Solving two long-stuck problems during a wait at the airport — how a non-engineer self-built Salesforce with Blaze

About Rhizome Co., Ltd.

— Tell us about your business.

We're a software development company of around 100 people. We handle everything in-house, from planning to development to operation and maintenance, and our biggest distinguishing feature is that we specialize in commercial facilities such as shopping centers and department stores.

Our flagship product is a data-analytics package called "Senryaku Kaigi Next." It lets people running commercial facilities visualize sales data with nothing more than clicks, and we designed it to be intuitive even for people who aren't highly IT-literate. We also provide services centered on building and rolling out data-analytics environments using Tableau.

Adopting Blaze to drive the build toward full-scale Salesforce operation with zero engineering resources

— What led you to adopt Blaze?

We were moving toward full-scale Salesforce operation starting March 1 of this year, and we were doing the build entirely ourselves, without using an implementation partner. Until then we'd used something like a third-party business-card management tool with SFA and MA bolted on, but we decided to switch over to Salesforce.

I was doing the build itself in spare moments while handling other work as well. Standard objects like the Opportunity object came with handy features preset, so those went relatively smoothly, but the moment we tried to realize our own custom specifications, situations arose that required knowledge of Flow and Apex.

What I struggled with most back then was billing management. At our company Salesforce also doubles as sales management — we manage billing amounts down to the yen inside Salesforce, and we were trying to realize a data flow where the CSV of that billing object goes straight into our accounting system. But billing management had to be built out with custom objects, and once we started nailing down the details around January, just before go-live, improvement points kept popping up one after another: "I want to automate this process," "I want this screen to look like that."

I come from an IT-solutions sales background — I'm not an engineer. When it came to situations that required building Apex or complex Flows, I was in no position to handle it myself. Just as I was feeling stuck, with no time left before the March 1 production launch, a Salesforce representative happened to introduce me to Blaze, and I looked into adopting it to make the launch in time.

More than just giving hints — operating directly on the live environment is the line that separates it from general-purpose generative AI

— What was the deciding factor in adopting it?

I was introduced to Blaze while on a business trip to Fukuoka. The very day I was told "an amazing tool called Blaze has come out," I installed it on the spot and tried it during the wait at the airport on my way back to Tokyo. In that hour and a half of waiting, Blaze developed and shipped to release the two things I'd been agonizing over how to build — a complex Flow and a screen I needed to add. And it did so within the trial conversation limit, with Blaze running on its own and completing the release in the background while I worked on something else. In that moment, I was convinced: "Let's adopt Blaze right away."

At about 30,000 yen a month, for just one person like me to use, the bar to adopt isn't that high. There was no need to even weigh the cost-effectiveness — the decision was easy.

— Compared with other AI tools, where were the advantages unique to Blaze?

Blaze has two strengths. One is that "it accesses and operates the live environment directly." The other is that "its knowledge of Salesforce is extremely deep."

If it's just a matter of checking Salesforce specifications or how to write a formula, general-purpose generative AI can do that to some extent too. In fact, before adopting Blaze I'd ask a lot about Salesforce configuration, and had it create copy-and-paste-ready formulas and the like.

But general-purpose generative AI can't reach into our own Salesforce live environment. Blaze answers based on how our org is actually built, and can even handle the implementation. That's the decisive difference.

On top of that, it also has the characteristic of being specialized for Salesforce, so I feel its problem-solving ability is extremely high.

From "how should I implement this?" consultations through to implementation — you can hand the entire Salesforce build over to it

— How are you actually using Blaze?

In the build phase, I started from consultations like "I want to do this — what implementation approaches are there?" The flow was: have it present multiple implementation patterns, hear the reasons it recommends each, get it to tell me the missing pieces for the pattern I picked, and then ask it to "implement it."

For example, I had Blaze build a process that reads the text of an Activity object and automatically reflects it into the Opportunity object. It's a mechanism that automatically registers what a sales member wrote in an Activity into each Opportunity field and leaves the change history in Chatter. It was a process combining Flow and Apex, but I left everything to Blaze, from consultation through implementation.

I also had Blaze build, from object creation all the way to field configuration, a custom object for accumulating VOC (Voice of Customer). It handles not just small additional implementations but the work of "standing up an object from scratch," so it was a huge help during the build.

Even the "pitfalls" invisible to a non-engineer — Blaze notices and deals with them

— You also used it for data migration. What did that look like specifically?

Blaze was a tremendous help with migrating data from the old system, too. The procedure required downloading data from the old system, cleaning it up locally, and uploading it to Salesforce, and I left that process to Blaze.

When you clean up data in Excel on a Windows PC, the character encoding can end up being Shift-JIS. Salesforce recommends UTF-8, so trying to upload it as-is throws an error. I didn't even have the notion of "watch out for character encoding" in the first place, but Blaze autonomously handled it, telling me "the character encoding doesn't match, so I'll convert it." Something an engineer would naturally be conscious of was a complete blind spot for me as a non-engineer.

For data in objects with parent-child relationships, when I handed it over split across multiple Excel sheets, it processed each one automatically.

What helped most was matching that doesn't use Salesforce IDs. The old system's data didn't contain Salesforce IDs, so it matched and reconciled the data using strings such as account names. It would list out the number of unmatched records — "30 records were unmatched" — and for records with similar names it would even suggest candidates: "Isn't it this one?" It really felt like having a talented engineer sitting right beside me.

From bug investigation to spec checks — for anything Salesforce, I ask Blaze first

— Tell us how you use it in the operation phase after go-live.

Even after going into production in March, as you'd expect in a "launch period," improvement requests keep coming in — "that should be this way," "this would be easier to use like that." Blaze helps when we make those revisions, too.

What's surprisingly helpful is bug investigation. For example, when a problem like "a field I newly created isn't showing up in a report" came up, I asked "with this report ID, this field on this object isn't displaying — why?" and it immediately identified that it was a permissions configuration issue. Fix the setting, problem solved — I can wrap that up in a single exchange.

If I'd asked generative AI about this, I might have spent about 20 minutes going back and forth, and if I'd tried to open a case with official support, it would have taken 10 minutes to create the case and the answer would have come the next day. With Blaze, what you want to solve really does get "solved in seconds." Thanks to that, I can move work along with a sense of speed.

— What changes did you see in your work and organization after adopting it?

Basically, I've reduced how often I reach out with questions or look things up with generative AI when there's something I don't understand or want to consult on. That's because, in many cases, just asking Blaze first is faster and more accurate. That's how much I rely on Blaze.

Going forward, we're also considering using marketing automation (Marketing Cloud Next) running on the Salesforce Data Cloud foundation. We want to use it in parallel with Account Engagement and ultimately consolidate into the core. I'm hopeful we can put Blaze to work on that build as well.

For those considering adoption

— A message for others self-building Salesforce in the same way.

I'm not an engineer, and it's not as if I have specialized Salesforce knowledge. Someone like that was doing self-implementation on the side, but with Blaze I could proceed feeling like "there's one talented engineer right next to me."

Here's the kind of image I have for how to use it. Before a web meeting starts, you toss a heavy-looking data-migration task to Blaze, and during the meeting you focus on that. When you glance at the screen, the processing is steadily moving along, and an hour later when the meeting ends, it's done — you can use it that way.

If you're stuck thinking "I can't write Apex, so it's impossible" or "Flows are complex and I don't get them," try consulting Blaze first. From proposing implementation methods to the actual build, it'll run the whole thing on a chat basis. For anyone trying to build Salesforce on their own, it should be the biggest help of all.

——

※ This article is based on an interview conducted in spring 2026.