Keep in mind HAL 9000 from “2001: A House Odyssey”? The eerily calm, near-sentient robotic? It was a reliable coworker for the people onboard. Nicely, a minimum of earlier than it turned rogue (extra on that later).
HAL represents certainly one of science fiction’s most iconic visions of synthetic intelligence (AI) with autonomy — an AI system that might observe, motive, and act independently to realize its targets.
That is a glimpse of agentic AI in motion. Because the identify already offers away, agentic AI has company. Operating on giant language fashions (LLM), it really works with out human prompts. It not solely creates and plans duties however executes them by itself. And all of this isn’t simply sci-fi anymore; it is the following huge factor for tech giants. Salesforce deems it “era-worthy,” and McKinsey touts it because the “subsequent frontier of generative AI.”
“The AI brokers market is predicted to develop at 44.8% CAGR between 2024 and 2030, pushed by technological advances in pure language processing (NLP).”
One notable growth within the agentic AI breakthrough is that it warps how we’ve seen AI to this point: hand-held instruments that support us. As an alternative, AI brokers are proactive digital coworkers which may simply be reshaping how groups are structured, how staff work together, how duties have to be divided, and extra.
This shift additionally raises an unsettling, maybe sinister, query: are so many human staff wanted within the first place?
Backed by information from 3,621 evaluations, this text explores how AI brokers are getting used within the office right now and what which means for the way forward for work. By analyzing their commonest functions, utilization patterns, and different insights, we are able to perceive not simply the place AI is making an affect however the way it’s reshaping staff buildings, decision-making, and the division of labor between people and digital coworkers.
Let’s get into it.
What’s agentic AI and the way does it work?
Let’s start with what agentic AI is.
Agentic AI was first developed within the 2000s when machine studying (ML) fashions allowed brokers to study and enhance utilizing giant databases. Immediately, the agentic AI panorama is predicated on superior autonomy, enmeshed with an moral and accountable AI-controlled setting.
Though we at G2 use the time period AI brokers to seek advice from “software program techniques that may motive, act, and automate autonomously,” that is under no circumstances the one present or accepted definition, explains Matthew Miller, analysis principal at G2.
Largely, AI brokers are autonomous digital staff that use instruments to perform targets. These brokers have the flexibility to recollect throughout duties and altering states, in response to BCG.
However what does this seem like in observe? To grasp their affect, let’s look at an business that’s already seeing AI-driven transformation: content material advertising and marketing.
The human advertising and marketing supervisor assigns the AI agent a imprecise situation: create a weblog put up on the most recent digital advertising and marketing developments. The agent then researches for it, submits a graphic design request, drafts a weblog, and phases it on a content material administration system. The supervisor lastly steps in to evaluate it earlier than the agent schedules it for publication. It additionally creates and schedules social media posts for promotions.
Whereas this instance demonstrates the autonomous capabilities of agentic AI, many individuals may marvel how these techniques differ from the AI assistants they’re already accustomed to. In any case, hasn’t AI been serving to with content material creation for a while now? This raises an necessary query concerning the distinction between actually agentic techniques and their extra restricted predecessors.
Is agentic AI totally different from chatbots?
Tim Sanders, vp of analysis insights at G2, believes there’s a gradient to agentic AI. “I wish to name it the Waze-to-Waymo continuum.”
He provides that the entry level to this spectrum is co-pilots or chatbots. And that the highest of this continuum is occupied by “system of brokers.” “Brokers right here, on the highest degree, span throughout techniques and collaborate and ship in your intent and never simply your acknowledged want,” clarifies Tim
Supply: LinkedIn put up by Tim Sanders
He additionally argues that agentic AI is an enormous deal because it solves the supply downside customers and organizations may face with chatbots and automation, “The supply downside is that particular person productiveness good points don’t essentially translate into organizational elevated velocity of outcomes.”
For instance, human content material writers may save 30% of their time utilizing writing assistants, however they could squander it on taking a nap, scrolling by way of social media, or attending extra conferences.
As customers of AI brokers — the persona remains to be evolving as sellers experiment with functions — have begun integrating them into their workflows, sure patterns have emerged. These point out how customers favor to make use of these brokers and their likes and dislikes, which in flip exhibits the affect AI brokers have had on work. Let’s talk about them based mostly on person evaluations.
Agentic AI in motion: who’s utilizing it, what they assume, and the affect
To grasp the usage of AI brokers, we analyzed 3,621 evaluations made by verified customers worldwide on G2. We discovered that agentic AI is mostly utilized in laptop software program, IT, and providers industries, adopted by monetary providers.
These brokers are additionally extensively accessible and helpful for companies of all sizes, not simply giant enterprises.
“AI brokers aren’t only for the large established firms. In response to G2 evaluations from the previous 12 months for AI brokers, over half (55%) of evaluations are from small companies.”
Matthew Miller
Analysis principal at G2
Moreover, the truth that “AI chatbots”, “AI brokers”, and “chatbots” are the most well-liked AI classes when it comes to site visitors on G2 testifies to the rising curiosity in agentic AI amongst customers.
Sentiment evaluation of the AI brokers class
After analyzing 500 person evaluations between September 2024 and February 2025, we discover that 87% are optimistic, whereas simply 4.6% are unfavorable. This means important satisfaction ranges and the ensuing adoption of AI brokers, that are nonetheless of their early kinds.
Let’s perceive the person sentiment extra deeply and its implications for each patrons and sellers of AI brokers.
What customers like essentially the most
- Ease of use: Customers admire that the software program on this class is intuitive and simple to navigate.
- AI capabilities: AI features are a robust level, with customers mentioning their effectiveness.
- Buyer help: Many evaluations spotlight glorious customer support and help.
- Effectivity: Customers word that the software program helps automate duties and enhance workflows.
- Integration: Seamless integration with present instruments and platforms is a plus.
- Reducing time to ROI: The time to ROI has decreased from 16.9 months in 2023 to 9.5 months in 2024.
G2 Take
The emphasis on ease of use is a driving think about agentic AI’s adoption. The democratization will speed up its use throughout sectors and features and attain non-technical customers.
Customers additionally worth buyer help, impressing the human-AI collaboration in onboarding and launching agentic AI options.
Stressing the lowering time to ROI, Yukta Rustagi, a market analysis analyst at G2, provides, “This additionally implies that organizations leveraging AI are gaining a aggressive edge by way of effectivity and innovation. It reinforces the assumption that AI brokers are actually a extra rapid and impactful funding for companies.”
ROI for AI Brokers class (Jan 2023 – Dec 2024)
Supply: G2 Market Analysis
Challenges and what they imply
Customers have highlighted characteristic limitations in AI brokers. However this doesn’t essentially point out performance gaps. As an alternative, the expertise’s infancy involves the fore, revealing that we’re nonetheless creating shared concepts of what agentic techniques ought to accomplish throughout domains.
Options for particular person customers might be constructed on common options that serve a number of wants. Nevertheless, enterprises want specialised options that clear up issues at scale. Thus, AI brokers for bigger firms should goal particular use instances equivalent to coding, stock administration, and lead qualification.
AI brokers current a customization problem, not like that seen with earlier AI makes use of. They’re rising as adaptive techniques that personalize by way of interplay slightly than configuration. Thus, firms might not have to customise AI brokers, that are thought of digital coworkers, any greater than they customise human colleagues.
Some customers wrestle with coaching AI fashions, which could hobble adoption and delay the time to ROI. AI brokers, as autonomous options, should want minimal coaching. They have to adapt to customers, not vice versa.
The training curve will stay steep if customers proceed to assume within the command-execution paradigm with AI brokers, which function by way of objective delegation and autonomous planning. Each sellers and patrons should reimagine onboarding as a collaborative alignment between people and brokers.
Value considerations amongst some customers point out low market maturity as companies wrestle to quantify agentic AI worth with out standardized metrics. Sellers have but to generate sufficient case research and predictable use instances to construct a stronger enterprise case for AI brokers.
Whereas AI brokers are getting used throughout industries, our evaluation of G2 evaluations reveals that buyer expertise is the realm the place their affect is most pronounced. Practically half of the person evaluations point out CX-related enhancements, making it a pure focus for understanding the real-world advantages of agentic AI.
Buyer expertise: commonest affect space
Whereas analyzing evaluations for AI brokers, we discovered that 217 of them, or 43%, point out buyer expertise, making it arguably the commonest space of affect for firms right now.
Echoing the sentiment, Tim believes brokers are essentially the most employable in two use instances to this point: buyer help and gross sales growth.
“Buyer help groups presently face excessive backlogs, which agentic techniques can rapidly scale back. Gross sales growth leaders see little draw back to agentic SDRs, given their hard-to-fulfill quotas of lead technology, booked conferences, and addition to the pipeline,” he explains.
“Within the subsequent 10 years, AI in CX will scale back the associated fee to serve by an order of magnitude, enabling manufacturers to increase touchpoints with prospects in a manner that has by no means been attainable.”
Jason Maynard
Chief expertise officer of AMER and APAC at Zendesk
Matthew has discovered that agentic AI has had the most important affect on the shopper journey across the consideration part. “Though the readiness of patrons to depend on brokers is rising, buying nonetheless stays a ache level,” he claims.
At the moment, 30% of customers would work with an AI agent for quicker service. “We predict that it will improve because the techniques turn out to be extra dependable and as customers and companies develop extra belief towards the techniques,” he says.
Most optimistic evaluations on G2 spotlight AI’s effectivity in buyer help and automation. Whereas unfavorable mentions (4.6%) embody considerations about AI errors, sluggish help response, and lack of customization.
To help our findings on AI brokers use instances, we requested two customers how their expertise with AI brokers had been in enhancing buyer expertise, a preferred use case:
G2 Icons Converse
G2 Icons are our most prolific contributors and energy customers of the G2 platform.
Person 1: Peter Gill, freight dealer
How do you employ AI brokers to reinforce buyer expertise?
We should perceive buyer necessities concerning pickup factors, supply factors, and commodity sorts to supply them with essentially the most aggressive charges and high-quality providers.
Earlier, we used to do that manually. Now, utilizing AI-driven chatbots and digital assistants, we are able to supply immediate responses to buyer inquiries, lowering resistance within the gross sales funnel. Moreover, AI helps us look at buyer habits patterns, enabling us to know them higher in order that we are able to get repeat enterprise from them.
Results of agentic AI interventions:
- Buyer satisfaction scores up by 23%
- Uptick in ticket decision charge by 19%
- Decreased guide work for groups by automating 33% of repetitive inquiries
Person 2: Luca Piccinotti, development marketer
How do you employ AI brokers to reinforce buyer expertise?
We use AI brokers to research market developments, phase goal audiences, and forecast buyer habits, which in flip helps the event of exact messaging and tailor-made campaigns.
Results of Agentic AI interventions:
- Decreased ticket decision time
- Improved buyer suggestions
At first look, AI’s most rapid affect appears to be on buyer expertise — dealing with help queries, automating duties, and bettering engagement. However what occurs when these similar AI capabilities are turned inward?
Simply as AI is remodeling totally different industries, it’s additionally redefining how work will get accomplished, how groups collaborate, and what it means to be productive. As AI brokers transfer past buyer help and into core enterprise operations, their function within the office is changing into unattainable to disregard.
Reimagining work: consultants weigh in
Agentic AI isn’t nearly mere automation however cognitive reallocation. It’s creating a brand new manner to take a look at the division of labor the place people are elevated to higher-order pondering roles.
To grasp this real-world affect, we flip to business consultants who’ve noticed AI brokers in motion, providing beneficial insights into how companies are integrating them, the place human oversight remains to be essential, and what expertise will likely be wanted on this AI-driven future.
Right here’s what they must say:
AI brokers are the always-on teammates
Some peddlers of AI brokers are branding them as digital staff, others as teammates, and others nonetheless as instruments that stand behind customers versus between them, says Matthew.
To this, Tim provides, “They require much less human-in-the-loop efforts than LLM chatbots. We must always consider these brokers as staff members that by no means take break day, get distracted, or develop dangerous attitudes.”
Mark Purdy, director of Beacon Thought Management, says AI brokers additionally carry out quite a lot of specialised features. For instance, brokers can collect info from a number of inner databases and exterior data sources, assessing and synthesizing the insights for enterprise analysts, attorneys, scientists, or different data staff.
AI brokers can act as casual sounding boards
AI brokers can perceive totally different enterprise issues and contexts, triggering actions and workflows that scale back the pressure on human staff,” factors out Mark. For instance, AI brokers can assess e-mail site visitors from prospects or purchasers, routinely responding to queries or complaints. They will monitor and observe up on gross sales leads.
“AI brokers might help human managers and leaders make higher choices by operating totally different situations or simulations to indicate the outcomes of other programs of motion.”
Mark Purdy
Director of Beacon Thought Management
On this sense, AI brokers can function casual sounding boards for various decision-makers, whether or not on the supervisor or board degree.
Nevertheless, Mark additionally emphasizes that whereas AI brokers can act autonomously, there almost all the time must be some human-in-the-loop ingredient to keep away from errors or unethical choices.
Human-(nonetheless)-in-the-loop
When requested how organizations should divide duties between brokers and people, Mark says the diploma of human involvement will rely upon many elements. These embody the choice’s significance, the diploma of belief within the AI agent’s suggestions, the implications of a mistake, and the human employee’s expertise and judgment.
“For instance, there’ll seemingly should be a excessive diploma of human supervision in areas equivalent to healthcare or protection the place the implications of errors by AI brokers could possibly be very important, and doubtless much less want in areas equivalent to customer support or back-office processing,” says Mark.
Echoing the sentiment, Sreelesh Pillai, co-CEO at Zepic, says that the corporate’s AI brokers function independently, mimicking and amplifying human functionality whereas permitting companies to configure human involvement the place vital.
Overworked? AI may simply be the productiveness increase you want
Leandro Perez, CMO for Australia and New Zealand at Salesforce, factors to a productiveness problem dealing with people. “For example, for Australia, the productiveness development has fallen to thirtieth out of 35 comparable international locations,” he says.
“Overworked staff want instruments that assist them work extra effectively.” And that is the place instruments like Agentforce, Salesforce’s agentic AI resolution, come into the image.
Leandro cites the instance of Fisher & Paykel, certainly one of Salesforce’s prospects, which has saved over 3,300 hours month-to-month by reimagining their processes by way of AI brokers and automation.
“Agentic AI is not nearly effectivity; it’s about unlocking potential with limitless digital labor…you may count on to see each worker main or working alongside groups made up of AI brokers and studying to extract most worth from them.”
Leandro Perez
CMO for Australia and New Zealand at Salesforce
Leandro argues that the agentic AI shift isn’t essentially about creating totally new roles however slightly about evolving present ones. With AI brokers dealing with routine duties, people will turn out to be orchestrators of intelligence.
He emphasizes strategic pondering, empathy, and resilience as expertise that can turn out to be extra necessary with the involvement of AI brokers.
CX execs should evolve with AI
In response to Jason, CX groups should reskill in the same option to advertising and marketing groups within the 2010s to scale back the associated fee to serve. “As search and social emerged as dominant channels for demand technology, the “technical marketer” grew to become indispensable: half system integrator, half information analyst, and half advertising and marketing strategist,” he says.
CX will see the identical transition with the necessity for technical CX professionals who can design the foundational parts of AI brokers — data, insurance policies, procedures, and techniques that help AI brokers. “They’ll then use qualitative and quantitative information to repeatedly enhance and optimize these techniques,” he says, including that such expertise will command a premium within the job market.
As professionals adapt to this AI-driven shift, their roles will evolve slightly than disappear. Nevertheless, with AI taking over extra duties, a urgent query emerges: what does this imply for the human workforce?
Is the menace to people actual?
Nicely, probably not.
Agentic AI can, in idea, perform autonomously and take over total processes and techniques. However will this integration be on the expense of human staff?
Employee considerations about AI are each justified and misplaced, believes Kate O’Neill, founder and chief tech humanist at KO Insights. “The menace is not that AI will change people wholesale — it is that we would fail to reimagine work in ways in which leverage uniquely human capabilities alongside AI.”
She claims that sensible firms are already reframing AI from a substitute expertise to an enhancement expertise. “This is not simply semantic gymnastics; it is a elementary shift in how we design and deploy these techniques,” she provides.
“The way forward for work is not a zero-sum recreation between people and machines. It is about creating synergies that make each extra succesful, extra productive, and in the end, extra human.”
Kate O’Neill
Founder and chief tech humanist at KO Insights
Kate calls upon AI agent distributors to design their instruments explicitly as human amplifiers, not human replacements. Which means constructing instruments that improve human judgment, creativity, and emotional intelligence — the very qualities that make us uniquely human.
Stressing synergy, Sreelesh says essentially the most thrilling shift is not simply in creating new AI-specific jobs; it is in how AI transforms present roles into their “augmented” variations or the next-strategic evolutions of them.
He predicts that as agentic AI turns into extra frequent, we’ll see customer support representatives turn out to be insight-driven engagement specialists, entrepreneurs turn out to be buyer journey architects, and operations people turn out to be automation strategists.
On the same word, Jason means that agentic AI will spawn extremely expert jobs unseen in CX to this point. These will give attention to designing and creating the foundations that help AI brokers.
He claims that traditionally, human brokers have been requested to be the “glue” throughout techniques and data sources — gathering related data and navigating an online of backend techniques to resolve issues with orders, merchandise, and providers.
Greater than your common publication.
Each Thursday, we spill scorching takes, insider data, and information recaps straight to your inbox. Subscribe right here
Nevertheless, right now, AI brokers related to data sources, techniques, and instruments are enabling groups to automate these repetitive steps and give attention to monitoring the standard and accuracy of responses, offering judgment on the following steps, and approving actions the place judgment is required.
As agentic AI evolves, each sellers and patrons should implement them in a human-centric manner. Intrinsic motivation, a key driver for workers, can take successful if agentic AI is allowed to take over duties which have supplied staff with a way of mastery and function. It’s no extra about ability adaptation however reconstructing an expert identification for staff as work will get divided between them and AI brokers.
Agentic AI can also be sure to flatten conventional hierarchies inside organizations. Junior professionals armed with AI brokers can carry out at senior ranges, compressing years-long studying curves into months. This creates unprecedented workforce agility. As well as, senior professionals who’ve historically prided themselves on their info mastery should redefine their model.
The ethics and security hurdles
AI brokers work throughout techniques. They don’t take breaks, go on trip, or want motivation to carry out duties. Not directly, they’re tireless and timeless. How secure is buyer information with this novel utility of AI? And what ought to a governance framework for agentic AI seem like?
Within the film “2001: A House Odyssey,” the story of the autonomous robotic HAL 9000 takes an ominous flip. It goes rogue, doubting people, blaming them for errors, and disconnecting life help.
The destiny of people with agentic AI is unlikely to be comparable, a minimum of within the close to future. Earlier than the expertise turns into actually autonomous, people should set up tips, test for potential hallucinations, and shield information.
Agentic AI use: an influence play?
In response to Kate, essentially the most urgent moral considerations round agentic AI transcend surface-level automation points to elementary questions of belief and decision-making authority.
“The core moral problem round Agentic AI is not about algorithms or automation — it is about energy.”
Kate O’Neill
Founder and chief tech humanist of KO Insights
Who controls these choices? How can we guarantee prospects retain significant company?
“Each time an AI agent makes a selection, it is basically making a small prediction about human habits and desire. Get sufficient of those micro-decisions fallacious, and we’re not simply failing at customer support — we’re undermining human autonomy. The stakes are larger than most firms understand,” warns Kate.
The answer? We’d like unprecedented ranges of transparency with agentic AI. “Clients want to know not simply that they are interacting with AI, however how and why these brokers make particular choices,” suggests Kate.
Multi-system hallucinations might be actual
As we progress alongside the agentic AI gradient, brokers will work with one another.
In response to Tim, there are just a few dangers to look at for when this occurs: they usually have to change credentials to truly carry out duties inside a multi-step course of. That would pose safety dangers as not all agentic platforms have the identical degree of trustworthiness.
“Reasoning errors (assume hallucinations) have exponential affect as they unfold throughout agentic groups,” says Tim. “Consider how statements can get distorted as repeated throughout a series of human beings.”
Agentic options are safe, declare sellers
Responding to those considerations, main AI agent sellers Salesforce and Zendesk declare their options characteristic safety plug-ins past these historically deployed for AI instruments. They are saying people nonetheless management the wheel, buyer information is secure, and workplaces are metamorphosing into extra related and productive areas.
Salesforce
“Context is the king for correct, personalised AI outputs,” says Leandro. “With out real-world information about what you are promoting and your prospects, agent responses are generalized or, worse, depend on hallucinations and guesswork. Knowledge is crucial, however so is its safe and moral dealing with.”
He explains that they developed the Einstein Belief Layer at Salesforce, which secures and anonymizes information to stop leaks. “Transparency can also be constructed into Agentforce. These with digital labor on their groups can simply evaluate the reasoning behind agent outputs and outline the scope of agent tasks in pure language,” he provides.
Zendesk
In Zendesk’s case, the human-in-the-loop method is integral to utilizing agentic AI.
Jason explains the method and says they’ve configurable thresholds that permit human brokers or directors to evaluate and approve AI-generated content material and prompt actions.
“Any high-risk motion, like issuing refunds or making account modifications, might be configured to all the time have a human operator evaluate and make sure it,” he provides.
Amplify human potential, not simply automate duties
As for governance round AI brokers, Kate says, “Cease ready for good laws — they will not come. As an alternative, construct governance frameworks that put human outcomes first.”
“Sure, doc your processes. Sure, set up clear accountability. However the true work is creating techniques that amplify human potential slightly than simply automate human duties.”
“Your ethics board ought to seem like your buyer base, not your govt staff. Deliver within the skeptics, the philosophers, the social scientists — and most significantly, representatives from the communities your AI techniques will have an effect on.”
Kate O’Neill
Founder and chief tech humanist at KO Insights
She believes that the businesses that thrive will not be those with essentially the most refined AI — they’re going to be those who constructed essentially the most considerate guardrails round it.
AI or human: study when to modify gears
The important thing ability of the long run is not writing prompts or managing AI — it is the flexibility to collaborate with AI to resolve more and more complicated challenges, believes Sreelesh.
Marshall McLuhan, a Canadian communications theorist, was prescient in his statement when he mentioned, ‘We form our instruments, and thereafter, our instruments form us.”
That is precisely what’s taking place with AI. We have created these instruments to reinforce our capabilities, and now they’re reshaping how we work, assume, and clear up issues, says Sreelesh. It’s altering how intelligence itself operates throughout organizations.
Having essentially the most superior AI will not assure success — what is going to set organizations aside will likely be their skill to stability human and synthetic judgment prudently.
This requires creating a option to orchestrate human intelligence and feelings into agentic AI-driven decision-making processes. This won’t solely delight prospects however assist improve worker expertise.
With inputs from Yukta Rustagi, Matthew Miller, and Brett Nehls of G2.
Do you know G2 has its personal AI SDR with a mean conversion charge of 30%? Strive it out right here.
Edited by Supanna Das