Gradial raises $65M Series C at $675M to run marketing on AI agents

Gradial, a Seattle-based AI startup that builds agents to run enterprise marketing operations, closed a $65 million Series C on Thursday at a $675 million valuation, according to an Axios exclusive first reported Wednesday. The round was led by Insight Partners with participation from existing investors VMG Partners, Madrona, and PruVen, and brings total funding to roughly $120 million in 16 months.

The platform Gradial ships today

The product sits between a marketing team and the dozens of tools that team already uses, including Adobe, Salesforce, ServiceNow, Databricks, and Sitecore, and gives the team a fleet of AI agents that can do the operational work that the tools were never designed to automate. Authoring, QA, brand compliance, routing updates through approval chains, and publishing changes across systems are all jobs the agents can run without a human queuing up an agency ticket. The pitch from CEO Doug Tallmadge is that brands do not need another agent for every step of the workflow, they need one agent that spans the whole workflow.

The customer list is the part of the announcement that says the most about where the agentic marketing market has actually landed. AWS, Prudential, T-Mobile, Vanguard, Kaiser Permanente, and U.S. Bank are all live. Several of the earliest adopters came from heavily regulated industries, including health care and financial services, where the value of encoding compliance rules into a workflow, so AI agents consistently apply requirements that humans often overlook, was the deciding factor. T-Mobile's senior director of digital business management, Nick Pappas, said the company cut marketing campaign execution time by 80 to 90 percent and now sees 99 percent accuracy on the workflows Gradial handles. ARR at Gradial has grown by more than 10x over the past 12 months, off a substantial enterprise base the company did not break out.

The four co-founders met at Dartmouth and launched Gradial in 2023, shortly after the public debut of ChatGPT. Tallmadge and CTO Deip Kumar both came from SpaceX's Starlink engineering team. Chief growth officer Anish Chadalavada worked on AI strategy at Microsoft and on deep tech investments at Point72 Ventures. COO Anup Chamrajnagar also worked at Point72. The 100-person company is hiring across engineering, sales, and marketing with the new capital.

Where marketing operations sits in the agent buyer wave

The deal lands at a moment when AI search has started to change what marketing operations has to do, and how fast it has to do it. People are now getting answers directly from large language models like ChatGPT and Claude instead of scrolling through Google results. When someone asks an LLM "what is the best health insurance for small businesses," they get a direct answer, and if the brand is not in that answer, it does not matter how the brand's website ranks. The challenge is that the models shaping those answers update constantly, and staying visible in them requires publishing and updating content at a pace that enterprise marketing operations were not built for. A campaign that once took a week to imagine now takes a quarter to ship, because every page, email, ad, post, variant, and update has to crawl through agencies, tickets, handoffs, reviews, and tools that were designed for monthly cycles, not minute-level cycles.

The strategic bet inside Gradial is that brands are about to spend the next several years rebuilding their marketing operations to be agent-native, and that the rebuild will look more like an operating system swap than a feature add. Teddie Wardi, managing director at Insight Partners, framed the round as a bet on a new operating model for marketing that lets enterprises move at AI speed. The alternative is what most large brand marketing teams are doing today, which is bolting AI features onto a stack that was built for the 2010s and hoping the seams do not show. The seams are starting to show. Every software vendor in the marketing stack is racing to ship its own agent features, and the marketing team is left with a separate agent for every step of the workflow, which is exactly the problem Gradial is built to consolidate.

The "AI glue" framing is the part of the announcement that should get enterprise software buyers thinking. Adobe, Salesforce, ServiceNow, Databricks, and Sitecore are all large platforms with their own AI strategies, and every one of them will tell a buyer that their native agent stack is the one to standardize on. Gradial's argument is that no single vendor in the marketing stack has the incentive to be the glue that holds the others together, so the buyer is going to need a neutral layer to keep the agents from stepping on each other. The category is closer to what Okta did for identity or what MuleSoft did for application integration than what any single marketing tool is going to do on its own.

The buyer's checklist for agentic marketing in 2026

For a marketing operations leader evaluating an agentic marketing platform in 2026, the questions that matter are about governance, not features. Here is the short list.

First, ask whether the agents can be governed by a control plane the buyer already trusts. If the agents created by the platform can be discovered, governed, and revoked by an external system, the buyer is in a defensible position. If they are locked inside the vendor's own identity layer, the buyer is making a 10-year bet on a single vendor's agent stack surviving regulatory pressure, market consolidation, and the next round of model upgrades. Gradial is positioning itself as the neutral control plane, but the question applies to every agentic platform on the buyer's shortlist.

Second, ask how the platform handles AI search visibility, not just AI search content generation. The reason Gradial's customers are putting real money into the platform is that the work is not just "write me a blog post." It is "find every place my brand is missing from the AI answers that matter to my funnel, draft the fix, route it through the legal review, and publish it." That is a different product surface than a content generation tool, and the buyer should know which surface they are getting.

Third, ask about the compliance encoding. The most under-discussed part of the Gradial story is the early regulated-industry adoption. Health care and financial services buyers are not the easiest sales cycles. They got there first because they were the buyers who had the most to lose from an agent publishing something that violated a compliance rule, and the most to gain from a system that encoded the rules so a human could not forget them. If the platform cannot show how a compliance rule is written, who can change it, and how the system proves it was followed, the buyer is buying a content tool with marketing on top, not an enterprise agent platform.

Fourth, ask about the integrations you actually run, not the integrations on the deck. The platform has to plug into the systems the marketing team is already paying for, and the integrations have to be deep enough that an agent can do the work end to end, not just push a webhook. The easiest way to test this is to pick a single real workflow, like a regulated product launch, and ask the vendor to run it on a real environment with the real systems the team uses. A demo on a sandbox will not show the seams.

The Gradial deal is the cleanest signal yet that agentic marketing has stopped being a frontier bet and started being an infrastructure category. It also lines up with what we have been seeing in our recent coverage of the MCP supply chain, where the same shift is playing out for the tools marketing teams plug their agents into. The list of customers that have already standardized on it, the size of the round, the fact that a $675 million valuation is now what a Series C looks like for an AI agent startup, and the fact that the round was led by one of the more disciplined growth investors in software all point in the same direction. Marketing operations is about to get the same kind of rebuild that customer support, sales, and security operations have already started, and the platform that becomes the operating system for that rebuild is going to be a large company by the end of the decade.