Kantata opens Agent Studio for plain-language custom PSA agents

Kantata introduced Agent Studio inside its Expertise Engine, letting professional services firms describe a job in plain language and publish a custom AI agent that runs their delivery standards every time — available and in use by customers now.

SOFTWARE desk — PSA vendors are flooding shops with one-off agents that become new silos; Agent Studio bets that a context-rich superagent plus plain-language authoring is how firms scale judgment without hiring an AI team for every playbook.

What the page says Agent Studio is. It is a conversational capability inside the Kantata Expertise Engine. A firm describes its delivery standards in plain language and gets a custom agent that runs those standards every time. The page calls that agent governed. Governed, here, means the agent follows the standard the firm wrote down, instead of improvising a new answer each week. The lede says this gives every project, every team, and every client the insights and outcomes the firm’s most experienced people would deliver. An agent, here, is software that does a recurring job, not only a chat reply. These lines are the company’s. This desk did not publish an agent.

The quote Kantata prints from Forrester, as a quote, not as this desk’s finding. Ted Schadler, named vice president and principal analyst at Forrester, is quoted from a piece the page titles “Turn Your Proprietary Knowledge Into AI Advantage.” The sentence: the only path to a lasting business advantage is capturing the organization’s own knowledge, expertise, and capabilities and putting them to work through AI models, apps, and agents. The page does not link that piece, and it does not print its date. File the sentence as the quote Kantata chose. This desk did not read the Forrester piece as a second primary.

Basware, as the customer paragraph the page prints. Kantata calls Basware a leading provider of invoice lifecycle management, software that follows a bill from arrival to payment. Leading, again, is Kantata’s word. The page says Basware is using Kantata to put its own knowledge to work. Mark Johnston, chief customer officer, said he wants the professional services team spending time with customers, not doing documentation, figuring out resourcing, and entering billable time. Resourcing is deciding who works on what. Billable time is the hours a firm charges a client. He said that with Kantata they can create AI agents to do that work in the context of their business and processes, and keep people on the expertise that moves the needle for customers. The page does not say Basware published an Agent Studio agent on 22 Sep, and it does not print hours saved. Do not invent either. A quote is not a measured result.

The problem the page says it is answering. PSA software, the page says, is moving toward specialized agents: one for resourcing, one for risk, one for status, one for estimating. Vendors are selling them by the dozen, with a promise of more efficiency. The problem, still the page: every new agent is another silo and another project to build and keep up. A silo, here, means a tool that does not share the rest of the firm’s context. Kantata’s different approach is a superagent that has context across the whole platform: the data, the relationships between that data, and how work gets done. With Agent Studio, a firm can put that context to work without waiting for a vendor to build a new agent for every job. These lines are the company’s. This desk did not count a vendor’s agent catalog.

The chief executive, as the company’s, not as a ranking. Michael Speranza said that in professional services, AI is not the differentiator. Expertise is. AI is what lets a firm unlock that expertise, scale it, and defend it. He said any vendor can hand a firm a tool to build agents. If all the firm does is automate processes, it is racing to the bottom and cutting cost, rather than building something a client can tell apart. He said Agent Studio is meant to help firms compound their expertise in a simple, repeatable way and change how they operate. Compound, in that sentence, means the standard gets reused and adds up, instead of being retyped each week. File the name, the title, and those sentences as the post’s. A quote is not a customer count.

How a build is supposed to run. Agent Studio extends Kantata’s Expertise Agent, which the page calls the professional services industry’s first AI superagent, launched in June. First is the company’s word. The page does not print the year or the day of that June launch. Do not add them. A user describes the job in plain language. The Expertise Agent builds the agent through a dialogue and refines it the same way. Every build and every edit is checked automatically so the inputs and the outputs are structured correctly before the agent is published. Structured, here, means the data going in and the result coming out have a set shape the platform can validate. The page says no forms, no logic trees, and no code are required. A logic tree is a diagram of if-this-then-that rules. Once published, a person can start the agent, or a workflow can start it, with a single command. A workflow is a sequence the platform already runs. These lines are the company’s. This desk did not watch a build.

Who the page says can write the standard. Practice leads, delivery directors, and project management office owners. A practice lead runs one line of the firm’s work. A delivery director owns how that work reaches the client. A project management office, which the page spells out in full, is the group that sets how projects are run. The page says those people can build agents around their own standards without a developer or an AI specialist. Three uses, in the page’s order. Write the standard down once: the format, the escalation rules, the level of detail, and the required outputs for a deliverable that repeats, captured in plain language and applied automatically, instead of rebuilt by whoever is staffed that week. An escalation rule is who gets the problem when it is stuck. Eliminate the prompt tax: work a team re-explains to AI becomes one command, which removes the cycle of drafting, correcting, and reformatting. Give every client the benefit of the firm’s best people: the judgments of the most experienced practitioners are built into the agent, so that expertise is available on every project, every team, and every client, every time. The third bullet on the page has no period at the end. File the three as the company’s.

The chief technology officer, as the company’s, not as a test this desk ran. Vikas Nehru said an agent that improvises is a liability in a business where the same deliverable has to hold up in front of a client every time. He said that by letting firms capture their standards once and run them through a single governed agent, consistency becomes a property of the platform, rather than a matter of who happened to be staffed. He said that is a different promise than adding another agent to the pile. File the name, the title, and those sentences as the post’s.

When a customer can use it, and what the page leaves blank. Kantata Agent Studio is available and in use by customers today. Today, on this page, is 22 Sep 2026. The “click here” link goes to the Expertise Agent and Workflows page at kantata.com/ai/expertise-agent-and-workflows. That is a product page the post points to. It is not a second announcement, and it is not a list of launch customers. The post does not name who is using Agent Studio today. The Basware paragraph says Basware is using Kantata. It does not say Basware published an Agent Studio agent on this date. Do not merge those. The page does not print a price. Do not add one.

The about box, and where the boast stops. Kantata says more than 1,500 organizations worldwide rely on it to assemble a team, amplify what the firm already knows, and forecast outcomes. More than 1,500 is more than fifteen hundred firms. It is the company’s count. This desk did not audit it. The box says Kantata is recognized as a Leader on G2’s PSA Software Grid and on G2’s Resource Management Software Grid, and that it is ranked among the top project management products in G2’s annual Best Software Awards. Leader, and top, are the categories as Kantata states them. This desk did not open the grids. The box says the product covers the services lifecycle from scoping and staffing to delivery and forecasting. Scoping is deciding what the work includes. The site it prints is www.kantata.com. The media address is in Sources.

Plain English for the rest of the card: PSA = professional services automation, the software that staffs people, tracks the project, and watches the money. Expertise Engine = Kantata’s AI layer, the page’s name for it. Agent Studio = the conversational tool, inside that layer, for publishing a custom agent. Expertise Agent = the superagent Kantata says it launched in June, which builds the custom agent through dialogue. governed = the agent follows the standard the firm wrote down. structured inputs and outputs = a set shape the platform can check before publish. logic tree = a diagram of if-this-then-that rules. workflow = a sequence the platform already runs. prompt tax = the time spent re-explaining the same job to AI. PMO = the project management office, the group that sets how projects are run. escalation = who gets a problem when it is stuck. 1,500 = the company’s count of organizations. This filing is the 22 Sep introduction. It is not a price, and it is not a named launch-customer list.

PRIMARY here: Kantata’s 22 Sep 2026 post, “Kantata Unveils Agent Studio: AI Agents That Go Beyond Automation to Compound Professional Services Expertise,” datelined Irvine — Tier A PRIMARY, the company’s own record. The page did not print an hour. The Expertise Engine placement, the plain-language line, the governed-agent line, the Schadler quote and the untitled-beyond-the-name Forrester piece, the Basware and Johnston lines, the silo paragraph, the superagent-with-context paragraph, the Speranza title and quotes, the June launch with no year printed, the dialogue build, the automatic validation, the no-forms line, the person-or-workflow trigger, the three roles, the three bullets, the Nehru title and quotes, the available-and-in-use line, the learn-more link, the more-than-1,500 line, and the G2 sentences are the post’s. NOT claimed: that this desk published an agent, ranked a “leading” or a “first,” read the Forrester piece, timed a Basware week, named a launch customer, treated Basware’s Kantata use as an Agent Studio launch, opened a G2 grid, found a price, a June day, a stock tip, or investment advice. Distinct from the already-filed findem-studio, digitalocean-managed-agents, and perform-ai-commerce.