Software meant to fit the way a business actually runs.

The problem we keep returning to

Most software is built by technical people, for technical people. The rest of us are left running our companies on a daily improvisation that happens around the SaaS we bought, the processes we invented on the fly to make that SaaS work, and a business vocabulary that came with the software and rarely fit the way we actually run.

The result is a quiet tax on attention. The missed follow-up. The Friday afternoon meeting whose only purpose is to bridge what the system says and what the team actually did. The new hire whose job is to translate between them.

We've been working on this problem in different forms for years. Now we are trying to see how far AI can take it.

What we've put out publicly so far

Our first open experiment is Things Have History, a 15-category history publication that an autonomous editorial engine runs end to end. It picks the topics, does the research, drafts the essays, generates the cover illustrations, narrates the audio, and ships everything to the live site without a human in the loop. If something turns out to be wrong, we correct it after the fact rather than reviewing before.

It runs in public on purpose, with the parts that don't work yet visible alongside the parts that do. That is a quieter way to learn about a system than running it behind closed doors.

Why a publication is our public showcase

The same engine architecture that runs Things Have History is what we are building for internal operating layers. Both are pipelines that take a question, do the research, assemble the artifact, ship it, and correct it after the fact when something is wrong. A publication is the version of this work we can run fully in public without exposing client information, which is why it became the demonstration we point people to. If the engine can keep a 15-category weekly publication coherent, readable, and factually careful on its own, the same engine can keep the moving parts of a business coherent too.

Read what the engine has been writing →

Who we are

Anand Krishnan and Sai Ganesh met in college in Bangalore in 1992 and have been friends since. We started our first company together, thinkbridge, in 2010. ZTRIC, which we co-founded in 2015 and where we now serve on the board, was our second. ZPQV, which we launched in 2024, is our third.

We launched ZPQV around two questions we kept returning to. The first was whether AI had reached the point where the kind of operating-layer work thinkbridge has been doing by hand for fifteen years could be turned into products that a non-technical operator could pick up and use on their own. The second was whether that work could be done from India by the team we already had in mind. We thought yes to both, and ZPQV is how we are trying to act on that. It is the products side of the same long question thinkbridge has been working on as a services firm.

Sai and Anand live in the US. The rest of the team is in India, working day to day alongside a set of AI assistants and agents we keep adjusting as we learn.

What we have learned at thinkbridge

The pattern on every thinkbridge client engagement has been the same. We sort through the client's existing SaaS and spreadsheets with them, deciding what to retire and what to keep as a system of record. Then we build a custom operating layer around what stays, shaped to how that specific business actually runs, in its own vocabulary. The companies we have worked with have generally gone on to higher valuation or better operating efficiency afterwards, which is the part that mattered to them. ZPQV is our attempt to turn that pattern into products an operator can run on their own, instead of a service the thinkbridge team performs by hand for one client at a time.