Skip to content
TL;DR
- The problem: At Speee, every closed deal had to be handed from sales to Customer Success, and that ran entirely on text. Sales flagged the win, CS sent a form, sales filled it in, CS asked follow-ups. Dozens of rounds per account, about 30 minutes each, with new contracts landing every month.
- Why it kept failing: The notes carried each rep’s own interpretation, and a day or two passed before the handoff, so memories had already faded. CS often met the customer to find the details didn’t match reality, and onboarding had to start over.
- What they did: Recorded every sales call in tl;dv. CS now pulls handoff notes straight from the transcript using Ask tl;dv AI, with over 20 purpose-built prompts, then confirms the output with the rep in a single pass.
- The results: Handoff effort down two-thirds. Onboarding completion 23 days faster. Feature adoption up 15%. Seven more deal opportunities per rep, per month.
“If we could just share things on a factual basis, the problem would be solved. Once we reached that hypothesis, we decided to adopt tl;dv,” says Taiki Minami, Customer Success (CS) lead for Speee’s in-house SaaS product. The organization had been spending enormous effort on sales-to-CS handoffs alone. When they started using recorded meeting logs — raw facts — as their foundation, everything began to change. Here is the story of how they put it into practice.
Company Profile & Background
Speee, Inc. was founded in 2007 in Japan. The company operates 24 businesses ranging from marketing DX consulting for enterprises to DX initiatives in legacy industries. Minami works in the home renovation DX division, where he serves as the first-ever Customer Success hire for Budii, the company’s proprietary SaaS sales enablement tool for the renovation industry. He also oversees CS strategy and operations design.
The trigger for adopting tl;dv was a handoff problem between sales and CS. Post-deal information sharing relied on back-and-forth text exchanges, creating bottlenecks in both effort and quality. To structurally solve this issue, the team decided to build a system based on recorded meeting logs of every sales call.
The Challenge Before tl;dv: Endless Text-Based Back-and-Forth Was the Norm
Speee’s Budii division runs a “The Model”-style organization: Marketing → Inside Sales → Field Sales → Customer Success (CS). In this model, the first critical gate that determines CS outcomes is the post-deal handoff from sales.
Before tl;dv, handoffs were entirely text-based. Sales would send a message saying “We closed the deal,” CS would send a template form, sales would fill it in, and CS would review it — a cycle that sometimes ran into dozens of rounds per account.
“Even when the form came back saying ‘The goal is to make sales activities more visible,’ we needed to understand why they wanted visibility and why they couldn’t achieve it now. We had to keep going back and forth to get to that depth. Doing it over text was incredibly tedious. We’d also set up face-to-face meetings, but even then, it took about 30 minutes per account.”
With new contracts coming in consistently every month, handoffs of 30 minutes each added up, eating heavily into the time CS should have been spending with customers.
But it wasn’t just about time — quality was also a problem. Meeting notes written by individual sales reps varied wildly, skewing either too detailed or too vague. When CS finally met the customer, the information they’d received often didn’t match reality, forcing them to redo the onboarding from scratch.
Minami boiled the root cause down to three points: sales reps’ subjective interpretations were mixed in; memories faded when one to two business days passed before the handoff; and on the receiving end, CS had no choice but to rely on their own interpretation. He then identified the common thread.
“It all came down to one thing: facts weren’t being shared. Flip it around, and if we could just share facts, the problem would solve itself. That’s the hypothesis that led us to adopt tl;dv.”
Why tl;dv?
Three services made the shortlist: tl;dv and two domestic tools.
The first candidate was ruled out due to Speee’s security governance — services that use customer data for AI training were not an option. “To opt out of AI training, you needed their enterprise plan, which pushed the price up significantly. That’s why we dropped it.”
The second candidate was primarily focused on sales rep skill development, which didn’t fit the team’s top priority at the time: fixing the handoff process. “For handoffs and meeting documentation, tl;dv had the best transcription accuracy and the easiest way to extract information. Above all, the Ask tl;dv AI feature was the deciding factor.”
After narrowing it down to tl;dv, the team ran a three-month trial on a monthly plan. To get the purchase approved internally, Minami estimated the handoff time savings and translated them into projected revenue from the additional sales capacity freed up. However, with AI tools emerging rapidly at the time, there were concerns about whether the tool would deliver and whether a cheaper alternative might appear. Instead of committing to an annual contract right away, the team opted for a three-month validation period.
The three-month trial confirmed the results, and the team moved to an annual contract. With outcomes exceeding expectations, they expanded to a larger number of licenses than originally planned.
What They Did Was Simple: Record Every Sales Call in tl;dv
“What we did was really simple — build our entire meeting log infrastructure on tl;dv. That’s it.”
Here’s the workflow. When a sales rep conducts a call on Google Meet, tl;dv automatically stores the recording. CS can then use the Ask tl;dv AI feature at any time to submit a prompt and extract the information they need. They show the output directly to the sales rep and ask, “Does this match your understanding?” — and that’s the end of it.
The dozens of text-based back-and-forth rounds disappeared. Sales reps no longer needed to write meeting notes by hand. CS could now pull the information they needed, when they needed it, based on facts.
Ask tl;dv AI: A Feature That Works Because the Facts Are Preserved
At the heart of this system is tl;dv’s AI feature. Because the raw source material — meeting recordings and transcripts — is preserved as-is, the team can extract what they need from any angle, at any time.
“When meeting notes are written by a person, their interpretation gets baked in. But when the recording and transcript — the facts — are preserved, AI can extract only the information you need. That’s what makes such a big difference.”
That said, simply deploying the tool didn’t mean the team adopted it overnight. During the initial onboarding, the focus was on explaining concepts, and it took time for members to experience the “aha” moment.
Minami took it upon himself to create different prompts for each type of meeting and personally distributed them to the team, again and again.
“When people actually see it in action and operate it themselves, it clicks — ‘Oh, this is actually useful.’ Getting hands on the tool mattered more than explaining the concept.”
Today, Minami uses over 20 different prompts: for new deal handoffs, regular check-in meetings, accounts with rising churn risk — each with its own specialized prompt.
His process for creating prompts is distinctive. First, he thoroughly analyzes the factors behind customers who achieved early success. From the perspective of “if we’d had this information at the handoff stage, things would have gone better,” he identifies what’s truly essential. He then feeds those requirements into a generative AI tool to produce a first draft of the prompt. That draft gets tested in tl;dv — sometimes a dozen times, sometimes dozens — while he refines and adjusts. Only when the output meets his standards does he share the final prompt with the team.
“Creating prompts was a pretty high hurdle at first. But once you’ve built one, it becomes an organizational asset. The key is to build them so the whole team can use them.”
Minami follows four rules for prompt design:
① Eliminate inconsistencies in output format. Specify everything down to font size and whether emojis are allowed, so the output is uniform every time. Without this, downstream processing becomes difficult.
② Pre-fill proper nouns with selectable options. Since tl;dv doesn’t use data for external AI training, accuracy for company names and other proper nouns can drop. Setting them up as selectable options in advance helps the AI return the intended results.
③ Structure each item as a “three-level breakdown.” Instead of long sentences, design prompts so information comes back short, dense, and in a consistent format — a structure that lets readers absorb it quickly.
④ Include the source recording timestamps. Have the AI output the timestamps of the statements it based its analysis on. “Sometimes you can’t gauge a customer’s emotional temperature from text alone. Being able to go back and listen to the recording — that’s the most important part.”
After Adoption: Not Just CS — Sales Changed Too
The impact showed up on both the CS and sales sides.
CS handoff effort was reduced by two-thirds. With higher-quality handoffs, onboarding no longer needed to be redone from scratch, and time-to-onboarding-completion was shortened by 23 days. Feature adoption rates also improved by 15%.
Sales changed too. With the time previously spent on handoffs now freed up, reps could redirect it to actual sales conversations, resulting in seven additional deal opportunities per person per month.
“By doing handoffs based on recorded facts, CS’s resolution when first engaging a customer is dramatically higher. You already know ‘this person was concerned about X’ before you even make first contact. Starting from that point has a direct impact on onboarding quality.”
The Next Challenge: Turning tl;dv into a “Sales Coach” — Evolving into a Skill Development Tool
What Minami envisions going forward goes far beyond handoffs and meeting notes.
“At first, I thought tl;dv was a meeting notes tool. But as I kept using it, I realized it could evolve into something like a senior colleague or coach for sales calls.”
What he’s currently testing is a “virtual manager” system powered by Claude. The idea is to have Claude adopt the persona of a manager, with tl;dv meeting data automatically fed in on a regular basis. The virtual manager would then provide feedback like “Your goal-setting lacked quantitative rigor in this call” or “Here’s the action you should take next” — all generated automatically.
The essence of this initiative isn’t scoring — it’s skill development. New team members could have experiences similar to shadowing a senior rep’s sales calls, delivered on-demand through AI. No need to borrow a senior colleague’s time. No need to coordinate schedules.
“In an era where understaffing is the norm, the key to survival is knowing which tasks to delegate to AI and which require a human touch. The fewer people you have and the more results you need, the more you should adopt tl;dv. It creates an environment where you can devote all your resources to the things only humans can do.”
Summary
For Speee, tl;dv started as a tool to fix a handoff problem. But its true value lies in preserving every sales conversation as a factual log and creating an environment where the organization can extract exactly the information it needs, whenever it needs it.
The answer Minami arrived at is simple: keep recording the facts. From there, sales and CS operate from the same information, and customers reach value faster. Beyond that lies an even bigger possibility — using accumulated meeting data as the foundation for organization-wide skill development.
“I think tl;dv has so much more potential beyond handoffs and meeting notes. I’d love for people to keep that broader vision in mind as they use it.” Minami’s words carry the conviction of someone who has lived the journey from tool user to platform thinker.
