How Will Sales AI Agents Change the Field? How They Differ from SFA and CRM

You introduced SFA or CRM, yet entering data after meetings, writing minutes, and following up still eat into the time you have for customers. A sales AI agent takes on this surrounding work so that salespeople can focus on conversations with customers.

In this article, we draw on the conversation at our joint webinar held on October 29, 2026, to cover how sales AI agents differ from SFA and CRM, three steps to make adoption stick, and the role salespeople will play from here.

Key points of this article

• A sales AI agent does more than record information like SFA and CRM: it carries out part of the work, such as meeting preparation, records, and follow-up drafts
• Three steps help adoption stick: decide what to delegate, define the conditions for judgment, and measure correction effort to improve
• Keep human review and approval for anything sent outside the company and for important decisions, so salespeople can spend their time with customers

A joint webinar on sales AI agents

Aitane Inc. held an online conversation on sales AI agents together with Iidabashi Cross Partners Inc., the corporate venture capital (CVC) arm of the PCA Group.

Item	Details
Title	How Will Sales AI Agents Change the Field? A CVC and a founder on where adoption stands and what comes next
Date and time	Thursday, October 29, 2026, 12:00 to 13:00 (JST)
Format	Online (Google Meet)
Co-hosts	Iidabashi Cross Partners Inc., Aitane Inc.
Speakers	Masaki Hasegawa (Representative Director, Iidabashi Cross Partners Inc.), Junichiro Aita (Representative Director, Aitane Inc.)

Drawing on his experience in back-office operations and corporate planning at a business software company, Mr. Hasegawa set out the challenges companies face when introducing AI, from the viewpoint of a CVC. Aita, from the side that develops a sales AI agent, explained how to think about handing work over to AI.

Below, we introduce the conversation in four themes.

How sales AI agents differ from SFA and CRM

The first theme was how sales AI agents differ in role from SFA and CRM, which many companies already use.

From systems of record to systems of action

Aita began by contrasting the roles the two play.

SFA and CRM are systems of record that store meeting history and customer information. An AI agent, by contrast, can be seen as a system of action that takes over part of the work people have done until now.

Item	SFA and CRM	Sales AI agent
Main role	Record and store information	Carry out part of the work
Who enters and drafts	Mainly salespeople	AI organizes and drafts, people review
Expected effect	Centralized information	Turning records into the next action

Note: Organized by the editorial team based on remarks at the webinar

What matters is not entering information for its own sake, but connecting it to the next action. For the basics of how a sales AI agent works, see What Is a Sales AI Agent?.

Examples of sales work AI can take on

Concrete examples include organizing information about the other company before a meeting, recording what was discussed, and drafting follow-up emails.

As an example of taking on actual work, Aita also introduced a development case in construction cost estimation, where AI reads a large volume of specifications and design documents and compiles a rough estimate. It shows the idea of handing over the work itself, not only generating text.

That said, an AI agent cannot run every sales task automatically and without conditions. The premise is to design what to hand to AI and where people check the results.

Three steps to make AI agent adoption stick

Many companies introduce AI only to stop at having tried it. Aita explained three steps that connect AI to real improvements in work.

Mr. Hasegawa also pointed out the need to design work across the whole organization, rather than leaving it to partial use by individuals. Even if one task becomes faster, the work as a whole improves little if manual steps remain before and after it.

Decide what work and judgment to delegate

The first step is to make clear what AI should judge and which work it should take on.

Instead of trying AI for its own sake, make the purpose and the target work specific. A good starting point is to list sales work one item at a time, such as meeting preparation, meeting records, customer updates, and follow-ups.

Define the conditions and rules for judgment

Next, organize the conditions AI needs in order to process the work.

At the webinar, Excel formulas and the definitions of figures were used as examples of why the criteria for judgment must be made explicit. With vague criteria, it is hard to evaluate consistently whether AI output is correct.

Measure correction effort and improve

The last step is verification.

Aita questioned the idea of measuring AI use only by the number of tokens consumed. What matters is how much people had to correct AI output and how long checking took.

Look back at whether the actual workload became lighter, and keep improving while revisiting the conditions and rules.

Balancing automation with human approval

The more you hand to AI, the more it matters where people stay involved. The conversation covered both keeping work flowing and keeping human judgment in place.

Keeping the flow of work moving

Sales work runs as a series: research before a meeting, the meeting itself, records, customer updates, and follow-up.

Mr. Hasegawa said that if people no longer have to stop again and again along the way to enter or check data, AI would likely settle naturally into daily work.

Rather than switching between separate tools, having the series of tasks move forward as one is a condition for staying in use on the front line.

Keeping moments for human review and approval

On the other hand, not everything should be left to AI, from replies and proposals to customers to decisions about contracts.

Mr. Hasegawa pointed to concerns about security when companies use AI and to situations that require human checks. Aita agreed, saying it is important to design an experience where AI drafts and people review, edit, and approve.

Especially for information sent outside the company and for important decisions, it helps to design the approval flow as well. Expanding automation and keeping human judgment are not in conflict; they are meant to work together.

The role of salespeople in the age of AI

The last theme was the role of salespeople once AI supports their work.

Conversations and trust with customers

When AI supports preparation and administrative work, what should people spend their time on?

Aita said he wants people to be able to spend their time on the work people should do, with people handling time with customers and AI taking on the preparation, records, and paperwork around it.

Mr. Hasegawa likewise expects that as automation advances, salespeople will put more energy into building trust with customers and into deeper conversations to understand their needs.

What an investor looks for in an AI agent

What Mr. Hasegawa valued as an investor was not technical novelty alone.

He said he sees potential in AI agents that understand a company's complex work and information and function as actual members of the workforce.

Not how many times AI was used, but what work moved forward for customers and the organization. That question looks set to become the yardstick for adopting AI agents.

Frequently asked questions about sales AI agents

Based on the conversation, we answer questions often raised by people considering a sales AI agent.

How does a sales AI agent differ from SFA and CRM?

SFA and CRM are systems for recording and storing meeting history and customer information. A sales AI agent differs in that it carries out part of the work, such as organizing information before meetings, recording meetings, and drafting follow-up emails.

Where should we start to make adoption stick?

First, decide which work and judgment to hand to AI. Next, define the conditions and rules for judgment, and improve while measuring how much people had to correct AI output.

Can we leave all sales work to AI?

There is no need to leave everything to AI. For information sent outside the company and for important decisions, such as replies and proposals to customers or contract decisions, it is important to keep a flow where people review and approve.

Summary: start with an inventory of sales work

What the conversation showed is that, when considering a sales AI agent, it matters more to look at the work you can actually hand over than at the number of features.

• SFA and CRM handle records, while an AI agent handles carrying out work
• Decide what to delegate, define the conditions for judgment, and improve while watching correction effort
• Advance automation while keeping human approval for external communication and important decisions

At Aitane, we see AI not as something that takes away people's work, but as support that lets salespeople spend their time talking with customers. Start by taking stock of your sales work and checking which tasks take the most time.

Watch the full conversation on demand

The recording also covers points we could not include here and answers to questions from participants. If you would like to watch it, please apply through the free form (in Japanese).

Apply to watch the recording (free)