An Hour a Day Was Evaporating — Why I Built Tasukiba, an AI Operations Secretary
“Wait — where was that file again?”
Someone in the meeting starts hunting for a document. Ten minutes pass. Six people are sitting around the table.
From the organization’s point of view, a full hour of payroll has just evaporated — not to make a decision, not to debate, just to locate a file. I think this is the moment I find most physically painful in work.
A meeting is supposed to be where opinions collide and decisions get made. Instead, the time that was reserved for creative thinking gets quietly burned by “wait, where was that…” — every single day.
That hour, evaporating daily, is what I’m trying to claw back with Tasukiba Knowledge Relay, the business SaaS I’m building.
”Great at storing data, weak at reusing it”
I spent years at my previous job building a low-code WebDB tool and supporting its deployment at customer sites. As a place to accumulate data, it was genuinely one of the best products I’ve ever used.
And yet, a quiet discomfort never left me.
Storage was excellent. Reuse was awkward.
Ten records, anyone can scan. A thousand records, and human eyes start dropping things. The data that gets missed is treated as if it didn’t exist — because the decision-maker’s eye never landed on it. I watched that happen, over and over.
Full-text search exists in every tool. But the moment a thousand hits come back and the UI essentially says “now sort them out yourself,” the search has effectively become an unused search.
Unformalized experience leaks out as someone else’s overtime
The second problem is harder to talk about: tacit judgment that quietly shifts the burden onto whoever’s downstream.
The PM (project manager) I currently work with is a genuinely kind person. So kind that, during user acceptance testing, they try to absorb feature requests on top of bug fixes.
“Take in defects, but don’t take in new specs — that path leads to regressions and quality collapse.” Anyone with experience holds that line as a gut feeling. The problem is that the line isn’t formalized anywhere. So when the PM’s kindness runs unchecked, no one can stop it.
The downstream cost is paid by developers: twelve-hour days, friction in the team, motivation bleeding out.
This isn’t a story I can shrug off as “one PM’s bad judgment.” What’s actually being lost is time, the team’s atmosphere, developers’ creative hours, and — on a longer horizon — the organization’s credibility.
And I don’t blame the PM, either. I think their kindness needed an objective check from past formalized knowledge — and that check just didn’t exist when they needed it.
What I’m actually trying to fix
If I compress my problem statement, it lands in one line.
Humans aren’t spending time on the work only humans can do.
Whether it shows up as wasted search time or as a kind PM’s judgment quietly draining developers, the underlying shape is the same. I touched on this in Reteal Organizations — what stops “self-managing teams” in the first place, where I wrote that delegation fails when the judgment criteria live only inside one person’s head. Same root, different surface.
Data on its own isn’t an asset. It becomes an asset the moment it gets reused. Which means Tasukiba isn’t really trying to “improve search precision.” It’s trying to surface buried past assets in front of the right person at the right moment, without them having to remember to ask.
Why I’m the one building it
The market has plenty of project management and knowledge tools. Notion, Jira, Confluence — all genuinely great products.
And yet, as far as I can tell, every one of them is better at storage than at retrieval. Full-text search collapses the moment data volume grows. I’ve seen this same pain across multiple sites, with different teams, different industries.
When you can’t find the tool you want, you build it. That’s the most honest answer I have for “why me.”
There’s a second reason. I’m aiming, longer-term, to be a project manager — and beyond that, to build the kind of psychologically safe organization I wrote about in Psychological safety isn’t reached by “knowing it”. Tasukiba is, in part, the apparatus I’m building to walk toward that.
Starting a series: “Launching a SaaS as a solo developer”
From today, I’ll be writing through Tasukiba’s philosophy, design, and operations across a long-running series.
| Chapter | Theme |
|---|---|
| A | Origin & philosophy (6 posts, including this one) |
| B | Product overview & release announcement (4) |
| K | Community, hiring, culture (4) |
| L | Post-release roadmap (3) |
| Bonus | Mascot / six-month retrospective (3) |
The public release is June 1, 2026. Until then I’ll be sketching the worldview; after, I’ll be writing about how the service keeps going.
It’s a deliberately quiet series. No hyper-growth narrative, no founder mythology. It’s aimed at people who genuinely want to reclaim the hour that evaporates from every workday — and don’t believe willpower is the right tool for the job.
Next post: the same problem, reframed as changing the texture of Sunday evening and Monday morning.
FAQ: Reclaiming the time you spend searching
Q. How can I cut down the time spent hunting for documents and past information?
A. The shortcut is to change the very structure of “going to look for it yourself.” Tasukiba remembers your past projects, knowledge, and retrospectives, and proactively surfaces what you need now through semantic search. Because it matches on meaning rather than exact keywords, you skip the whole step of “remembering where it was.”
Q. Why does project knowledge fail to accumulate — or stay impossible to find?
A. Usually because it is scattered across chat, email, and Excel, and search doesn’t surface it. Information stored is not yet an asset; it only becomes one when it is handed to you at the moment you need it. The “one hour evaporating every day” in this article is exactly what that “can’t-find-it” problem adds up to.
Q. How do I carry retrospectives forward into the next project?
A. The key is keeping retrospectives (KPT, risks, and issues) in a state where they resurface automatically in similar situations later. Tasukiba keeps past retrospectives within reach of semantic search and surfaces the relevant ones, along with why they are related.
Related posts
- Trial run of Tasukiba Knowledge Relay — a relay baton for project knowledge — the first public announcement
- Reteal Organizations — what stops “self-managing teams” in the first place — the structural cost of tacit knowledge
- Psychological safety isn’t reached by “knowing it” — the “organizational dream” Tasukiba is walking toward
About Tasukiba
Tasukiba Knowledge Relay is a business SaaS that semantically surfaces past project assets when you need them. Public release: June 1, 2026. See the product page for the full picture.
