Top 10 Agentic AI Keynote Speakers for 2026
57% of teams building AI agents already run them in production, and these are the 10 speakers qualified to tell your audience who built the agents, who governs them, and who has actually measured what they get wrong.
Three filters, applied in order
Have they built, deployed, or measured agentic systems?
Commentary about agents, however sharp, did not qualify anyone on its own.
Can they point to published or documented work?
A book, a benchmark, a widely used repository, or a research report a meeting planner can read before signing.
Are they confirmed active and speaking in public right now?
Agentic AI moves fast enough that a 2024 title is a liability, so we checked 2026 roles and real speaking records.
Why this {list is scoped to agentic AI}
Our other AI speaker lists are scoped by who is in the audience: higher education, enterprise, public sector, the future of work. This one is scoped by subject instead. Agentic AI has become its own discipline, with its own frameworks, failure modes, evaluation methods, and governance questions, and a speaker who is excellent on machine learning in general can still leave an audience with nothing usable about agents that plan, act, and touch live systems. No name here appears on any of our other lists. Alex Goryachev opens the list on the workforce-design side of that question: he shaped a $1.1 billion innovation portfolio at Cisco and now advises the California State University system on AI governance.
Top 10 Speakers {Ranked)
Alex Goryachev
Plenty of speakers can explain how an AI agent plans a task and executes it. Far fewer can tell a leadership team what happens to the 4,000 people whose work that agent now partially absorbs. Fewer still can say who is accountable when it acts on a customer record at 2 a.m. Goryachev works the second question, which is the one that decides whether an agent program survives its first quarter. His subject is governance, workforce design, and the shrinking half-life of skills once software starts doing multi-step work on its own.
The operating record behind that is 20 years at Cisco. He shaped a $1.1B innovation portfolio and built 12 university-anchored co-innovation centers across 14 countries. Those programs produced more than $400M in revenue and over 1,100 ventures. The Innovate Everywhere Challenge he built pulled a 72,000-person workforce into the process. He is a US TAG member for ISO 56002, the international innovation management standard, which means he has argued about governance language line by line rather than reading it off a slide.
His current work sits where agentic AI actually lands. As AI Governance Advisor to the California State University system he advises on AI use across 22 campuses and roughly 460,000 students. That includes a ChatGPT Edu deployment that is among the largest in American higher education. At Tulane's A.B. Freeman School of Business he holds the Innovator-in-Residence seat. Both give him a live view of curriculum half-life: the interval between teaching a skill and watching an agent absorb it.
The published work is *Fearless Innovation* (Wiley, 2020), a Wall Street Journal bestseller, plus 212+ Forbes contributions and a book proposal in circulation, *The Great Relearning: Why Your Skills Expire Faster Than Ever*. On stage he runs 4 agentic sessions built for different audiences: "The Agentic Enterprise: Winning the Next Wave of AI Transformation," "The Future of Work: Leading Human + Agent Teams," "Revenue, Reinvented: Selling in the Age of Agentic AI," and an Agentic AI Governance Workshop for teams that already have agents running and no rules written for them. He has delivered 310+ keynotes across 6 continents, with 98% of clients saying they would recommend him.
Book him when your agent pilots work fine in demos and your executive team still cannot say who approves an agent, who audits it, or which skills the organization now needs to buy, borrow, or rebuild.
Bret Taylor
Taylor runs the largest commercial bet on customer-facing AI agents currently in the market, and he talks about it in terms an executive can use: pricing, accountability, and what a company owes a customer when the agent gets it wrong. That combination is rare. Most agent voices are either builders who cannot speak to a P&L or strategists who have never shipped one.
Sierra builds AI agents for customer experience, and the numbers are public. In May 2026 the company raised $950M led by Tiger Global and GV at a valuation above $15B, giving it more than $1B in available capital. Sierra reported $150M in annual recurring revenue in early February 2026, up from $100M in late November, with more than 40% of the Fortune 50 as customers. Its agents handle mortgage refinancing, insurance claims, returns processing, and nonprofit fundraising at a volume Sierra counts in billions of interactions.
The company also popularized outcome-based pricing for agents, charging for resolved issues rather than seats or tokens, which is the sharpest business idea in the agent economy right now and the one most likely to reset how software gets bought. Taylor argues the case in public rather than in a sales deck, including a widely circulated TechCrunch appearance in March 2025 laying out the bull case for agents. In April 2026 Sierra shipped Ghostwriter, an agent that builds other agents from a plain-language description of the outcome you want, which is a useful, concrete answer to the question of who staffs all this.
His earlier record explains the credibility with enterprise audiences: co-CEO of Salesforce, CTO of Facebook, and co-creator of Google Maps. He chairs the OpenAI board, so he sees the model layer and the application layer at the same time, and he is candid about which problems belong to which. Event history includes TechCrunch Disrupt, Skift Live, and a steady run of conference conversations and podcasts on agent economics.
Booking reality: he is a sitting CEO, so his stage time concentrates in marquee industry events and fireside formats rather than a general keynote circuit. Ask early and expect a moderated conversation.
Book him when your executive team has to decide how to price, staff, and stand behind agent-delivered work before a competitor sets the customer's expectation for them.
Harrison Chase
If your audience builds agents, they are almost certainly using something Chase's company made. LangChain started in October 2022 as an open-source project for chaining language-model calls, and it became the default toolkit for agent development: LangGraph for orchestrating multi-step agent workflows, LangSmith for tracing and evaluating what the agent actually did. In October 2025 the company raised $125M at a $1.25B valuation. The work is open source, which means his claims about how agents behave in production are checkable by anyone in the audience.
Chase is also the person publishing the field's numbers. LangChain's State of Agent Engineering report, released June 12, 2026, surveyed 1,340 practitioners and found 57% running agents in production, up from 51% the year before, with quality (accuracy, consistency, following instructions) named as the top barrier by 32%, and nearly 89% of teams now running observability on their agents. Those figures are the closest thing the industry has to a shared scoreboard, and they are more useful to a planning audience than any vendor forecast.
His public argument has an edge worth putting on a stage. Chase's position is that when agents fail, the cause is usually the context they were given rather than the model itself, an idea he helped name as context engineering. That claim reorganizes what a team should spend money on: data plumbing, retrieval, evaluation, and permissions rather than another model upgrade. He has been equally direct about ambient agents, meaning systems that run continuously in the background and interrupt a human only when judgment is required, which is a far more realistic picture of the near term than full autonomy.
On speaking, he hosts Interrupt, LangChain's agent conference, running in New York on September 24, 2026 and London on October 13, 2026, and he is a regular on the technical circuit, including the Databricks Data + AI Summit. He is listed with speaker bureaus, so external bookings are possible, though a founder's calendar sets the limits.
Book him when your engineering and product leadership need a clear-eyed account of what breaks when agents move from a demo to thousands of daily runs, and what the teams already past that point spent their money on.
Michael Wooldridge
The word "agent" did not arrive with ChatGPT. It arrived with a research field that Wooldridge has been building since the early 1990s, and he wrote the textbook that trained most of the people now using the term. *An Introduction to MultiAgent Systems*, now in its second edition, is the standard reference on how autonomous software entities negotiate, cooperate, compete, and fail to coordinate. Every hard problem the current agent wave keeps rediscovering, delegation, trust, conflicting objectives, who is accountable for a decision no single component made, has a 30-year literature behind it. He can hand your audience the short version.
He is also one of the few researchers at that level who can hold a non-technical audience for an hour. He delivered the 2023 Royal Institution Christmas Lectures, broadcast by the BBC in the 198th year of that series. It is the most demanding public-explanation assignment in British science. His trade books, *The Road to Conscious Machines* (published in the US as *A Brief History of Artificial Intelligence*) and a Ladybird Expert Guide to AI, are written for readers with no computing background at all. He holds a UKRI Turing AI World Leading Researcher Fellowship, awarded in 2021.
His current speaking is squarely on this list's subject. In May 2026 Barclays Private Bank published a session with him titled "Agentic AI: In conversation with Professor Michael Wooldridge," following an earlier conversation on the future of AI. He takes the deflationary position where it is warranted, which is exactly what a leadership audience needs after 18 months of vendor promises. Agents are useful, the autonomy claims are ahead of the evidence, and the coordination problems are old, well-studied, and unsolved.
That combination makes him the right corrective voice on a program that already has a builder or a vendor on it. He supplies the intellectual history, the failure modes, and the vocabulary distinctions (an agent, a workflow, a tool-calling loop) that let an audience judge a supplier's pitch on Monday rather than absorb it.
Book him when your audience has heard the promises and needs a scientist with no product to sell. He explains what these systems can do, what the research says they cannot do yet, and why the difference is measurable rather than a matter of opinion.
Sangeet Paul Choudary
Choudary is on this list because he answers the question executives actually lose sleep over: when agents can do the knowledge work, who captures the value that gets freed up? His 2025 book *Reshuffle: Who Wins When AI Restacks the Knowledge Economy* argues that agentic systems do not simply make existing work cheaper. They restack it, moving control from the firms that own the work today toward whoever owns the coordination layer above it. Insurance, banking, logistics, staffing, and professional services are all essentially a coordinated sequence of human judgments. For any company built that way, that is the strategic question of the decade.
He earned the standing to make that argument the long way. He is a co-author of *Platform Revolution* with Marshall Van Alstyne and Geoffrey Parker, and the author of *Platform Scale*. Those books gave a generation of strategists the language for network effects and multi-sided markets. Platform economics turns out to be the right analytical toolkit for agentic AI, because agents are a coordination technology before they are an automation technology. He is a Thinkers50 listed thinker and a senior fellow at the University of California, Berkeley. His ideas also get tested through the BCG Henderson Institute rather than only on a stage.
His speaking record is unusually institutional for a strategy author: keynotes at the G20 Summit, the United Nations, the World Economic Forum, and the World50 Summit, plus the standard corporate and industry circuit. That range matters for a program with a mixed audience of operators and policy people, because he holds the altitude without drifting into futurism.
What he gives an audience is a way to look past the productivity math. Agent deployments get justified on cost per task. Then the real change shows up 2 layers away. A supplier becomes a commodity, a distribution channel disappears, and an in-house team that used to hold the customer relationship now holds a queue. Choudary maps those shifts before they show up in the numbers, which is the only point at which a strategy can still be changed cheaply.
Book him when your leadership team is planning agent investments industry by industry. He helps you understand which parts of your value chain get reshuffled, which get commoditized, and which are worth defending before the reshuffle reaches you.
Dharmesh Shah
Shah has done something almost no other keynote-level figure in agentic AI has done: he built a consumer-facing agent marketplace himself, in public, while running the technology function of a company serving hundreds of thousands of businesses. Agent.ai, the network he launched in 2024, passed 1,000 prebuilt public agents and more than 1 million users, and he ships to it personally. When he tells an audience which agent use cases hold up and which collapse on contact with real users, he is reporting from his own logs.
His stage record on this specific subject goes back further than most. His INBOUND 2024 keynote, "The Future of A.I. Agents," was among the earliest big-tent business keynotes to treat agents as the main event rather than a closing slide, and included a live look at agent-building tooling. He followed it with "You To The Power of AI" at INBOUND 2025. HubSpot's event now runs as UNBOUND, September 16 to 18, 2026, where agent sessions carry the program. He is also an early investor in the agent infrastructure market, including CrewAI, which gives him a view across dozens of deployments rather than one.
The substance is unusually practical for a founder keynote. Shah's argument is that most companies overinvest in the model and underinvest in context and permissions: what the agent is allowed to see, what it is allowed to do, and how a human takes back control mid-task. He is also blunt that most agents shipped in the past 2 years are workflows with a language model attached, which is worth saying to a marketing audience that has been sold otherwise.
The audience fit is commercial rather than technical. He speaks to marketing, sales, service, and general-management audiences in plain language, with jokes, live demos, and a bias toward what a 40-person company can do next quarter. For associations and mid-market conferences, that combination lands better than a research talk.
Book him when your go-to-market organization needs to see working agents, understand what they do to pipeline, service load, and headcount planning, and leave with a realistic view of what a small team can build without a data science department.
Silvio Savarese
Savarese leads the research organization behind Agentforce. That puts him in charge of one of the largest programs anywhere aimed at a single question: how do you know an agent is safe to point at a paying customer? His team's answer has been to build the measurement instruments in the open. CRMArena and its successor CRMArena-Pro, both published on arXiv, put language-model agents into realistic sales, service, and CRM environments and score what they actually complete. The results have been sobering in a useful way. They gave the industry a shared way to argue about agent competence with evidence instead of demos.
He also named the failure pattern that most executives have felt and could not describe. Salesforce AI Research calls it jagged intelligence: a system that handles a difficult reasoning task and then fails a trivial one a human would never miss. That single idea reframes an entire deployment conversation, because it explains why an impressive pilot tells you almost nothing about production behavior. His team's work on agentic simulation testing follows directly from it, running agents through synthetic environments before customers meet them.
His academic credentials are real and current: tenured at Stanford through winter 2021 and still adjunct faculty there, with a long computer-vision research record before this chapter. That combination, a research leader operating inside a company with a large installed base of enterprise customers, is what makes him credible to a technical audience and legible to a business one. He speaks and publishes constantly through Salesforce's research channels and in press interviews. He also takes conference stages on enterprise AI research.
For a program, he fills the slot no vendor keynote usually covers: how to test the thing. Most organizations deploying agents in 2026 have a governance policy and no evaluation practice, which means the policy is decorative. Savarese can walk an audience through benchmark design, simulation, and guardrails in language a non-researcher follows.
Book him when your organization is past the pilot stage. You have to prove, to a regulator, an auditor, or your own executive team, that an agent performs reliably on the specific work you are about to hand it.
Graham Neubig
Software engineering is where agentic AI stopped being a demo and started changing headcount plans. Neubig sits at the center of that shift with one foot in academic research and one in a shipping product. OpenHands, the open-source coding agent platform his company builds, has passed 81,000 GitHub stars and is used by engineers at TikTok, Amazon, Netflix, Google, NVIDIA, Apple, and MongoDB. It plans, writes, and applies changes across a real codebase rather than suggesting snippets in an editor. That is the practical dividing line between an assistant and an agent.
The academic side keeps him honest about results. His group's work has repeatedly topped SWE-bench, the benchmark that measures whether an agent can resolve real GitHub issues. He talks openly about the distance between a benchmark score and a merged pull request. That candor is the reason to book him. CMU's own expert directory points to a Business Insider piece headlined "AI agents aren't ready to do your job," with Neubig as a source, published while his company was shipping agents that do a growing share of it. An audience gets both halves of the argument from one person, with the receipts.
His conference record is substantial and technical. It includes "Deploying Autonomous Coding Agents" for the MLOps Community's Agents in Production series, "The State-of-the-Art in Software Development Agents" at MLOps World, university lectures, and a long run of workshop and program-committee roles in the research community, including agent-focused NeurIPS workshops. He has been publishing and teaching in natural language processing at CMU for over a decade. The explanations are built for people who will be quizzed later.
He gives a leadership audience a calibrated answer to the question every CTO is being asked right now: how much of our engineering work can agents take, on what timeline, and what does the team look like afterward. He answers it with measured results rather than a projection, including where agents plateau, where review costs rise, and which tasks stay human.
Book him when your engineering organization is deciding how far to trust coding agents and needs someone who has measured them, shipped them, and will name the limits without hedging.
João Moura
Moura built the framework that taught a lot of teams to think in crews rather than chatbots. CrewAI was first released on December 4, 2023. It lets developers define a set of agents with distinct roles, tools, and goals, then run them as a coordinated team against a business process. CrewAI states that 65% of the Fortune 500 use it. The company raised $18M across seed and Series A rounds in October 2024, from Insight Partners, Boldstart Ventures, Craft Ventures, and Earl Grey Capital. Andrew Ng and HubSpot co-founder Dharmesh Shah were among the individual investors.
What makes him worth a stage slot is the design argument he has been refining in public. Most business processes do not want one large autonomous agent. They want several small, narrow, well-supervised ones with a clear handoff between them, wrapped in a deterministic flow for the parts that must never improvise. CrewAI's split between crews, agents collaborating with latitude, and flows, structured and repeatable execution, is the clearest expression of that idea in any current framework. It maps directly onto how an operations leader already thinks about work.
He also has the deployment data to back the design. CrewAI runs inside a large number of enterprise pilots and production workloads, so Moura can speak to the boring, decisive details. He covers where multi-agent systems burn cost, why an agent loop stalls, what human approval steps belong in the middle of an automated process, and how teams instrument the whole thing so a failure is diagnosable rather than mysterious.
His speaking is steady and technical. He appeared at the Databricks Data + AI Summit 2026 on a panel about AI agent infrastructure alongside founders from LangChain and other agent companies. He has also spoken at The AI Conference and ODSC events, and does a regular circuit of developer conferences and podcasts. Before CrewAI he was an engineering leader, so the talks are built for people who will implement, not just applaud.
Book him when your technology and operations teams have automated the easy tasks and are looking at a multi-step process that crosses 3 or 4 systems. He supplies a working design pattern for agents that hand work to each other without losing the audit trail.
Qingyun Wu
Wu co-created AutoGen, the multi-agent conversation framework that came out of her research collaboration with Microsoft Research. It became one of the most widely used foundations for agent development. She now leads AG2, the open-source platform that carries that work forward, while holding a faculty position at Penn State. AG2's own speaker materials put AutoGen at more than 700,000 monthly downloads and over 20,000 developers. That makes her one of the few people on this list whose framework shaped how the rest of the field talks about agents at all.
The specific contribution is worth naming for an audience, because it is subtle and it is everywhere. AutoGen framed agent work as a structured conversation between specialized agents, with a human able to enter that conversation at any point. That pattern, agents talking to agents with a person in the loop by design rather than as an afterthought, is now the default assumption in most enterprise deployments. It came out of a research group, not a product roadmap, and she can explain why the choice was made and where it strains.
Her academic record is real: a 2019 SIGIR Best Paper Award, and a Best Paper Award at the ICLR 2024 LLM Agents Workshop, plus continuing research on automated machine learning and agent evaluation. That means she talks about failure cases with the precision of someone who has measured them, including cost, latency, and the ways multi-agent systems produce confidently wrong consensus.
She speaks regularly on the technical circuit: the 2026 Agent Conference, ODSC events including sessions on building effective agents with AG2, the DataHack Summit, and the Agentic AI Summit programs. Her sessions are built for practitioners who will go build something. She is unusually good at the transition from research finding to implementation choice, which is the hardest handoff in this subject.
Book her when your technical audience needs the design principles behind multi-agent systems from someone who wrote one of the originals. She fits best when your organization is choosing between orchestration frameworks and wants that decision made on architecture rather than on marketing.
Why Agentic AI Keynote Speakers Matter in 2026
Who should stand in front of your audience and talk about agents
Sierra's AI agents now work for more than 40% of the Fortune 50, and the company's annual recurring revenue went from $100M in late November to $150M by early February, roughly 10 weeks.
The wider picture is calmer and more instructive. LangChain surveyed 1,340 practitioners for its State of Agent Engineering report in June 2026 and found 57% with agents already in production, up from 51% a year earlier. The barrier they name most often is quality, at 32%: accuracy, consistency, whether the agent follows the instruction it was given. Nearly 89% now watch every step their agents take, because the alternative is finding out later.
So the capability is arriving faster than the organizational competence required to run it responsibly. That distance between the two is the real subject of an agentic AI keynote, which is why we scoped this list by subject instead of by audience.
Watch any service supervisor during the first week an agent starts closing tickets on its own, the queue shrinks, which is the part that makes the slide. Then one case comes back wrong, and the supervisor learns that nobody on the team can reconstruct why the agent decided what it decided, or who authorized it to decide anything in the first place. The software worked; the design around the software had never been written.
This has happened before, at national scale. American factories owned electric motors for decades before productivity moved, and Paul David's research on the dynamo explained the delay: the motors were installed in buildings still laid out around a central steam shaft. The payoff came only when the work itself was rebuilt around the new power source, floor plan by floor plan, by people who had to relearn their trade in the middle of their careers. The relearning was the hard part, and it took a generation.
Agents put the same bill in front of every organization, and the invoice lands on individuals first. When software absorbs the multi-step tasks, the half-life of skills gets shorter across whole job families at once, and what remains for people is the work above the algorithm: judgment, taste, accountability, and the ability to direct a set of agents toward an outcome and know when they have gone wrong. That is a workforce design question long before it is a technology question. It is decided in budget meetings this quarter, and felt at kitchen tables for the next 10 years.
An agent can be deployed in a week; the judgment it borrows took a decade to build.
Which is why the person you put on your stage matters more this year than last, since a general AI speaker will only describe the wave. The 10 people below have built the systems, measured them, governed them, or studied the multi-agent problem since before it had a market, and each one can hand your audience something to test on Monday.
One question to take into your next leadership meeting: if an agent made a call in your name yesterday, could anyone in the organization tell you why? If the answer takes more than a minute, that is the topic your audience needs. I'm easy to find.
Sources: LangChain, "State of Agent Engineering" (June 12, 2026); TechCrunch, "Sierra raises $950M" (May 4, 2026); Paul A. David, "The Dynamo and the Computer" (American Economic Review, 1990).
How we picked this list
Ranking speakers on followers produces a list of famous people. Ranking them on what they have actually run produces a list your audience can use. 3 filters decided this one, applied in order.
The operating and product record came first. Every person here has built, deployed, governed, or measured agentic systems that other people depend on. That includes agent platforms with named customers and public revenue, open-source frameworks with verifiable adoption, benchmarks that the field argues about, and institutional advisory work on how agents get authorized and audited. Commentary about agents, however sharp, did not qualify anyone on its own.
Published or documented work came second. Each entry points to something a meeting planner can read before signing a contract: a book, a peer-reviewed paper or benchmark, a widely used repository, a research report with a sample size, a keynote recording. Where a number appears in an entry, it comes from the person's own organization, an employer page, a funding announcement, or a named publication. Claims we could not source, we left out, including a few we suspected were true.
Verifiable current activity came third, and it removed more candidates than the other 2 filters combined. Agentic AI moves fast enough that a 2024 title is a liability. We confirmed each person's role and affiliation as stated in 2026, and confirmed each one speaks in public: conference programs, event listings, recorded sessions, bureau representation. Several well-known agent builders were dropped for the simple reason that they do not take stages.
2 things this list deliberately does not weigh. Fee level is not a proxy for quality, and we do not rank by it. Company size is not either: the founder of a 30-person framework company can be more useful to your audience than an executive from a firm 100 times larger, depending on the question you are trying to answer.
The order reflects breadth of usefulness to a mixed leadership audience rather than technical depth alone. A CTO planning an engineering transition and a chancellor writing an AI policy need different people from this list, and the entries say which is which.
How is this different from your general AI keynote speaker lists?
Our other AI speaker lists are organized by who is sitting in the audience: higher education, enterprise, public sector, and the future of work. This one is organized by subject. Agentic AI has its own frameworks, its own failure modes, its own evaluation methods, and its own governance questions, and a speaker who is excellent on AI in general can still leave an audience with nothing usable about systems that plan and act on their own. Nobody on this list appears on the others. If your program needs one AI keynote for a broad audience, use the audience lists. If your program is about agents specifically, use this one.
Should we book a technical agent builder or a business speaker?
It depends on which decision your audience has to make when they get home. If they are choosing an orchestration framework, sizing an engineering transition, or deciding how much to trust coding agents, book a builder or a researcher and let the session get specific. If they are deciding what to fund, how to price agent-delivered work, who signs off on an agent's actions, and what happens to the workforce underneath it, book the operator or the strategist. Mixed audiences usually do best with 2 sessions rather than 1 speaker asked to cover both, because the compromise version tends to satisfy neither half of the audience.
What should we ask an agentic AI speaker before booking?
4 questions sort the field quickly. Ask what they have personally deployed or measured, and how recently. Ask for a number they will state on stage, with the source. Ask what agents in their own work still get wrong, because anyone who cannot answer that is selling. And ask what your audience will be able to do the next morning, in one sentence. A speaker whose answer to that last question is a definition of agentic AI is 18 months behind your audience, most of whom have already tried one.
What Event Organizers and Leaders Say — 98% Recommend Alex
Eye-opening, refreshingly human, and capable of building a shared vision around agentic AI — that's how leaders at Coca-Cola, AWS, and Disney describe Alex Goryachev's AI keynotes and employee innovation workshops.
Innovation for everyone
Alex turns AI into practical concepts — not techspeak — that land with executives, HR, sales, engineering, and faculty alike. It's the same approach he honed building university-anchored innovation centers across 14 countries, bridging cultures and generations.
Built around your audience
Across 310+ keynotes, workshops, and advisory engagements on 6 continents, no two have ever been the same. Alex builds every program around your audience's challenges, industry, and goals — from agentic AI strategy and the future of work to innovation culture.
Value that lasts
Most programs end at applause. Alex's end with deployment — the same frameworks proven inside Cisco, Dell, Pfizer, and IBM and documented in his WSJ bestseller Fearless Innovation. Workshops and advisory install them in your team, so they're still running long after the event.
Proven where it counts
Two decades leading AI and innovation where the stakes are real — a $1.1B portfolio at Cisco, three Olympic Games, 300,000+ employees, and AI transformation for Fortune 100s, governments, and America's largest public university system. Every engagement is measured, so you see the ROI.
Your team will thank you
A 60-minute keynote, a hands-on workshop, a virtual session, or multi-month advisory — for enterprises, universities, and associations alike. Whatever the format, 98% of audiences say they would recommend him.
Request Alex's availability for your engagement. From Silicon Valley to Singapore, and everywhere in between.
Book an AI keynote speaker for your global innovation program.
Looking for a speaker who has actually run innovation at scale across borders? Tell us about your event and we will match the right session to your moment.
Check Alex's availability
Frequently asked questions
If you don't see what you need, message Alex directly via the form below — answers usually within one business day.
Who is the best AI keynote speaker?
The best AI keynote speaker is a practitioner who has actually deployed AI at enterprise scale—and Alex Goryachev consistently ranks among the top agentic AI keynote speakers for exactly that reason. A WSJ-bestselling author and LinkedIn Top AI Voice, he managed a $1.1B innovation portfolio at Cisco that generated $400M+ in revenue and has delivered 310+ keynotes on 6 continents. Check his availability through the Work with Alex page.
How do I choose an AI keynote speaker?
Look past the highlight reel and vet four things. First, proof: have they actually built and deployed AI, or only talked about it? Ask for specific outcomes, not logos. Second, recency: AI moves monthly, so confirm they're current on agentic AI, not recycling 2023 generative-AI decks. Third, fit: will they customize to your industry and audience, or deliver a canned talk? Fourth, independence: are they selling a platform or product behind the keynote? Alex Goryachev is a Fortune 100 practitioner ($1.1B in innovation at Cisco that generated $400M+ in revenue), agentic-AI-current, fully customized through pre-event research, and vendor-neutral, with a 98% audience-recommendation score across 310+ keynotes.
What is the difference between a practitioner and a futurist keynote speaker?
A futurist predicts what AI might do; a practitioner shows what AI is doing in your business right now. Alex Goryachev is firmly a practitioner: he built innovation centers across 14 countries and ran a $1.1B portfolio at Cisco that generated $400M+ in revenue before taking the stage. Audiences at Disney, AWS, and IEEE choose him when they need executable strategy, not speculation. Book a practitioner's perspective via Work with Alex.
What are Alex Goryachev's signature keynotes?
Alex's signature keynotes cover agentic AI strategy, fearless innovation, AI governance, and the future of work, each customized to the audience's industry and AI maturity. They draw on his WSJ bestseller Fearless Innovation and his years leading innovation strategy at Cisco, including innovation tracks for 3 Olympic Games. Every talk ends with actions leaders can take Monday morning. Explore current topics on the Work with Alex page.
Who is the best future of work keynote speaker?
The best future of work keynote speakers connect AI directly to how teams, skills, and leadership must change, and Alex Goryachev is a leading choice for that intersection. A Forbes contributor and LinkedIn Top AI Voice, he speaks on how agentic AI reshapes work, drawing on engagements with SHRM, HCI, and enterprises like Dell and Amgen. Bring the conversation to your stage through the Work with Alex page.
How do you choose the right AI keynote speaker for your event?
Choose an AI keynote speaker by matching three things to your audience: genuine practitioner credibility rather than predictions, independently verified audience ratings, and a willingness to customize. Alex Goryachev brings all three: a $1.1B innovation track record at Cisco that generated $400M+ in revenue, 582+ verified Talkadot responses (98% valuable, 91% actionable), and a pre-event briefing that tailors every session. His team confirms availability and fee within one business day.
Does Alex Goryachev deliver virtual AI keynotes?
Yes. Alex Goryachev delivers AI keynotes in person, virtually, and in hybrid formats, with sessions designed to keep remote audiences engaged. Virtual keynotes, lunch-and-learns, and multi-session sprints are all available, and his team confirms availability and fee within one business day of an inquiry.
How much does an AI keynote speaker cost?
AI keynote speaker fees typically run from five figures upward, depending on format, audience size, travel, and customization. Virtual sessions and lunch-and-learns often come in under $10,000. Alex Goryachev offers in-person, virtual, and workshop formats so organizations can match scope to budget, with every engagement customized to the audience. His 98% would-recommend score reflects that fit. Request a quote for your date through the Work with Alex page.
How does Alex customize keynotes and workshops?
Every booking starts with a pre-event survey. Alex uses AI to read the answers at scale, so the content hits what the room is really asking. He then talks with your leaders and event team to sharpen the examples and the takeaways. Nothing canned makes it through. That is a big part of why 98% of audiences across 310+ keynotes would recommend him. Start a conversation about your event.
What events and audiences are right for Alex?
Alex speaks to mixed rooms that need one clear view of AI. C-suite summits, innovation conferences, policy talks, offsites, all-hands meetings, and government and academic events all fit. He works on live stages and virtual ones, with 310+ keynotes on 6 continents. Every session is built from pre-event research, so the examples fit the room. 91% of audiences rate his sessions actionable, with steps they can use the next day. Tell Alex about your event and he will suggest the right format.
.svg)
