GTM 1.0 vs GTM 2.0: what actually changes
In GTM 1.0, growth meant more reps and more tools: a data provider, an intent feed, a sequencer, a LinkedIn tool, a dialer, and an operations person keeping them in sync. Every tool worked. The pipeline still depended on a rep finding the time to run them all, and on someone logging the result.
In GTM 2.0, AI for GTM stops being a feature inside each tool and becomes the layer that runs the work between them. Agents decide which accounts are in market, research them, write and send, dial, and write the outcome back to the CRM. People stop operating the machine and start supervising it, and the scarce human hours go to live conversations, deals and decisions.
| GTM 1.0 | GTM 2.0 | |
|---|---|---|
| Unit of capacity | Rep hours | Agent capacity plus rep conversations |
| Who finds accounts | Reps and ops pull lists | Agents build and re-score the TAM from your ICP and wins |
| Who decides "why now" | Reps scan alerts when they have time | Agents act when a signal crosses a threshold you set |
| Who sends and dials | Reps, one sequence step at a time | Agents run the steps; reps take live conversations |
| CRM | Updated when reps remember | Updated by the agent after every touch |
| Where humans add the most | Everywhere, thinly | Conversations, approvals, strategy |
Adoption is already broad. In Salesforce's State of Sales survey (a vendor survey of 4,050 sales professionals, August to September 2025), 87% of sales organizations use some form of AI and 54% of sellers say they have used agents. Gartner also predicts that by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024. The open question is not whether agents arrive in your stack, but whether they do real work.
The GTM AI maturity model: stage 0 to stage 4
Use this model to place your team honestly. Each stage is defined by who does the work, not by which tools you own. Many teams own stage 3 software and run it at stage 2 because nobody trusts it to act.
| Stage | Research and signals | Outreach and dialing | Conversations and decisions | CRM | What holds you back |
|---|---|---|---|---|---|
| 0. Manual | Reps build lists by hand | Reps write and dial one by one | Reps | Logged by hand, often not at all | Rep hours |
| 1. Tool-assisted (GTM 1.0) | Data and intent tools; ops stitches them | Sequencer and dialer; reps run every step | Reps | Partly synced by each tool | Glue work and stale data |
| 2. Copilot | AI summarizes accounts and suggests contacts | AI drafts emails and call notes; reps send and dial | Reps | AI suggests updates; reps accept | AI saves minutes, not steps |
| 3. Agent-assisted | Agents watch signals and research on a trigger | Agents run defined plays with approval gates | Reps, plus approving agent work | Agents write back with an audit log | Trust, approvals, data quality |
| 4. Agent-run, human conversations | Agents build and re-score the TAM continuously | Agents run email, LinkedIn, SMS and the dialing; humans take every live conversation | Humans decide, coach agents and close | Kept current by agents; humans review exceptions | Strategy and conversation quality |
Two things stay constant at every stage. The conversation with a buyer is held by a person, and a person decides what the agents are allowed to do. Stage 4 is not "no humans". It is humans spending their time only on the parts a buyer can tell a human did.
What changes for each role in agentic GTM
| Role | Less of | More of | Measure them on |
|---|---|---|---|
| SDR | List building, research tabs, manual dials, logging | Live conversations with in-market accounts; reviewing agent drafts | Conversations held, meetings held |
| AE | Prep, follow-up drafting, CRM updates after calls | Discovery, multi-threading, negotiation | Win rate, cycle time |
| RevOps | Fixing duplicates and stage data by hand | Defining fields, permissions and approval rules agents follow | Data completeness, forecast accuracy |
| GTM engineer | One-off enrichment scripts | Building signal plays, routing and agent workflows | Plays shipped, pipeline per play |
| GTM leader | Adding headcount to add capacity | Choosing plays, setting guardrails, deciding where humans stay | Pipeline per dollar, conversations per rep |
The SDR job shrinks in admin and grows in conversation. Salesforce's survey reports the average seller spends 40% of their time selling and that nearly half of reps name cold calling as the worst part of the job; in GTM 2.0 the agent takes the dialing and the dead time, and the rep keeps the part that needs a human. The GTM engineer becomes the person who designs what agents do, which is why we wrote a separate guide on what a GTM engineer does.
For leaders, the planning question changes from "how many reps" to "how many conversations, and who holds them". Gartner warns of a value ceiling: by 2028 it predicts AI agents will outnumber sellers tenfold, yet fewer than 40% of sellers will report that agents improved their productivity. More agents is not the goal. Fewer, better plays are.
Must-have agent capabilities vs the agentic features buyers now ask for
Separate the foundations from the features. Foundations decide whether an agent is safe to let act; features decide how much it can do once it is.
Must-haves (do not buy without them)
- Two-way CRM write-back to your real objects, custom fields and stages, deduped against what is already there.
- Approval controls you can set per play: review before send, banned phrases, budget or volume caps, and an instant pause.
- An audit trail of every signal, message, call and field change, with the reason the agent acted.
- Signal intake from more than one source (hiring, funding, job changes, site visits, tech stack) with thresholds you control.
- Every channel your buyers use, including the phone. An agent that cannot get a buyer to a live rep leaves your best channel to manual work.
- Deliverability infrastructure: authenticated domains, warm mailboxes and volume pacing.
Agentic features buyers are asking for
- Next best action: a recommended, and optionally executed, next step per deal or account from its calls, emails and signals. Deep dive: next best action in sales.
- Signal-triggered plays: when an account crosses a threshold, a specific sequence starts, with the signal written into the first line.
- CRM hygiene: agents that find duplicates, stale fields, missing contacts and wrong stages and fix them. See agentic CRM.
- Speed to lead: an inbound lead gets a call or text within minutes and is booked or routed, instead of waiting in a queue.
Where AI agents fail in go-to-market
Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of escalating costs, unclear business value or inadequate risk controls. It also estimates only about 130 of the thousands of vendors selling agentic AI are real, calling the rest "agent washing". In GTM, failures cluster in four places.
1. Bad data
An agent acting on duplicates and stale fields sends the wrong message to the wrong person, faster. In Salesforce's survey, 51% of sales leaders with AI say disconnected systems are slowing their AI initiatives, and 74% of sales professionals are focusing on data cleansing. Fix ownership, duplicates and stage definitions before agents write to the CRM.
2. No approvals
Agents switched to fully autonomous on day one are how brands end up apologizing in public. Start every play with review before send, then loosen approvals per play only once quality holds for several weeks. If a vendor cannot show you per-play approvals and a pause button, it fails the risk-control test Gartner describes.
3. Email-only agents
Most AI SDR products execute email and stop there; on the phone they give reps a dialer or nothing. That leaves the buyers who decide by conversation to manual work. AI voice is not a shortcut for cold calls either: the FCC treats AI-generated voices as artificial under the TCPA, so they generally need prior consent. The workable pattern is an agent that dials and a human who talks; our AI SDR guide covers the detail. This is not legal advice.
4. Deliverability
Agents can send far more email than a team ever could, which is exactly how domains get burned. Google's Gmail sender guidelines require SPF or DKIM for all senders, add DMARC and one-click unsubscribe for anyone sending more than 5,000 messages a day to Gmail, and say to keep reported spam rates below 0.3%. Pace volume, warm new domains and treat spam rate as a stop signal for the agent.
A 30-60-90 day plan to adopt GTM AI
- Days 1-30: fix the data and pick one playWrite your ICP down. Dedupe accounts and contacts, agree stage definitions and field owners, and decide which fields agents may write. Pick one play with a clear trigger and outcome (for example, a hiring signal in your ICP leads to a three-touch sequence and a call). Set up authenticated domains and warm mailboxes. Write down the baseline: meetings held per week and pipeline created.
- Days 31-60: run the play with approvals onLet the agent research, draft, sequence and dial, with review before every send. Reps take every live conversation. Review weekly: approval edits, reply quality, spam rate, meetings held and CRM accuracy. Kill or rewrite anything that is not working.
- Days 61-90: widen channels, loosen approvals where earnedAdd a second play and a second channel. Move approvals from every message to sampled review only on plays that held quality. Add CRM hygiene and speed-to-lead workflows if the data holds. Compare against the baseline: pipeline per dollar and conversations per rep, not messages sent.
Frequently asked questions
What is agentic GTM?
Agentic GTM is go-to-market where AI agents act, not just advise: they build the account list, watch signals, research, run outreach and update the CRM inside limits you set. Humans hold the conversations, approve what matters and make the decisions.
How is AI used in GTM today?
Most teams use AI as a copilot: summarizing accounts, drafting emails and call notes. In Salesforce's vendor survey, 87% of sales organizations use some form of AI. Fewer let agents run whole plays, which is where the bigger change, and the bigger risk, sits.
Will AI agents replace SDRs?
They replace much of the SDR's admin: list building, research, dialing and logging. They do not replace the live conversation. In GTM 2.0 SDRs spend more of their time talking to in-market buyers and reviewing agent work.
What is the difference between GTM 1.0 and GTM 2.0?
GTM 1.0 adds reps and tools, and people stitch the tools together. GTM 2.0 has agents run the work between the tools, research, signals, outreach, dialing and CRM updates, while people hold conversations and decide.
How long does it take to adopt AI agents in go-to-market?
Plan on about 90 days for the first real results: 30 days to fix data and pick one play, 30 to run it with approvals, and 30 to widen channels and loosen approvals where quality held. Data quality decides the pace more than software does.
- 1.Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (prediction and webinar poll), checked 2026-11-10 (2025)
- 2.Gartner, Gartner Predicts By 2028 AI Agents Will Outnumber Sellers by 10X, Yet Fewer Than 40% of Sellers Will Report AI Agents Improved Productivity, checked 2026-11-10 (2025)
- 3.Salesforce (vendor survey), State of Sales report announcement (survey of 4,050 sales professionals, Aug to Sep 2025), checked 2026-11-10 (2026)
- 4.Google, Gmail email sender guidelines, checked 2026-11-10 (2026)
- 5.FCC, FCC Makes AI-Generated Voices in Robocalls Illegal (2024)
- 6.TwinsAI, GTM Agents (2026)
Spend your reps' day on conversations, not ringing phones.
TwinsAI parallel-dials up to 10 lines at once, validates every number with the carrier before it rings, and bridges a live prospect to your rep in under 400ms. It works inside HubSpot, Salesforce, Close, Attio, Outreach, Zoho, HighLevel, Pipedrive, Apollo and SmartMoving, with no seat minimum.