A Familiar Team Places a New Bet
On July 14, Thira emerged from stealth with a $21 million seed round led by Madrona Venture Group, with participation from FUSE. The draw here is the pedigree as much as the capital. Sunny Gupta and Kurt Shintaffer, who co-founded technology business management pioneer Apptio and sold it into a multibillion-dollar exit to IBM, are back together, with Gupta serving as co-founder and executive chairman. Matt McIlwain, the Madrona managing director who backed Gupta at both iConclude and Apptio, is joining Thira's board, and FUSE founding partner Kellan Carter comes on as a board observer, reassembling a proven investor-founder alliance.
That shared history is the reason this seed round deserves attention beyond its dollar figure. Apptio built a durable enterprise software category by attacking an unglamorous operational problem, the management of technology spend, with rigor and disciplined executive-level selling into the office of the CIO. Thira is now aiming that same playbook at agentic AI and the back office. When founders with a proven enterprise exit choose to reunite around workflow automation rather than a flashier consumer or model play, it is a useful signal about where seasoned operators believe the near-term, defensible money in enterprise AI actually sits.
What Thira Is Building
Thira describes its product as a secure, self-learning AI system of execution for enterprise back-office operations, initially focused on IT. In plain terms, it targets the invisible work that keeps large companies running day to day: setting up a new hire's laptop, resetting a locked account, approving a software purchase, and the many multi-step processes that quietly touch several systems along the way. The platform combines self-learning agents, enterprise-grade security and governance, connectivity across existing applications, and execution powered by an enterprise knowledge graph that encodes how a given organization's processes and systems actually fit together.
The framing choice is the phrase system of execution, and it is doing deliberate work. Plenty of tools already offer visibility, dashboards or suggestions, and Gupta is explicitly positioning Thira past that crowded line into acting on the work itself. He argues that CIOs are realizing the real opportunity is executing complex processes that touch disparate systems with enterprise-level controls. That is a direct claim on the contested ground between a passive chatbot and a fully autonomous agent, the zone where tasks actually get completed under supervision and permissioning that a large, audited organization is willing to accept in production.
The Team Behind the Product
Thira has assembled a founding group that blends enterprise delivery experience with hands-on AI engineering. Beyond Gupta and Shintaffer, the early team includes Mudit Goel, a former senior vice president of engineering at Atlassian, Grant Neuman, a former AI engineer at Oracle Cloud Infrastructure, Tarek Madkour, a former director of product management at Databricks, and Shashwat Srivastav, a former founder and lead AI engineer at AIVOT. McIlwain called the group the perfect blend of proven enterprise product builders and AI-native innovators assembled to deliver durable solutions, which reads as an investor betting on execution capacity as much as on the idea itself.
We put real weight on that composition. Enterprise back-office automation fails more often on integration, security and change management than it does on raw model capability, and this team is stacked with people who have shipped at exactly that layer inside demanding organizations. The combination of Apptio-era enterprise credibility with engineers drawn from Atlassian, Oracle and Databricks suggests a company deliberately built to sell into skeptical, control-minded IT organizations. That is precisely the constituency that ultimately decides whether AI agents earn production access or stay penned in a sandbox, and it is a hard audience to win with a demo alone.
Why IT Operations Is the Wedge
Starting with IT operations is a shrewd entry point rather than an obvious one. IT service desks handle high volumes of repetitive, well-documented tickets, provisioning, access resets and approvals, that follow known rules and touch systems the IT team already owns and understands. That makes them close to ideal for agents: the tasks are frequent enough to justify automation, structured enough to execute reliably, and governed by the same team that would sign the purchase order. Early success there produces measurable ticket deflection and cycle-time gains that build internal credibility and fund expansion into adjacent domains.
The broader architecture also matters to the strategy. By grounding execution in an enterprise knowledge graph and connecting across multiple systems, Thira is building for the reality that back-office work rarely lives inside a single application. A routine password reset or a software approval quietly spans identity, HR and procurement tools, and the orchestration between them is where the difficulty and the value both concentrate. Solving that cross-system coordination for IT first gives Thira a foundation it can extend into finance, HR and procurement once it has earned trust in the most controllable and best-instrumented domain available to it.
The Trust Problem It Has to Solve
The central challenge for Thira is less about capability and more about permission. Autonomous execution against production systems raises genuine exposure, from misconfigured access grants to unauthorized spend, and CIOs have repeatedly watched promising agentic pilots stall at the exact moment of granting write access to real systems. Thira's heavy emphasis on enterprise-grade security, governance and controls is a direct answer to that institutional hesitation, and it is the feature set that will determine adoption far more than the raw autonomy or sophistication of the underlying agents doing the work.
Gupta frames the mission squarely around this shift in expectations, saying AI systems of execution are core to the future enterprise and that the company is building Thira to act on that belief and change outcomes for every enterprise back office through agentic execution. To make good on it, Thira will have to prove that its guardrails, detailed audit trails and fine-grained permissioning satisfy security and compliance teams before those teams will ever hand over meaningful control. The vendors that win the agentic enterprise will be the ones that make governed, reversible execution feel boring, predictable and dependable.
What CIOs Should Watch
Thira is currently working with 10 design partners ahead of a broader launch this fall, so prospective buyers have a real window to evaluate the approach before it becomes widely available and heavily marketed. The useful diligence here is to probe the controls in detail: how permissions are scoped and approved, how every action is logged, how a mistaken or harmful execution is detected and cleanly reversed, and how the underlying knowledge graph stays current as connected systems change. Those answers are what separate a compelling demo from something a security and risk team will actually approve for production use.
The wider signal worth reading is where experienced enterprise founders are choosing to place their chips in this cycle. Rather than chasing another foundation model or general-purpose copilot, the Apptio team is betting that the durable, defensible value lies in the governed execution of the specific workflows that already consume large IT and operations budgets. If Thira can deliver measurable ticket deflection under real enterprise controls, it will validate a thesis many CIOs are quietly testing already: that the genuine payoff from agents comes from safely doing the work across the systems they already run and trust.
)
