Latest Update: September 30, 2026. Details about Meta Muse and ChatGPT dots come from each company’s announcements and may change as the products roll out.
Caption: A person sets a goal, and a personal AI agent works through the steps between the request and the result.
Key takeaways
An agentic journey is the path a goal takes when an AI agent does the work between a person’s request and the result. The agent plans, researches, compares, acts and follows through.
It runs as a loop of six stages: goal, plan, discover, evaluate, act and follow through. The person sets limits and approves sensitive steps.
Agentic commerce is the buying part of that path. An agentic journey includes commerce and also covers planning, research and scheduling that never reach a checkout.
Meta Muse (introduced September 8, 2026) and OpenAI’s dots in ChatGPT (announced September 29, 2026) are personal agents built to keep working while the person is away.
Opttab’s verdict: agentic journey matters more in 2027 and beyond as personal agents improve.
What is an agentic journey?
An agentic journey is the path a goal takes when an AI agent does the work between the request and the result. The person states what they want and sets limits such as budget, dates and preferences. The agent researches options, compares them, plans the steps, takes actions such as booking or buying, and follows through when something changes. The person approves the steps that matter.
For a brand, the same path looks different. The agent is the visitor. It reads your site and your data, compares you with competitors, and decides whether to recommend you or complete a purchase. Opttab defines the agentic journey from this side as a structured path with four pillars: discoverability, visibility, optimization and agentic commerce. The original framing appears in Opttab’s guide to agentic commerce. The phrase is recent, so this article uses that definition throughout.
Compare it with a traditional journey. A person opens a browser, runs a search, opens a dozen tabs, compares options by hand, fills in forms and completes checkout. In an agentic journey the person describes the goal once. The agent handles the tabs, comparisons and forms, then asks for approval before the final action.
Caption: A traditional journey has many manual steps. In an agentic journey the agent handles them and the person approves.
Why personal agents make this real now
Earlier assistants could already book a reservation or buy a plane ticket. The past month brought two products built around an agent that stays with a person and keeps working in the background.
Meta Muse
Meta introduced Muse on September 8, 2026 and describes it as a personal AI agent that does the work instead of only answering questions.
Muse runs on its own secure virtual machine with its own browser. People talk to it in the Muse app or in WhatsApp.
After a person shares a goal, Muse builds a plan and advances the work on its own. Meta says it can open a browser, fill out forms and negotiate on the person’s behalf.
It keeps working after the app closes and comes back when something changes or when it needs approval, for example before it sends an email or makes a purchase.
Muse checks out with Link by Stripe, and Meta says it is the first AI agent covered by Link’s purchase protections.
It is rolling out in the US on iOS, Android and muse.ai, with AI glasses listed as coming soon. Most features are free, with subscriptions for more.
Watch:Take the full tour of Muse, Meta’s personal AI agent (2 minutes 37 seconds, posted by Muse on YouTube). It shows the chat tab, proactive ideas and the goals tab where Muse tracks what a person is working toward.
ChatGPT dots
OpenAI announced dots on September 29, 2026 and describes them as always on agents.
Each dot has its own cloud computer and browser, learns from feedback over time and works toward a person’s goals around the clock.
Dots connect to more than 4,000 apps through plugins. People reach them through ChatGPT, Slack, Teams or a voice call.
A dot can act before it is asked. In OpenAI’s example, an early tester’s dot noticed he had forgotten to invoice a publication, prepared the invoice and sent it after his approval.
Dots are rolling out across Pro, Business Premium and Enterprise plans in eligible markets. OpenAI says about 1.2 billion people use ChatGPT each week.
Watch:OpenAI DevDay 2026 keynote (53 minutes, posted by OpenAI). Dots are the first announcement, including a scene where a dot finds a backup wedding cake vendor after the first one cancels and then proposes a tasting time.
What the two agents share
Own computer and browser. The agent works without the person watching.
Memory. Both learn a person’s preferences over time, and Muse acts on details a person mentioned only once.
Approval gates. Muse asks before sending an email or making a purchase. OpenAI says people can set boundaries on app and computer use for dots, and in its invoice example the dot sent the invoice after approval.
Connected apps. The agent works inside tools the person already uses.
These traits turn a single chat answer into a journey that unfolds over days.
How an agentic journey works
Every agentic journey follows the same loop, whichever agent runs it. The diagram below shows the six stages.
Caption: The six stages of an agentic journey: 1 goal, 2 plan, 3 discover, 4 evaluate, 5 act with approval, 6 follow through and learn.
Goal. The person states a goal in plain language and sets limits. Some agents propose goals too. Muse has a tab of proactive ideas, and OpenAI says dots can bring work done “sometimes before you even think to ask.”
Plan. The agent breaks the goal into tasks and constraints. Meta says Muse develops a personalized plan and coordinates the person’s time and resources.
Discover. The agent searches websites, catalogs, reviews, apps and connected tools for options. A brand the agent cannot reach or read drops out here.
Evaluate. The agent compares options against the person’s criteria. Opttab’s guide to agentic search explains that the agent researches a brand across several sources and judges it against competitors, using information the brand may not control, such as reviews and third party articles.
Act. The agent books, registers, fills in forms or buys, within the limits the person set. It asks for approval before sensitive steps.
Follow through. After acting, the agent watches for changes such as price drops, cancellations or schedule conflicts, updates the plan and learns from feedback. Google’s Universal Cart, for example, tracks price changes and alerts shoppers when an item is back in stock.
An example: training for a half marathon
This is an illustration, not a customer case. A person tells their agent: “I want to run a half marathon in April, and I am away the first weekend of March.”
Stage
What the agent does
Who it touches
1. Goal
Captures the goal, the date limit and a budget
No brand yet
2. Plan
Drafts a training schedule and a task list for race, shoes and travel
Training content
3. Discover
Finds races that fit the date, shoe options and hotels near the start
Race sites, retailers, hotels
4. Evaluate
Compares courses, reviews, prices, return policies and delivery dates
Review sites, articles, product pages
5. Act
Registers for the race, books a hotel and buys shoes after the person approves the total
Checkout systems, payment providers
6. Follow through
Flags a price drop, adjusts the plan after a missed week and sends race day reminders
Email, SMS, loyalty programs, support
Purchases sit in one stage of six. The other five involve research, planning, comparison and follow through, which is why a term wider than agentic commerce is useful.
Agentic commerce vs agentic journey
Opttab defines agentic commerce as an AI agent acting as an autonomous handler that does product research, discovery, comparison and even the purchase on behalf of a buyer. Its agentic commerce guide gives an example: a shopper asks for noise cancelling earbuds under $200 that arrive by Thursday, and the agent interprets the request, queries catalogs, curates options, refines them with feedback and completes checkout.
Standards for this layer are taking shape. The Agentic Commerce Protocol (ACP), developed by OpenAI and Stripe, powers Instant Checkout in ChatGPT, the first product built on it. Google has announced the Universal Commerce Protocol (UCP) and, in May 2026, a Universal Cart built on it. Google said the cart would roll out in the US to Search and the Gemini app that summer, with YouTube and Gmail to follow.
Watch:The new era of agentic commerce is here (2 minutes 18 seconds, Google Ads on YouTube). It covers Universal Cart and UCP, the commerce layer of the journey.
Agentic commerce sits inside the agentic journey. Every agentic commerce transaction happens within a journey, and a journey can finish without any transaction.
Caption: Agentic commerce is one stage inside the wider agentic journey.
Dimension
Agentic commerce
Agentic journey
Scope
Finding, comparing and buying products or services for a buyer
The whole path from goal to result, including planning, research, booking, buying and follow through
Starting point
A shopping request such as “find noise cancelling earbuds under $200 by Thursday”
A goal such as “run a half marathon in April”
Ends with
A completed order
The result the person wanted, with or without an order
Standards and tools
ACP, UCP, payment tools such as the Agent Payments Protocol (AP2), catalog access through the Model Context Protocol (MCP)
The same commerce standards for the purchase stage, plus agent access, visibility in AI answers, content and data quality, and follow through
Brand question
Can an agent complete a purchase with us?
Can an agent find us, understand us, choose us, transact with us and keep the relationship afterward?
Agent and crawler visits, AI visibility and citations, agent transactions, recovery of abandoned carts
Why the distinction matters today
Shoppers say they trust agents with the front half of a journey and hesitate at the last step. Acosta Group asked 1,271 US shoppers in May 2026 which tasks they would trust a generative AI assistant to handle, assuming it follows their criteria. These are the answers from shoppers who use generative AI for shopping:
Task
Share who would trust an agent
Finding the best deal on a product
71%
Comparing product reviews
70%
Product recommendations
69%
Gift suggestions
41%
Reordering groceries or household essentials
24%
Making purchases with their account or card
22%
Source: Acosta Group Shopper Insights, AI-Driven Shopping Evolves, survey fielded May 15 to 21, 2026.
Most of the trust today sits in research and comparison. A brand that focuses only on checkout misses the stages where the agent shapes the shortlist. OpenAI made a related point in its DevDay keynote, describing tasks like booking a reservation or buying a plane ticket as useful to hand off but small next to what the technology can do.
What agentic journeys mean for brands
Opttab positions its platform as an Agentic Journey OS that lets brands manage four pillars. Each pillar matches something an agent needs at a point in the journey.
Pillar
What the agent needs
What to do
Discoverability
To reach your site, catalog and tools
Allow and monitor agent access, serve pages prepared for bots, offer an MCP server
Visibility
To find and cite you when it answers buyer questions
Track prompts, competitors and cited sources for each AI engine
Optimization
Complete, accurate content and product data
Fix feeds, answer buyer questions, keep price and stock current
Agentic commerce
A way to complete the purchase
Publish an MCP server, WebMCP, plugins or connectors, and track agent transactions
Discoverability. An agent has to reach you before anything else happens. Opttab’s Agent Analytics shows how human visitors, AI crawlers and LLM bots access your site. AXP generates versions of your pages prepared for bots and points bots to them through a proxy setup.
Visibility. Engines cite very different sources. In Opttab’s study of 403,588 citation events in the AI visibility software category (June 12 to September 8, 2026), the number of distinct pages cited ranged from 730 for Grok to 30,628 for Google AI Mode. Video made up 5.56% of Google AI Mode’s citations and 0.03% of ChatGPT’s, and effectively all of it was YouTube. Measure visibility for each engine separately, because a blended number hides these gaps. For the basics, read what AI visibility is, and track your own results with AI visibility, which covers competitors, sentiment, citations and analytics.
Optimization. Data quality decides whether an agent can use what it finds. One independent 2026 benchmark, cited in Opttab’s guide to AI search for ecommerce, put AI search at 14.7% of product discovery traffic in Q1 2026, up from 4.2% in Q1 2025. In the same dataset only 23% of Shopify stores had adequate structured data for AI engines. The Feed Optimizer in Opttab’s AI Commerce suite validates catalog quality and raises feed health scores so the catalog suits ACP and UCP.
Agentic commerce. This pillar covers the purchase and what follows it. Opttab’s agent tools let brands publish a remote MCP server, WebMCP and a ChatGPT plugin or Claude connector so the catalog and agent tools are available where buyers ask. Transactions completed through agents appear in Agent Insights. The AI Commerce suite also includes an Agentic Commerce Readiness Audit, which scores readiness from 0 to 100 across feeds, policies, promotions, checkout and MCP server maturity. When a buyer abandons a cart, Opttab can connect that journey to a customer data platform so Insider One, Klaviyo or Salesforce Marketing Cloud can continue it. For a wider view of the tooling, see Opttab’s list of agentic AI tools for commerce automation in 2026.
Verdict: agentic journey matters more in 2027 and beyond
Opttab expects agentic journey to become a central part of how people find and choose brands from 2027 onward, as personal agents such as Meta Muse and ChatGPT dots improve. This is a forecast. Five developments support it.
Personal agents now sit inside large apps. Muse works in the Muse app and in WhatsApp. Dots live inside ChatGPT, which OpenAI says about 1.2 billion people use each week. The two were announced three weeks apart.
The agents work out of sight. Both run on their own computer, continue after the person leaves and return for approval. More research and comparison happens before a brand sees a human visit.
Shopper habits are moving. In Acosta Group’s survey, 61% of US shoppers used generative AI in 2026, up from 49% in 2025. Among AI tool users, 39% said AI influenced something they bought in the past three months, including 52% of Gen Z and 50% of Millennials.
The purchase step is getting shared rails. ACP, UCP and Link by Stripe give agents defined ways to pay, and Muse pairs checkout with purchase protections from Link. When checkout works the same way across brands, the stages before and after it decide who wins.
Trust in purchases has room to grow. Only 22% of shoppers who use generative AI for shopping would trust an agent to buy with their account or card, compared with 69% to 71% for deals, reviews and recommendations. Among Millennials, 35% would. If trust in payments moves toward the level of trust in research, more of the journey will be delegated.
What could slow this down. Three facts point the other way. Only 22% trust agent purchases today. Among AI tool users, 56% are highly concerned that AI recommendations are influenced by ads. Both products are early: Muse is rolling out in the US first, and dots are rolling out on paid plans in eligible markets. Familiar brands also keep an edge. When AI recommends a brand a person does not know, 45% of AI tool users said they would definitely or probably stick with a brand they know, and 4% said they would buy the unfamiliar one.
The direction is clear even if the timing is not. As personal agents improve, more journeys will start with a goal handed to an agent, and brands will be found or skipped in stages they never see. Opttab expects that shift to be material in 2027 and to keep growing after that.
What to do now
Map the journey. For your five most valuable customer goals, list what an agent would search, compare and need in order to complete a purchase or booking.
Check agent access. Review bot and agent visits with Agent Analytics, and consider serving pages prepared for bots through AXP.
Measure visibility by engine. Track prompts, competitors and cited sources with AI visibility, and compare your results with the patterns in how LLMs cite sources.
Clean your data and publish agent tools. Fix titles, attributes, price, stock, shipping and return policies, then test readiness with AI Commerce.
Connect follow through. Send agent transactions and abandoned carts to your customer data platform so your marketing tools can continue the journey. The agentic commerce guide explains the setup.
An agentic journey is the path a goal takes when an AI agent does the work between a person’s request and the result. The agent plans, researches, compares, acts and follows through, and the person sets limits and approves sensitive steps.
How does an agentic journey work?
It runs as a loop of six stages: goal, plan, discover, evaluate, act and follow through. The agent uses its own computer, browser and connected apps, and it goes back to the person for approval before actions such as sending an email or making a purchase.
What is the difference between agentic commerce and agentic journey?
Agentic commerce covers an agent finding, comparing and buying products or services for a buyer. An agentic journey covers the whole path from goal to result, and commerce is one stage inside it. A journey can also include planning, scheduling and support that never reach a checkout.
Is an agentic journey the same as agentic search?
Agentic search is AI that retrieves, evaluates and acts on information on behalf of users. It covers the discover and evaluate stages of a journey. Opttab explains it in What is agentic search?.
How is an agentic journey different from a customer journey?
A customer journey maps what a person does across stages such as awareness, consideration and purchase. An agentic journey maps what an agent does for the person across the same ground and marks where the person still approves. Brands need both views because the agent may make the first visit to your site. For how this connects to search strategy, see SEO vs GEO vs AEO.
Which brands does this affect?
Any brand that sells something an agent could research, compare or book. Retail and ecommerce teams can start with Opttab’s ecommerce page.
When will agentic journeys matter most?
Opttab expects their importance to grow from 2027 onward as personal agents such as Meta Muse and ChatGPT dots reach more people and as trust in agent purchases rises. Today 22% of shoppers who use generative AI for shopping would trust an agent to buy for them, and 35% of Millennials would. This is a forecast.
How do I measure an agentic journey?
Track agent and crawler visits to your site, your visibility and citations in AI answers, transactions completed through agents and recovery of abandoned carts. Agent Analytics and AI visibility cover the first two.
To provide the best experiences, we use technologies like cookies to store and/or access device information. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
Functional
Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes.The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
Sep 30,2026 
By Emily Danilova 
