiet operational gain makes technological maturity concrete: impressive demonstrations do not guarantee dependable performance in a real institution. Smaller applications also offer a way to learn about data, workflows, and accountability without betting the enterprise on a single system. Yet incrementalism should not become an excuse to avoid strategic change. The strongest portfolio can pair low-risk applications with explicit experiments aimed at larger opportunities, using evidence from each to update the organization’s expectations.

Match the tool to the task before buying the AI story

AI is a toolbox, not one machine. The book groups several distinct methods under a broad label. Machine learning finds patterns in data and predicts outcomes. Natural language processing works with speech and text. Deep learning, a layered form of machine learning, is particularly useful for images and voice, though its decisions can be difficult to interpret. Rule-based systems apply human-written logic, while robotic process automation follows instructions across digital systems.

The distinctions change the result. A bank may use machine learning to predict fraud, but a rule-based tool may be preferable when decision logic must be transparent. DBS discovered that a system advertised as an intelligent adviser could not handle the many formats of charts and reports its staff relied on. A technology can be powerful in general and still be wrong for a specific task.

Analysis

Choosing by task rather than by brand is a form of disciplined problem definition. It forces managers to ask what input the system must handle, what output counts as success, and how much explanation users or regulators require. Those questions also prevent a common category error: fluency in recognizing speech does not automatically mean competence at answering complex questions. The book’s inclusive use of AI is practical for business, but it can blur differences in risk, cost, and capability. Clear technical distinctions make comparisons more honest.

Start with business bottlenecks, then build a portfolio of modest wins

Choose the problem first. A cognitive strategy connects AI projects to business priorities, such as improving customer offerings, making better decisions, or reducing friction in operations. In a 2017 survey of 250 managers already familiar with cognitive technologies, 88 percent said the tools mattered to products and services, and 93 percent said they mattered to internal processes. These results show enthusiasm among active adopters, not universal readiness.

Look for blocked knowledge. The book points to three useful openings: expertise trapped in silos, a valuable service that is too costly to scale, or more data than people can analyze. A company might begin with an internal help desk or sales prediction rather than an all-purpose virtual adviser. Select pilots for business value and feasibility, then carry successful components into production instead of collecting demonstrations that never change the work.

Analysis

The bottleneck lens makes AI strategy more concrete than a list of fashionable technologies. It asks where better access to knowledge could change a decision or service, and whether the proposed system can reach that point in the workflow. The survey figures are suggestive, but their respondents were screened for AI awareness and use, so they cannot establish broad corporate readiness. A stronger strategy treats executives’ confidence as a hypothesis to test, with measures of customer value, operating performance, and adoption rather than enthusiasm alone.

Automate tasks, then redesign the whole process around them

Automation is not redesign. Robotic process automation, or RPA, uses software to perform structured digital tasks across systems, much like a person following a script. In a review of 152 consulting projects, process automation was the most common AI activity, appearing in 71 projects. NASA tested four RPA pilots in administrative functions; one human-resources application completed 86 percent of transactions without human intervention.

Fix the workflow as well. Automating every existing step can preserve needless complexity or outdated rules. The book argues for reviewing the process before deployment, removing steps that no longer add value, and deciding which exceptions still require people. Integration is a major hurdle: 47 percent of surveyed AI-aware managers said connecting projects with existing systems and processes was difficult. A few bots may be simple; managing thousands can become an architectural problem.

Analysis

Automation changes a system of work, not merely a task list. When one step becomes faster, the constraint may move to approvals, exceptions, or the next department, so local efficiency does not guarantee end-to-end improvement. Process mapping and worker participation help reveal those dependencies before they are encoded. There is also a risk in treating software as inherently mistake-free: a bot can repeat a flawed rule consistently and at scale. Effective redesign therefore includes exception handling, monitoring, and responsibility for correcting errors.

Use AI to amplify workers before assuming it will erase jobs

Jobs are bundles of tasks. Estimates of future automation differ sharply. Some studies cited in the book predict that as many as 47 percent of jobs could be automated, while another estimates that about 9 percent of jobs across studied countries are at high risk. The difference partly reflects whether researchers classify whole occupations or individual tasks, and whether they account for cost, regulation, and social acceptance.

Augmentation is common in practice. In the Deloitte survey, only 6 percent of respondents named headcount reduction as their main goal for AI, and 22 percent included it among their top three. Across the projects reviewed, job reductions were limited and often involved outsourced work. At Vanguard, automated investment tools handled routine portfolio tasks while human advisers shifted toward coaching clients and helping them stay with their plans.

Analysis

The task-versus-job distinction is essential, but it does not settle who bears the cost of transition. A role can remain while its status, pay, autonomy, or entry-level pathways change substantially. Augmentation is a promising design choice, not an automatic outcome: it requires training, usable interfaces, and a real plan for work displaced inside the job. The account is strongest when it treats workforce effects as something organizations can shape, rather than predicting that technological capability alone determines employment.

Build the data foundation before expecting machine intelligence

Models need usable evidence. Machine learning depends on data that are relevant, sufficiently plentiful, and often labeled with known outcomes. Yet large organizations commonly hold fragmented or inconsistent records. GSK’s research teams found that scientists struggled to use data beyond their own departmental silos. The company chose machine-learning tools to match and consolidate research data rather than relying only on a slower, manual data-management approach.

Preparation unlocks new questions. GSK initially prioritized ten research use cases, then expanded to 250 after building an integrated data environment. Its clinical-trial data could be mapped to shared industry formats, making comparison across studies more practical. BMO Bank likewise invested in data governance, common records, and infrastructure before expanding its use of analytics and automation. Data preparation is not glamorous, but without it, even sophisticated models may produce unreliable or unusable results.

Analysis

Data quality is not a preliminary chore that ends when a model is built. It is an ongoing organizational capability involving definitions, access, provenance, and responsibility for updates. The GSK example shows that connecting information can create scientific questions that were previously impractical to ask. Still, a larger data pool is not automatically a better one. If records systematically omit certain groups or contexts, scale can reproduce that gap with greater confidence. Data readiness should therefore include scrutiny of what the organization has failed to record.

Put fairness and explainability inside the model's design brief

Prediction can distribute harm. A criminal-sentencing tool called COMPAS was criticized after an investigation found that it appeared to overestimate recidivism risk for Black defendants and underestimate it for white defendants. Its developer disputed the analysis, illustrating that fairness depends partly on which measure is used. Facebook also faced criticism when its advertising tools permitted targeting based on hateful categories, showing that automated systems can reproduce damaging patterns without anyone explicitly programming that intention.

Build review into deployment. Determine how much explanation a decision requires before selecting a model. High-stakes uses such as credit, health, or sentencing need more scrutiny than low-stakes ad placement. Review inputs for unjustified proxies, examine missing or unusual data, test outcomes across affected groups, and reassess models as data change. When possible, use methods that can explain the factors behind an individual prediction.

Analysis

Fairness is not a single technical setting. Different definitions can conflict, so teams need to state which harms they are trying to prevent and who has authority to make that judgment. Explainability makes scrutiny possible, but a readable model is not necessarily fair; an opaque model may also perform well on average while failing particular groups. The design process therefore needs both technical testing and institutional accountability. This is especially important when an automated score influences a person’s access to rights, services, or opportunity.

Earn trust by disclosing limits and keeping humans accountable

Do not disguise a machine. In a consumer survey described in the book, 87 percent supported rules against AI systems posing as people. Customers may accept automated help, but they need to know whether they are speaking with a system and what it can actually do. The Swedish bank SEB introduced its digital agent as a trainee, describing a limited role rather than implying human-level competence.

Keep a responsible fallback. Facebook’s efforts to identify fake news and harmful content showed the limits of relying on automated filters alone. The company added human reviewers and tested other interventions, though the results remained uncertain. A responsible service makes escalation possible, tells users what the system cannot handle, and checks whether errors or harmful outcomes are appearing. Overstating capability may win attention initially, but failed promises make later adoption harder.

Analysis

Trust is partly a calibration problem: users need a realistic sense of what a system can and cannot do. Disclosure supports that calibration, while a human fallback prevents an automated interaction from becoming a dead end. The Facebook example also shows that safety cannot be reduced to a classifier that labels content correctly; incentives, platform design, and human behavior shape outcomes too. Human review is not a universal fix, since reviewers need time and authority, but accountability cannot be outsourced to the model.

Bring workers into redesign before bots reach production

Participation changes adoption. AI can threaten people whose expertise is being captured or whose routines are changing. The book recommends involving affected employees in designing how humans and systems divide the work. At one apparel retailer, buyers resisted an AI pilot that made merchandising recommendations, asking what their role would be if the system took over familiar tasks. The chief merchandising officer kept the project, but had to show how buyers could focus on understanding customers and manufacturers.

Prepare people for new responsibilities. Workers may need to interpret recommendations, handle exceptions, explain decisions, or monitor system performance. That requires practical training, clear communication, and enough time to adapt. SEB’s employees helped train its digital agent and saw the effort as experimentation rather than an immediate cost-cutting exercise. Participation does not guarantee agreement, but excluding workers makes it harder to build useful processes and identify problems early.

Analysis

Involving workers is both a change-management practice and a source of operational knowledge. Employees often understand exceptions, informal customer needs, and workarounds that process diagrams miss. Their input can reveal where automation will fail before those failures reach customers. However, participation should not become a symbolic consultation after the important decisions are settled. Credible involvement includes influence over task allocation, training, and the treatment of roles that change, especially when the organization has real incentives to reduce labor costs.

Analysis

The book’s central contribution is a shift in what counts as an AI decision. Rather than asking whether a company should adopt artificial intelligence, it asks which tasks are suitable, what data and organizational changes they require, and how the resulting capability will enter real work. That shift punctures both technological spectacle and passive skepticism. The Watson cancer projects show that a celebrated system can fail to cross the distance between demonstration and clinical use. DBS’s ATM forecasting, by contrast, illustrates how a modest application can produce durable operational value.

The argument also treats AI as a management problem, not simply a software purchase. Models need integration, data preparation, redesigned processes, skilled translators between business needs and technical methods, and employees who can work alongside automated systems. This is a more realistic account of implementation than claims that a model alone transforms an enterprise. Its examples repeatedly suggest that organizational friction, rather than algorithmic novelty, determines whether a pilot becomes useful infrastructure.

The book’s emphasis on incremental gains is persuasive, but it creates a tension. A collection of small efficiencies may improve performance without changing who benefits, whose work is devalued, or how decisions are governed. The author acknowledges bias, privacy, job displacement, and trust, yet these concerns are not peripheral safeguards. They shape the definition of success itself. A system that reduces costs while imposing opaque errors on customers is not simply a technically successful deployment with an ethical issue attached.

The book is also a snapshot of 2018, when the limits of particular platforms and enterprise adoption were especially salient. Its durable lesson is less about any vendor or model than about disciplined matching: begin with a consequential business problem, select methods whose strengths and weaknesses fit it, prepare the organization to absorb the result, and measure outcomes beyond the pilot. AI’s advantage is not that machines replace judgment wholesale. It is that organizations can reorganize information and work, provided they remain accountable for what the systems do.

Summary in 30 Seconds
AI creates business value when its method fits a specific task and its data, systems, and workflows are ready. Start with a consequential problem, test practical applications, and scale pilots that work. Automation can handle routine tasks while employees shift toward judgment and customer support. Fairness, explainability, human review, and worker participation belong in implementation, not as afterthoughts.
Contains spoilers

Last updated:

Report Issue

Review Summary

3.54 out of 5
Average of 149 ratings from Goodreads and Amazon.

Readers found The AI Advantage to have a strong underlying message but criticized its lack of depth and reliance on widely-known examples. Some felt it offered practical insights for businesses considering AI implementation, while others found it outdated and repetitive. Positive reviews praised its accessibility and overview of AI's business applications. Criticisms included its focus on superficial AI technologies, consultant-like writing style, and limited exploration of more complex AI concepts. Overall, opinions were mixed, with readers seeking more in-depth analysis of AI's impact on work and industry.

Your rating:
4.18
242 ratings
Want to read the full book?

Glossary

Cognitive technologies

AI for narrow tasks

Davenport’s broad label for technologies that perform tasks associated with human knowledge, perception, or insight, usually within limited domains. The category includes machine learning, natural language processing, expert systems, physical robots, and robotic process automation. The term is deliberately inclusive, although the systems grouped under it differ greatly in how intelligently they operate.

Robotic process automation

Software for structured tasks

Software that performs repetitive digital work by following rules and interacting with existing information systems as a human user might. Despite the name, it does not usually involve a physical robot. RPA is comparatively inexpensive and straightforward to configure, but it does not inherently learn, and large deployments can create maintenance and integration challenges.

Cognitive insight

Patterns extracted from data

A category of business AI applications that uses algorithms, most often machine learning, to identify patterns in large volumes of structured data and use them to predict or classify outcomes. Examples include detecting fraud, forecasting customer behavior, and anticipating equipment failure. These tools can analyze more data and create more detailed models than conventional approaches, but often need labeled examples.

Cognitive engagement

Personalized machine interaction

A category of AI applications that interacts with customers or employees through language or images. It includes chatbots, intelligent agents, recommendation systems, and some treatment-advice tools. These applications can make services available around the clock, but their competence is usually limited to particular topics or tasks, and smooth escalation to a human remains important.

Digital twin

Data model of an asset

A digital representation of a physical machine or collection of machines, updated with data from sensors and other sources. An organization can use the model to detect anomalies, assess performance, and predict maintenance needs. The approach is useful when equipment produces continuing data, but its predictive value depends on data quality and suitable models.

Knowledge graph

Connected facts and entities

A structured collection of entities, such as people, places, or products, together with facts about them and their relationships. Knowledge graphs can help natural language systems interpret terms in a particular domain and support question answering. Creating or adapting one often requires substantial work to define concepts, relationships, and relevant language.

Probabilistic matching

Likely record identification

A method for identifying records that probably refer to the same person, organization, product, or other entity, even when their names or details differ. Machine learning can estimate these matches across large databases more efficiently than manual review or extensive rule writing. The method supports data integration, but uncertain matches may still require validation.

LIME

Local prediction explanation

Locally interpretable model-agnostic explanations, a technique described in the book for identifying which features most influenced a particular model prediction. It can help explain an individual outcome, such as why a bank flagged a transaction as potentially fraudulent. It does not make every model globally transparent, but can provide a user-specific account of a decision.

FAQ

What is "The AI Advantage: How to Put the Artificial Intelligence Revolution to Work" by Thomas H. Davenport about?

  • Enterprise AI focus: The book examines how large organizations can effectively adopt and leverage artificial intelligence (AI) and cognitive technologies in business processes, products, and strategies.
  • Practical over hype: Davenport emphasizes real-world applications and realistic expectations, steering clear of exaggerated claims about AI’s capabilities.
  • Historical and current context: It traces the evolution of AI, from early expert systems to modern machine learning, using case studies from companies like Amazon and DBS Bank.
  • Strategic and organizational insights: The book explores how AI impacts organizational structures, jobs, and business models, and provides guidance on managing these changes.

Why should I read "The AI Advantage" by Thomas H. Davenport?

  • Balanced, realistic perspective: Davenport offers a nuanced view of AI, focusing on what it can realistically achieve for businesses in the short and long term.
  • Actionable advice for managers: The book provides practical guidance for business leaders on developing AI strategies, selecting technologies, and scaling projects.
  • Comprehensive coverage: It addresses AI technologies, business applications, workforce impacts, and ethical considerations, making it valuable for professionals and decision-makers.
  • Preparation for change: Readers gain insights into managing organizational change, workforce transitions, and responsible AI adoption.

What are the key takeaways from "The AI Advantage" by Thomas H. Davenport?

  • AI augments, not replaces: The book argues that AI is more likely to augment human workers than fully automate jobs, leading to new roles and skill requirements.
  • Integration is challenging: Embedding AI into existing systems and processes is often more difficult than developing AI models themselves.
  • Data and talent are critical: High-quality data and skilled AI practitioners are essential for successful AI adoption.
  • Ethics and transparency matter: Addressing algorithmic bias, explainability, and responsible use of AI is crucial for long-term success.

How does Thomas H. Davenport define AI and cognitive technologies in "The AI Advantage"?

  • Broad, inclusive definition: AI and cognitive technologies are systems that use human-like capabilities—such as knowledge, perception, and insight—to solve narrowly defined tasks.
  • Task-focused approach: The book emphasizes that current AI technologies are best at augmenting specific human tasks rather than replicating entire jobs or full human cognition.
  • Technologies included: Davenport includes machine learning, natural language processing (NLP), robotic process automation (RPA), expert systems, and physical robots under the AI umbrella.
  • Evolving intelligence: The definition acknowledges the growing sophistication and integration of these technologies in business.

What are the main business applications of AI discussed in "The AI Advantage"?

  • Process automation: Using RPA and AI to automate repetitive, structured back-office tasks, improving efficiency and reducing errors.
  • Cognitive insight: Applying machine learning to analyze large datasets for purposes like fraud detection, predictive maintenance, and personalized marketing.
  • Cognitive engagement: Deploying chatbots and intelligent agents to interact with customers and employees, enhancing service and support.
  • Product and service enhancement: Embedding AI into products (e.g., autonomous vehicles, smart appliances) and creating new AI-driven business models.

What are the key AI technologies and methods explained in "The AI Advantage"?

  • Machine learning and deep learning: Statistical methods that train models on data to predict or classify outcomes, with deep learning using neural networks for complex tasks.
  • Natural language processing (NLP): Techniques for analyzing and understanding human language, including chatbots and speech recognition.
  • Robotic process automation (RPA): Software robots that automate structured digital tasks by mimicking human interactions with IT systems.
  • Rule-based expert systems and physical robots: Older AI forms using if-then rules and robots for physical tasks, now increasingly integrated with cognitive capabilities.

What are the main technical approaches to AI implementation in "The AI Advantage"?

  • Embedded AI in software: Integrating AI capabilities into existing enterprise systems (e.g., Salesforce Einstein) for easier adoption.
  • RPA as a gateway: Using RPA to automate structured tasks, often serving as an entry point to more advanced AI.
  • Cognitive platforms and multi-vendor strategies: Larger firms may build or buy broad AI platforms or combine proprietary and open-source tools, requiring sophisticated integration.
  • Data readiness: High-quality, well-integrated data is essential; companies may use AI itself to improve data management and integration.

What organizational and workforce changes does "The AI Advantage" predict due to AI?

  • Task augmentation: AI will change the nature of work by augmenting many human tasks, requiring new skills and roles.
  • New roles and skills: Demand will grow for data scientists, AI specialists, and “translators” who bridge business and technology.
  • Process and structure redesign: Organizations must rethink business processes and structures to optimize human-machine collaboration.
  • Gradual workforce impact: Displacement is expected to be slow and marginal initially, with productivity gains supporting economic growth.

How does Thomas H. Davenport advise companies to develop an AI strategy in "The AI Advantage"?

  • Align with business objectives: Identify key problems or opportunities where AI can add value, focusing on strategic priorities.
  • Prioritize and pilot projects: Evaluate use cases for feasibility and impact, and create a portfolio of pilots and proofs of concept.
  • Build internal capabilities: Develop or acquire AI talent, invest in data infrastructure, and foster organizational understanding of AI.
  • Plan for scaling: Prepare for integration challenges, change management, and continuous improvement to move from pilots to production.

What challenges and limitations of AI adoption are highlighted in "The AI Advantage"?

  • Integration difficulties: Embedding AI into existing systems and processes is often the most significant challenge.
  • Talent scarcity: There is a shortage of skilled AI practitioners, making it hard to build and sustain AI capabilities.
  • Data quality and availability: Poor or siloed data can hamper AI effectiveness, requiring investment in data infrastructure.
  • Organizational resistance: Employees and managers may resist changes, necessitating thoughtful change management and communication.

What ethical and social implications of AI are discussed in "The AI Advantage"?

  • Algorithmic bias and fairness: AI systems can perpetuate or amplify biases, raising fairness and discrimination concerns.
  • Transparency and explainability: The book highlights the need for explainable AI, especially in regulated industries and high-stakes decisions.
  • Privacy and security: AI’s ability to analyze vast personal data raises privacy issues, and AI is both a tool for and a target of cybersecurity threats.
  • Responsible adoption: Davenport stresses proactive management of AI’s organizational, social, and ethical implications.

What are some notable real-world examples of AI applications in business from "The AI Advantage"?

  • Amazon Go stores: Use of image recognition and sensor fusion to enable checkout-free shopping, transforming retail operations.
  • Bank of Montreal: Implementation of machine learning and RPA for customer segmentation, fraud detection, and process automation.
  • GlaxoSmithKline R&D: Use of AI and machine learning to integrate research data and accelerate drug discovery.
  • Facebook’s content moderation: Challenges in using AI to detect fake news and harmful content, illustrating both potential and limitations.

What are the best quotes from "The AI Advantage" by Thomas H. Davenport and what do they mean?

  • Amara’s Law: “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” This cautions readers to have realistic expectations about AI’s immediate impact while recognizing its transformative potential.
  • Kevin Kelly on AI: AI is “cheap, reliable, industrial-grade digital smartness running behind everything, and almost invisible except when it blinks off.” This highlights AI’s future as a ubiquitous, behind-the-scenes utility.
  • Jeff Bezos on AI impact: “Much of what we do with machine learning happens beneath the surface... quietly but meaningfully improving core operations.” This underscores that AI’s most significant benefits may be operational rather than flashy or customer-facing.

About the Author

Thomas H. Davenport is a distinguished scholar in information technology and management, holding the President's Chair at Babson College. He has authored numerous influential books and articles on business process reengineering, knowledge management, and analytics. Davenport's work has significantly contributed to establishing various business ideas and practices. His research background includes leadership roles at major consulting firms and research centers. With degrees from Trinity University and Harvard University, Davenport continues to share insights through his website and regular contributions to the Harvard Business Review blog, cementing his position as a thought leader in the field of business technology and management.

Download PDF

To save this The AI Advantage summary for later, download the free PDF. You can print it out, or read offline at your convenience.
Download PDF
File size: 0.40 MB     Pages: 20

Download EPUB

To read this The AI Advantage summary on your e-reader device or app, download the free EPUB. The .epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.
Download EPUB
File size: 1.39 MB     Pages: 27
Want to read the full book?
Follow
Listen
Home
Swipe
Library
Get App
Try Full Access for 3 Days
Listen, bookmark, and more
Compare Features Free Pro
📖 Read Summaries
Read unlimited summaries. Free users get 3 per month
🎧 Listen to Summaries
Listen to unlimited summaries in 40 languages
—
❤️ Unlimited Bookmarks
Free users are limited to 4
—
📜 Unlimited History
Free users are limited to 4
—
📥 Unlimited Downloads
Free users are limited to 1
—
Risk-Free Timeline
Today: Get Instant Access
Listen to full summaries of 26,000+ books. That's 12,000+ hours of audio!
Day 2: Trial Reminder
We'll send you a notification that your trial is ending soon.
Day 3: Your subscription begins
You'll be charged on Nov 25,
cancel anytime before.
Consume 2.8× More Books
2.8× more books Listening Reading
Our users love us
1M+ readers
Trustpilot Rating
TrustPilot
4.6 Excellent
This site is a total game-changer. I've been flying through book summaries like never before. Highly, highly recommend.
— Dave G
Worth my money and time, and really well made. I've never seen this quality of summaries on other websites. Very helpful!
— Em
Highly recommended!! Fantastic service. Perfect for those that want a little more than a teaser but not all the intricate details of a full audio book.
— Greg M
Save 62%
Yearly
$119.88 $44.99/year/yr
$3.75/mo
Monthly
$9.99/mo
Start a 3-Day Free Trial
3 days free, then $44.99/year. Cancel anytime.
Unlock a world of fiction & nonfiction books
26,000+ books for the price of 2 books
Read any book in 10 minutes
Discover new books like Tinder
Request any book if it's not summarized
Read more books than anyone you know
#1 app for book lovers
Lifelike & immersive summaries
30-day money-back guarantee
Download summaries in EPUBs or PDFs
Cancel anytime in a few clicks

Why one summary?

Scanner
Find a barcode to scan

⭐
⭐
⭐
We have a special gift for you
Open
38% off
DISCOUNT FOR YOU
⭐
⭐
⭐
$79.99
$49.99/year
Only $4.16 per month
Continue
2 taps to start, super easy to cancel
Settings
General
Widget
Loading...