Thira Raises $21M Seed Round to Build Enterprise AI Execution Layer

Thira has come out of stealth with a $21M seed round. The Bellevue, Washington enterprise AI startup was founded by Apptio founders Sunny Gupta and Kurt Shintaffer, who spent years helping large companies understand technology spending before turning their attention to the next problem: getting enterprise work to actually execute.

The company is not pitching another chatbot dressed up as enterprise software. Thira is building a secure, self-learning AI system of execution for back-office workflows, the kind of repetitive work that moves through IT, finance, HR, procurement, and operations without ever becoming glamorous. The seed round matters because enterprise AI investment is moving away from demos and toward systems that can complete governed work across the tools companies already use.

What Happened

Thira announced $21M in seed funding on July 14, 2026. The round was led by Madrona with participation from FUSE and a group of global advisors and CIOs. The company emerged from stealth as a new enterprise AI startup focused on automating back-office execution.

The founding team also includes enterprise software operators from companies such as Databricks and Atlassian. Tarek Madkour serves as Chief Product Officer, while Mudit Goel is part of the company's senior engineering leadership. Thira is still early, but the founding pattern is clear: the company is leaning on enterprise software credibility rather than consumer AI noise.

The round will support product and engineering hiring, deeper work with enterprise design partners, and preparation for broader availability later in 2026. The company says it is working with roughly 10 enterprise design partners ahead of broader availability later in 2026.

Why This Matters

The first wave of enterprise generative AI was built around conversation. Companies tested assistants, copilots, summarizers, and search layers that could answer questions faster than the old internal wiki ever could. Those products improved access to information, but they still left the hard part with employees: someone had to take the answer and push the work across systems.

Thira is going after that handoff. The company describes its product as an AI system of execution that can operate across existing enterprise applications, learn from workflow context, and complete multi-step back-office tasks with governance and visibility. That is a very different buyer promise from "ask better questions." It is closer to "make the process run without turning the CIO into a risk manager with a fire extinguisher."

That distinction is why this funding round is worth watching. Enterprise buyers do not pay premium software prices because a demo looks clever. They pay when a product removes work, lowers risk, and fits into the controls they already answer for every quarter.

Market Context

Large organizations spent the last decade buying systems of record and systems of intelligence. They can see more, measure more, and report more than ever, but many operational workflows still depend on humans stitching together ServiceNow, Workday, Salesforce, Jira, SAP, email, spreadsheets, and approval chains that were never designed to move as one process.

The opportunity is not simply to build a better AI interface. It is to build a governed execution layer that can sit across fragmented systems and turn procedural work into something closer to infrastructure.

That is also why the Apptio lineage matters. Apptio helped enterprise technology leaders understand spending, operations, and accountability. Thira asks the next question in that sequence: once a company understands the work and the cost, why should so much of the execution still depend on manual coordination?

Competitive Landscape

Thira enters a crowded enterprise AI market, but its position is narrower than the category label suggests. The company is not trying to be a general-purpose productivity assistant or a front-office content machine. Its stated focus is back-office execution, including workflows such as onboarding, device provisioning, account resets, software approvals, and other operational tasks that touch multiple systems.

That focus creates a harder product problem and a clearer enterprise value proposition. If Thira can prove that self-learning agents, an enterprise knowledge graph, security controls, and workflow orchestration can safely complete recurring work, the platform becomes less of a novelty layer and more of an operational substrate. Enterprise software companies love that kind of boring sentence because boring is where the budget usually lives.

The design-partner strategy also matters. Rather than chasing public logos before the system is ready, Thira appears to be refining the product with a small group of enterprise operators who can expose edge cases early. For AI systems expected to act inside sensitive workflows, that discipline is not a luxury. It is the price of trust.

What This Signals

Madrona leading the round signals continued investor appetite for repeat enterprise founders with deep CIO relationships. FUSE adds another Pacific Northwest enterprise software investor to the syndicate, while the participation of advisors and CIOs suggests the company is turning buyer trust into product feedback from the outset.

The bigger signal is about where enterprise AI capital is moving. The market is growing less impressed by models that can describe work and more interested in systems that can execute it inside real companies. That shift puts pressure on startups to prove integration depth, governance, auditability, and operational reliability instead of relying on broad claims about automation.

It also rewards founders who understand how slowly enterprise trust develops. Gupta and Shintaffer are not selling AI into a vacuum. They are returning to the same class of customer they spent years serving through Apptio, this time with a thesis that the next platform layer will not just measure work. It will do more of it.

The Bigger Industry Shift

Thira's $21M seed round is part of a broader move from AI as an answer engine to AI as an execution layer. That shift is where the enterprise market becomes more serious and less theatrical because products have to survive procurement, security review, workflow complexity, and the quiet terror of being responsible for a system that acts instead of advises.

If Thira succeeds, the meaningful innovation will not be that an AI agent can talk like a consultant. It will be that enterprise software finally starts absorbing the repetitive operational work people have normalized for decades. The best version of this market does not make companies louder. It makes the back office quieter, faster, and less dependent on heroic manual coordination.