Tome's Next Chapter

Chapter 1: What we learned

You may know of Tome as the tool that could help you turn a prompt into a slide deck. We enabled people around the world to easily express their ideas in a visual format by typing in a sentence or pasting in a document, skipping the time and effort needed to create great-looking slides.

The product went viral in 2023 — hitting 20M users in just 18 months. Tome became one of the fastest growing consumer applications of all time. Millions of people were excited to see how AI could help with things they didn’t enjoy doing, like crafting decks.

We looked at who was using Tome and for what. Most people signed up for Tome because of the promise of useful generative AI at work. They wanted a copilot that would help them do research, start a story, and help shape it to something compelling.

In looking at the entire research, ideation, synthesis, and shaping process - it became clear to us that general reasoning wasn’t going to be enough.

To make a compelling presentation for a banker, analyst, lawyer, salesperson, or founder - one needs to know a lot more about their company, their job, their prospects, and what they believe to be true than what’s available to a foundation model. More over, presentations are so drastically different across professions that we found there was very little carryover work from one function to another.

When faced with the question of “How might we deliver on the promise of AI helping people tell compelling stories?”, it became evident to us that we needed to focus on one persona at a time with a narrower set of use cases. The thesis of building applied AI for everyone at once seemed fundamentally flawed to us.

Applying AI to a single persona

LLMs have unlocked an incredible amount of productivity, but require a narrow focus to handle more complex and meaningful work. They can automate manual tasks and help people visualize and draft out their ideas in mere seconds. But to be applied reliably to professional, daily work, LLMs need to be implemented with a focus on a specific set of tasks, powered by a specialized set of training data to perform well.

If you were to use AI to help tax lawyers, you would need a deep understanding of tax laws, legal precedents, and regulatory frameworks.

For accountants, you would need to grasp financial reporting standards, auditing procedures, tax regulations, and data analysis techniques, with some context-specific judgments required in financial decision-making.

For salespeople, AI needs to understand sales strategies and frameworks, buying behaviors, product knowledge, market trends, and effective communication strategies, while also grasping the nuances of relationship building and negotiation tactics.

Why we're going all-in on Sales

We decided to focus on the users that needed to build presentations for work most often. We looked at our most highly retained users: Founders, Freelancers, Creatives, Marketers, and Sales reps. The biggest area of opportunity for creating demonstrable value for was clearly in Sales. Here’s why we think it’s a great opportunity:

• Revenue and growth are the lifeblood of every company, driving continuous investment in sales team efficiency to boost performance.

• Companies spend millions codifying sales frameworks and processes, aiming for perfect execution by their reps.

• Sales blends art and science, creating a data-rich environment where AI can learn and improve.

• Repetitive sales tasks can be automated, freeing up reps for high-value work.

• AI's strengths in pattern recognition and prediction align perfectly with sales needs.

• Clear metrics make it easy to measure AI's impact on revenue, making sales AI a high-ROI investment for companies.

This creates a clear opportunity for AI that deeply understands and enhances the sales playbook.

Bringing AI slides to Sales

We brought our generative slides product to the world of sales and go-to-market. It helped with filling RFP templates, generating slides based on call recordings (e.g. What We Heard From You), and updating customer logo walls. But before long, it became evident that this product wasn't ROI positive enough to warrant an organizational tool change.

• For larger teams, they’re already using a CMS like Seismic or Highspot for customization.

• For smaller teams, automated deck customization isn’t compelling because they don’t have a repeatable enough go to market to want AI customization.

• Even when we had deals that were far along, the idea of changing a tool that spanned different functions (eg. marketing, sales, customer success) got in the way of adoption

• We’ve made more progress with companies wanting to try Tome because they prefer the aesthetics; however, there’s not enough urgency or demand for this to be our core business.
  
  

Diving into the biggest priorities for sales leaders today

As we spent time with over 100 sales leaders, we realized that the most important part of storytelling in sales is great discovery, powered by great prospecting and great research. That got us thinking about how to help sales leaders and account executives with pain they felt the most.

The top priorities we heard were:

1. Bringing in as much qualified pipeline as possible

2. Expanding deal size by attaching to bigger problems

3. Bringing consistency to their sales process, raising the floor of AE performance

4. Using technology to drive efficiency and and find the biggest opportunities

We learned that successful sales starts from the top of the funnel. Effective pipeline generation is a massive competitive advantage.

If there can be a healthy pipeline from the beginning with influential buyers who need a solution to a burning problem, AEs can use their time efficiently on the accounts that matter — leading to higher win rates and more revenue.

Discovering top pain points for sales reps

We spent a considerable amount of time interviewing AEs to understand what's preventing them from nailing their leaders' priorities, and here's what we learned:

Analyzing the sales-tech landscape

Why existing solutions fall short

While “Contact Databases” and “Intent Signals & Contact Enrichment” tools help with building extremely detailed lists of accounts and prospects to go after with signals on which ones to prioritize, those signals just tell them when to reach out and how to reach out.

They simply aren’t enough to effectively break into an account, and focus on a spray-and-pray tactics that end up providing a worse customer experience, rather than quality of prospecting understanding.

What they clearly lack is who to reach out to, why to reach out to them specifically, and what to say that will actually move the needle. If you just tell a sales rep that someone viewed your product page or searched for your product — will it actually convert to a meeting?

To focus on the right accounts and create outreach that resonates, you need to deeply understand what’s going on in the account, what their priorities are, what their financials are like, what the organization looks like, who makes decisions in the organization, any recent news, and more. You need to then think of the best way to tie your product’s value proposition to what matters for the specific person you are targeting.

This “account research” work is done completely manually today, where salespeople are analyzing company 10Ks/financials, trudging through their CRM, listening to earnings calls or relevant podcasts, diving into their website & blog posts, clicking through different LinkedIn profiles, and even searching through community forums like G2 or Reddit.

Somehow, the most critical part to breaking into an account isn’t solved for today, and that’s because it was impossible until the latest advancements in LLMs.

Additionally:

• Outreach Sequencing tools mainly help with email deliverability and sequencing, they’re not as effective with who to reach out to, why to reach out to them specifically, and what to say that will actually move the needle

• Deal Execution give insights into calls that happened in the past — their core business isn’t to give insight into who to reach out to, why to reach out to them specifically, and what to say that will actually move the needle

• AI Agents have great promise, but right now focus on spray-and-pray tactics meant for Founders or very early stage companies who desperately need leads. They’re not meant for more established sales organizations who want to be precise about how they should break in. AI Agents are good at ingesting your sales playbook but lack abilities in account research across a multitude of online data sources

• GTM AI Platforms require a heavier lift from the customer-side and rely on off-the-shelf GPT models not tuned to the context of your sales playbook

Tome’s opportunity to create a wedge in SalesTech

We see an opportunity to use LLMs to acquire, process, and synthesize large bodies of company data to uncover “company-level” intent signals—helping you understand if this company needs what you’re selling.

By using a large context window and a model with extensive reasoning capability, we’re able to read thousands of documents and look for patterns. Combine that with web crawling and data enrichment, and we’re able to extract insights that would’ve taken days of research and effort in the past, that no sales rep has the time to do effectively.

We believe we can become a daily-use research and knowledge base for sellers, and one day their preferred interface to their system of record, where you might think of Salesforce as a data warehouse for storing important data rather than a primary interface you spend time in to answer questions.

Below are example company signals that can be found for some of our customers, that can only be found in aggregated research:

• Are execs talking about AI or do they have an AI center of excellence?

• Are they interested in “low code”, “RAD” or “digital transformation”?

• Do they have a Product Ops function (existing team or job openings)?

• What named initiative inside of a company can we sell into?

• National and international companies with a local strategy to sell ads into

• Does the company spend $3M+ on travel each year and is it growing?

• Is the company growing their cloud spend this year substantially?

• Does the bank have a blockchain initiative we can sell into?

• Is the company growing their headcount internationally?

• Is the company trying to move upmarket with their product or service?

• Does the company have a high traffic website where better site performance can create immediate ROI?

Our research engine will manifest into an AI-native system of intelligence for leaders to codify every part of their sales process and how the ideal rep should run a deal.

Tome: The assistant for enterprise sales

As we build on top of our “research engine” — we want to make this knowledge conveniently available and interactive for sellers. Therefore as an interface atop it, we will have applications that either completely automate work or augment decision-making across the entire sales process.

Eventually, we believe this approach can apply to several other verticals, once done right. We could help financial analysts deep dive into companies to drive trading decisions, or recruiters learn about someone’s background and ideal placements or roles, or even venture capitalists who need to do diligence on companies in the context of moving trends.

It’s exciting how broadly applicable the concept can be, and we’re confident that this is one of the ways the world will change as AI offerings mature — revolutionizing how we consume and process information.

The future can pull us into many directions, but the biggest learning we’ve had with AI is to start vertical to go horizontal.