👋 Hi, it’s Kyle Poyar and welcome to Growth Unhinged, your source for practical advice on marketing, pricing, and growth.

Up today: I’m teaming up again with a favorite collaborator, Brendan Short of the excellent The Signal newsletter. Brendan and I revisit our favorite automated GTM plays from last year, and show what’s changed for AI-native teams.

ICYMI: Claude for GTM Pulse Report | Claude Skills for GTM & Pricing Analysis

Outbound has been around for, well, a long time. It feels fresh again because of two forces, data and automation.

There’s so much better data on both who’s your target customer and who’s ready to buy. When asked which GTM channels B2B companies are investing the most into for 2026, intent-based outbound was #2 on the list and warm outbound took the #5 spot. Buyers may hate AI slop cannons, but (good) outbound still pays the bills.

In February 2025, Kyle and I published The best automated GTM plays you’re not running. It was a menu of 25 plays like website visitor outbound, champion tracking, closed-lost re-engagement, and micro-campaigns. That post still holds up. The plays are still good. But the way you build them has changed… a lot.

14 months ago, "automation" meant setting up a signal that gets triggered, enriching a list in Clay, and having AI attempt to write an email. Today, it means AI agents that monitor signals continuously, pull research from your CRM and the web in real-time, and draft outreach that references what the prospect actually said on their last call with you.

Four things shifted:

So we're back. Instead of 25 plays at a surface level, we picked 5 workflows to show you how AI has changed how you set up GTM systems in 2026.

Important to notice: The workflows covered here include a combination of deterministic steps (if-then) and agentic steps (AI acts autonomously).

Original version (2025): Set a CRM automation to flag opportunities closed-lost 9 months ago. Rep reviews, writes a re-engagement email. Maybe they use the OpenAI API to summarize the last call and insert that summary into the email with a line like: "To jog your memory, here's what we talked about last time."

That was a solid play, but it had two problems. The timing was arbitrary. Why 9 months? What if the right moment to re-engage was 4 months (because they got a new VP) or 14 months (because they just raised a round)? And the summary was generic. It told you what was discussed, but not why the deal died.

The AI-native version: An agent monitors your closed-lost pipeline continuously.

The trigger is a cluster of re-engagement signals at the account: leadership change, new funding, job posting for a relevant role, champion who killed the deal left the company. When enough signals stack, the agent fires.

When it fires, the agent pulls the call transcript from the last conversation. It identifies the specific objection that killed the deal. "They went with [competitor] because of [specific feature gap]." Or "budget got pulled because of [reorg]."

Then it drafts an email that references the objection and what's changed. For Tier 1 accounts, the email routes to the rep for review. For Tier 2 and below, it sends automatically.

The key shift: Timing is dynamic instead of calendar-based. And the email references the real reason the deal died, not a generic recap.

Tool chain: CRM (closed-lost pipeline) → Signal sources (job changes, funding, leadership) → Call transcript via API and/or email replies (transcript + objection extraction) → Clay (enrichment + orchestration) → Claude (email drafting) → SEP, CRM sequence, or Slack.

Pro-tip: Stack this with champion tracking. If the person who killed your deal left the company, that's a completely different play than if they're still there. Your agent should know the difference and adjust.

Original version (2025): Run outbound campaigns aimed at small, highly-targeted lists of 50-250 contacts with very specific filtering and signals. Each list is relevant this week, but won't be next month. We gave an example: a company with a database of pre-vetted sales candidates scans target accounts daily, and when any post an AE job, reaches out to the VP of Sales with three candidates.

The concept was right. The execution was still mostly manual. Someone had to decide when a micro-campaign should exist, build the list, write the copy, and launch it. That took hours per campaign.

The AI-native version: The agent decides when a micro-campaign should exist and surfaces it themselves.

You define your ICP and your signal sources (job postings via Sumble, G2 reviews, LinkedIn engagement, first-party data from CRM/MAP/DW, etc.). The agent scans your target account list continuously.

When a signal cluster fires, meaning multiple signals stack on the same account or set of accounts within a short window, the agent:

A concrete example: You sell a data integration tool. Your agent detects that 13 companies posted 2 or more Data Engineer roles mentioning Fivetran (a competitor), and a Key Buyer either liked a LinkedIn post about migrating off Fivetran or found this in call transcripts in your CRM. Three signals stacked, finding 13 companies. The agent auto-prospects the right buyers at these 13 companies and compiles a micro-list, writes copy referencing the migration signal, and stages it. That campaign didn't exist yesterday and won't be relevant in 3 weeks.

The key shift: Micro-campaigns used to require a human to spot the pattern and build the campaign. Now the agent spots the pattern, builds the campaign, and presents it to you for approval (or just runs it).

Tool chain: Signal sources (Sumble, G2) → Clay (orchestration + enrichment + ICP scoring) → Claude (copy generation) → SEP.

Pro-tip: Start with one signal source and one ICP segment. Go with something highly specialized that’s unlikely to be immediately copied. Get the agent working on a narrow slice before expanding.

Original version (2025): Some of the best performing GTM plays are where there’s a pre-existing relationship that can be leveraged like a past champion or a mutual investor relationship. We listed three separate relationship-based plays: past champions change jobs, existing champion's prior company is ICP-fit, and mutual investor or advisor intro.

Each play had its own trigger, tool setup, and manual triage. Three separate workflows and three separate lists to maintain.

The AI-native version: Multiple plays triggered from the same system using a single agent and relationship graph.

The agent builds and continuously updates a relationship graph that maps your warm network: current champions, past champions, their job changes, their prior companies (scored for ICP fit), mutual investors and advisors, and LinkedIn engagement patterns (who's engaging with your content or your competitors' content).

When any node in this graph changes, the agent evaluates all possible warm paths into target accounts and ranks them by likelihood of success.

A few scenarios:

The key shift: You're running a single system that continuously evaluates the best path into every target account. The agent picks the play. You (or your rep) execute it.

Tool chain: Sumble, UserGems, Clay, etc. (job change tracking) → Clay (enrichment + ICP scoring) → The Swarm (relationship graph) → Claude (strength reasoning + outreach drafting) → SEP or Slack alert.

Pro-tip: The most underused warm path is "champion got promoted at the same company." They now have more budget authority and more influence. Your agent should treat a promotion the same way it treats a job change: as a trigger to re-engage with a bigger ask.

Original version (2025): Tech stack data makes for a great signal. We previously recommended using data scraped from job postings to find companies using a competitor or complementary product. Traditional technographic providers scrape websites, which works for tools with a web pixel. For everything else, job postings are the signal (because a Data Engineer role will mention "proficiency in our stack, which includes Snowflake").

That was a good start. But the play was essentially: build a list of competitor users → send competitive messaging. The messaging was usually based on your marketing team's battlecard, not on anything specific to the prospect.

The AI-native version: The agent monitors multiple displacement signal sources continuously:

Any single signal is interesting. When signals stack (negative G2 review + new hire + job posting dropping a competitor from the required skills), that's a displacement opportunity.

When the agent detects a stacked signal, it:

A concrete example: Your agent finds a G2 review from an ops manager at a target account saying their current tool "breaks every time we try to sync more than 10K records." A week later, the company posts a job for a RevOps lead mentioning experience with your product category. The agent builds a 5-person list, and the email to the ops manager opens with a reference to the scaling challenge (without quoting the review verbatim, obviously) and links to a case study from a company that migrated to you and solved that exact problem.

The key shift: The messaging is built from what the prospect's team is actually saying about the competitor, not from your internal positioning doc.

Tool chain: Sumble (tech stack signals from job postings) → G2/Capterra (review monitoring) → Clay (orchestration + enrichment) → Claude (pain point mapping + outreach generation) → SEP.

Pro-tip: The most effective displacement emails lead with the prospect's specific pain and offer a path forward. "I noticed your team is scaling fast and running into [specific challenge pulled from OKR/goal on the job description]. Here's how [similar company] handled that transition..." works better than "Ready to switch from [competitor]?"

Original version (2025): You can usually afford to invest extra effort in personalization for your best Tier 1 prospects. We previously recommended pulling together a research brief before a first call: recent news, competitive intel, relevant context.

In practice, most reps skipped this step. It took 20-30 minutes of manual research per call. The ones who did it were disproportionately effective, but the majority showed up underprepared.

The AI-native version: 30 minutes before every first call (triggered by the calendar event), an agent auto-generates two deliverables:

Both deliverables are pushed to the rep via Slack or email, 30 minutes before the call. They open the brief, skim the deck, and walk into the meeting with more context than 90% of reps who "did their research."

The key shift: Every rep now shows up as prepared as your best rep. The floor rises across the entire team. And the prospect gets a deck that actually reflects their world.

Tool chain: Google Calendar (trigger) → Clay (company + contact enrichment) → Claude (brief generation + deck generation) → Google Slides API or PDF generation → Slack/Email delivery.

Pro-tip: After the call, feed the call transcript back into the agent. Have it update the brief with what was actually discussed, what objections came up, and what the next steps are. Now you have a living account brief that gets smarter after every interaction.

14 months ago, these were plays you built. Today, they're systems that build themselves.

We think in another 12 months, most growth-stage GTM teams will have 5+ of these agent workflows running continuously. Automated GTM plays will become core revenue infrastructure. The closed-lost agent will be as standard as a drip campaign. The micro-campaign agent will replace the Monday morning "what should we target this week" meeting.

The teams that build these systems now will compound their advantage every month. The teams that wait will feel it in their pipeline, their win rates, and their rep productivity.

If you found a play here worth building, start small. Pick one. Get it working on a narrow segment. Then expand. And if you build something creative that we didn't cover, tell us. We want to know.

See you out there.

Brendan 🫡

This was a collaboration between The Signal and Growth Unhinged. Subscribe to both.