§ 01 — DATELINE NOWThe company that wants the bill to make sense.
It is a Tuesday in Redwood City, and inside a glass-walled conference room off Bradford Street, a Gruve engineer is on a call with a Fortune-500 security team. The customer is not asking how many people Gruve is putting on the project. They are asking how many incidents per quarter the AI SOC will close. The bill, when it comes, will be tied to that number. Not to seats. Not to hours.
This is the small, unglamorous, faintly heretical idea at the heart of Gruve: that enterprise AI work - the messiest, most expensive line item in any CIO's budget - should be priced like software. You buy a result. If the result doesn't show up, neither does the invoice. The consultants in the room have spent their careers in the opposite arrangement. They know what they're proposing.
"AI should create outcomes, not overhead. That's why we built Gruve." — Tarun Raisoni, Co-Founder & CEO
Two years ago, Gruve didn't exist on a single business card. Today it has 510 employees in five countries, $37.5 million in fresh capital from Mayfield and Cisco Investments, a private AI infrastructure product called PulseAI, and a small but ferocious belief that the global IT services industry has been quietly mispriced for about forty years.
§ 02 — THE PROBLEMAn industry built on time, not truth.
The enterprise AI bill, as it currently arrives, is a marvel of indirection. A Global 2000 CIO wires off a seven-figure SOW to a brand-name systems integrator. Eighteen months later, slide 47 of the steering-committee deck shows green checkmarks next to phrases like "POC complete" and "framework established." Nothing is in production. Nobody resigns. The retainer renews.
The Gruve founders watched this loop run, up close, for the better part of two decades. Raisoni and Sushil Goyal had built Rahi Systems, a global data-center integrator that scaled to nine-figure revenue before Wesco International bought it in 2022. They had run the projects, paid the consultants, and stared at the same status reports.
"We were a services company that thought like a product company. Then we asked: what if the contract did too?" — Gruve company narrative, paraphrased from public interviews
The complaint is not new. What is new is that the model finally has a forcing function: agentic AI. If a customer-service workflow can be run by an LLM-and-agent pipeline that costs cents per ticket, the old logic of charging by the analyst-hour stops being slightly inefficient and starts being indefensible. Someone was going to attempt the obvious thing. Gruve raised its hand.
§ 03 — THE FOUNDERS' BETSoftware margins for a service business.
The pitch Raisoni took to Mayfield in 2024 was deceptively short. Build an AI services company, but architect it from day one to ship productized offerings - not statements of work. Hire engineers who can write the agents and the infrastructure underneath them. Sign contracts that tie revenue to KPIs the customer cares about: time-to-deploy, ticket deflection rate, mean-time-to-detect, model uptime. Then run it like a SaaS company would: with gross margin, retention metrics, and a roadmap.
It is the kind of plan that looks tidy in a deck and ugly on a Monday morning. Outcome-based pricing assumes you can measure the outcome cleanly, that the customer's data is not actively on fire, and that your delivery team can keep its promises when the Q4 reorg arrives. Gruve's answer is to bring those problems in-house: it pulled in operators from Cisco, Google Cloud, NetApp and SecurView who have spent careers cleaning up exactly that mess.
"The unit economics of consulting have not been reinvented since the mainframe. We think they're about to be." — Tracxn / TechCrunch coverage, April 2025
§ 04 — THE PRODUCTFive things, sold as outcomes.
What Gruve actually sells is unusually concrete. PulseAI is a private AI infrastructure platform - GPUs, orchestration, retrieval, observability - that the company will stand up inside an enterprise's environment in under two weeks. AI SOC is an agentic security operations service, billed against incident volume and detection time. CAISS bundles Cisco's AI security tooling with SecurView's delivery muscle. DFIR handles the forensic cleanup after the worst has happened. A managed AI security service keeps the lights on for production AI workloads.
None of this is unprecedented in isolation. The strange part is the wrapper. Each offering ships with a target metric, a pricing model that follows that metric, and an engineering team accountable for hitting it. The customer's procurement department does a brief double-take, then orders more.
"Production-ready in under two weeks" is the kind of claim that either ends a sales call early - or, occasionally, wins it. — Pull quote, YesPress editor's note
Capital, stacked.
$ millions raised, by round, per public disclosures (2024-2025).
§ 05 — THE PROOFCustomers, partners, and a quietly grown-up cap table.
Cisco Investments rarely shows up on a Series A purely for the optics. That it returned for Gruve's seed and again for the A is, in venture grammar, an underline. The CAISS offering - Cisco AI with SecurView, packaged and resold by Gruve - is the product expression of that relationship. It is also the kind of arrangement that only works if the delivery team can actually deliver. Cisco's brand does not survive a botched implementation, and Cisco knows that better than anyone.
Mayfield led both rounds. The Korea investment in Pricow is the first piece of Gruve's geographic expansion - quiet, targeted, paid for in cash. The 510-person headcount, spread across the US, India and three other countries, is what makes the round-the-clock outcome contracts plausible to a buyer in financial services or healthcare who has been burned before.
"Cisco does not invest twice in companies that are still trying to figure out the org chart." — Industry analyst, paraphrased
§ 06 — THE MISSIONSpeed to scale, minus the overhead.
Read Gruve's own mission statement and you will find an unusually engineering-grade sentence: "unlock innovation's true speed to scale by helping enterprises architect infrastructure-to-agent AI solutions where technical performance and business ROI converge." Unpacked, it is the same complaint the CIOs have been quietly making over their second espressos: AI's promised speed keeps getting eaten by the wrapper around it. Procurement. Governance review. Vendor management. The actual model spends ninety days in a queue.
Gruve's argument is that the wrapper is itself a product problem. Build the platform, ship the agents, take responsibility for the integration, charge for the outcome, and most of the wrapper dissolves. It is the kind of argument that is easy to make in a keynote and brutally hard to keep on a Tuesday. The company's bet is that it has the right operators - Rahi alumni, ex-Cisco engineers, security veterans - to keep it on enough Tuesdays in a row.
§ 07 — WHY IT MATTERS NEXTThe day the rate card died.
If Gruve is right, the implications are not subtle. Roughly a trillion dollars of global IT services revenue is currently priced by the hour. A meaningful slice of that work is about to be done by agents that do not, by definition, bill by the hour. Someone is going to have to translate between the two pricing universes. Whichever firm builds the translator at scale gets to set the new market price.
That is the strategic prize Gruve is reaching for. The 510 employees, the PulseAI ship dates, the Cisco bundle, the Korea investment, the careful insistence on outcome contracts - these are the pieces of a single bet. The bet is that the next decade of enterprise AI will be sold the way software is sold, not the way consulting is sold, and that the firms that learn that grammar first will get to write the dictionary.
The interesting question is no longer whether enterprises will buy AI. It is who, exactly, gets to write the invoice. — YesPress, closing argument
Back in the glass-walled conference room in Redwood City, the call ends. The Fortune-500 security team has signed. The contract names a number of incidents per quarter, a target detection time, and a monthly fee that will adjust if either number drifts. No analyst-hours appear anywhere in the document. The Gruve engineer closes her laptop. Outside, the Bradford Street traffic continues, indifferent, as it should be - the small interesting things in business usually happen in rooms that look exactly like every other room.