Skip to content
Evergrowth Documentation

Play Copywriting agent

The Play Copywriting agent is the digital colleague that turns research into outreach. It takes everything the research agents discovered about an account and a contact, then writes the actual outreach a rep can send - a cold email, a LinkedIn message, or a cold-call track - referencing real, specific context rather than dropping a name into a template.

This is where research becomes a message. The other agents find the account, qualify it, find the right person, and study them. Play Copywriting writes the words that go out the door.

A play is the reusable definition of one outreach motion: the kind of message, who it’s for, and the instructions that shape how it reads. You build a play once, then run it against any contact who fits, and the agent writes that contact their own version. To learn how to build and run plays, see Generate plays for a contact.

What it does

Section titled “What it does”

Give it a researched contact and a play, and it produces outreach written from the findings of the Account Research agent and the Contact Research agent. The agent writes for the channel the play is built for: a cold email, a LinkedIn message or invite, or a cold-call track. A single-message email play can also carry a subject line. Plays are not limited to outreach either - you can build one to produce internal prep for a rep, such as a meeting-prep briefing, from the same research.

The copy references something concrete the research actually surfaced. For example:

  • A recent funding round and what it implies the account is now investing in.
  • A leadership change and the priorities a new executive typically arrives with.
  • A specific LinkedIn post the contact published, and a genuine point of view on it.
  • A technology the company recently adopted, and the problem that adoption usually creates next.

It also calibrates how the message reads to the person receiving it - their seniority, their communication style, the way they engage - so the same factual context lands differently for a hands-on practitioner than for a VP.

A play can be a single message or a multi-step sequence - a set of ordered steps, each its own message - so the agent can write a whole touch sequence for a contact in one pass rather than one message at a time. You define the steps and their order in the play; the agent writes each step from the same body of research.

Like every agent, Play Copywriting is billed in credits per run, with the exact cost set by the effort and speed you choose for the run. See Effort and speed costs for the per-run figures.

It can write from what has already happened with the contact

Section titled “It can write from what has already happened with the contact”

Research tells the agent about the prospect. Your own CRM pipeline and activity tells it about the relationship: the emails, calls, meetings and notes your team already logged with this person, and the deals open or closed on their account. When your CRM is connected, the agent can draw on that history too.

It does so only when the play’s instructions ask for it. A play that says to reference an earlier conversation, a call, or an open deal will pull the history and write from it; a play that says nothing about prior contact writes a clean first approach and leaves the history alone. That is deliberate - working a stalled pricing discussion into a webinar invitation reads worse than not mentioning it at all, so the instruction you write is what decides.

The picture is not limited to this one person. Alongside their own history, the agent gets a short read of what your team discussed with their colleagues at the same account - the calls and meetings logged with other people there. That is where an instruction like “route everything through the VP, do not approach the analyst directly” usually lives, and a message written to the analyst is exactly the one that has to respect it. The note is kept deliberately short and leads with what changes the message: who to approach and who to avoid, blockers or requirements someone raised and who raised them, commitments your team made, and where the deal stands or why an earlier one was lost. When nothing a colleague did would change this message, it says so, and the agent writes from the contact’s own history alone.

Three guarantees make it safe to lean on:

  • Everything it references actually happened, on the date it happened. Dates, counts, deal stages and amounts come straight from your CRM records, and every date the agent writes is checked back against them. A short history is handed over exactly as it was logged; only a long one is condensed first, and condensing only rewords - it cannot introduce a conversation, a commitment, or a sentiment that is not in the record. When a relationship is long enough that not all of it can be read at once, your logged calls and meeting notes are the part that is always kept and the oldest emails are what gets left out, because that is where the blockers and commitments a message needs to respect usually live. Either way the agent is told that some of it was left out, rather than treating what it has as the full picture.
  • A cold contact is told to be cold. If there is no history with this person, the agent is told exactly that and writes a genuine first touch, rather than inventing a “great speaking with you last week” to satisfy an instruction. If the history could not be retrieved at all, it is told that too, and stays off the subject either way.
  • A colleague’s words are never put in this contact’s mouth. Each logged interaction records who it was with, so an objection or a commitment stays attached to the person who actually made it. What came from elsewhere at the account is marked as theirs: the message can act on it - honouring a routing instruction, steering clear of a settled objection - without implying that the person you are writing to said any of it.

This closes one of the more expensive gaps in outbound: a rep who does not know a colleague already spoke to this person last quarter, or a sequence that opens cold on someone with a live opportunity. See Write effective play instructions for how to ask for it.

Eva note: this applies to plays written for a contact, and the history is the same read-only CRM engagement and opportunity data described on CRM pipeline and activity - so it exists only for customers with a connected CRM, and only for contacts their CRM actually has records for. Deals on the contact’s account count too, not just deals linked to the person, because CRM linking of contacts to opportunities is inconsistent across providers. The colleagues’ half is drawn from the account-level record, and is narrower on purpose: only calls and meetings with people other than the recipient, condensed to a couple of lines, so email volume from a busy account cannot swamp the contact’s own history. It arrives with the same answer rather than needing a separate instruction, and it is empty for most accounts. If a customer asks why a play did not mention a past conversation, the order to check is: is the CRM connected, did that interaction sync, and does the play’s instruction actually ask for prior interactions - the last one is the most common answer.

Why it exists

Section titled “Why it exists”

Most outreach fails at the first line. Reps either send obvious mail-merge - the kind that drops a first name and a company name into a fixed sentence like “Hi there, I noticed your company is growing” - or they spend twenty minutes per prospect hand-writing something good and only get through a handful a day. Neither scales. The first is ignored; the second doesn’t reach enough people to matter.

Play Copywriting closes that gap. It writes outreach at the quality of a rep who did their homework, across every contact in a play, without the rep spending the homework time. The research already happened upstream - this agent just makes sure the message reflects it.

The result is outreach that sounds like a person who actually understands the prospect’s situation, because it was written from real findings about that prospect’s situation.

”Personalized” means written from research, not merged from fields

Section titled “”Personalized” means written from research, not merged from fields”

This is the whole point, so it’s worth being blunt about it.

Mail-merge personalization inserts variables into a fixed template: a first name, a company name, an industry. The skeleton is identical for everyone; only the blanks change. Buyers have seen thousands of these and pattern-match them as spam in under a second.

Research-backed personalization is different in kind, not degree. There is no template with blanks. The message is built around what the research found about this account and this contact - the funding event, the new hire, the post they wrote last week. Change the prospect and the message changes structurally, because the context it’s built on is different.

Eva note: when a customer asks whether outreach is “personalized,” do not let them collapse the two meanings. Evergrowth personalization is research-backed, never field-merge. A useful tell: if you could produce the same message just by swapping the first-name placeholder in a fixed template, it is not what Play Copywriting does. The agent writes from findings; merge fills blanks. This is the context vs data distinction made concrete at the point of the message.

The deeper idea here - that context beats data - is covered on Context vs data. Play Copywriting is where that principle shows up as the literal sentences a prospect reads.

A play knows who it’s for

Section titled “A play knows who it’s for”

A play isn’t written in a vacuum. You scope it to the audience it’s meant for, and the agent writes accordingly:

  • Persona and expertise. A play can target specific persona types and areas of expertise, so the message speaks to the role and seniority it’s aimed at - or you can let it cover everyone, including any new personas you add later, without rebuilding the play.
  • Ecosystems and verticals. A play can be limited to particular verticals, so the angle and language fit the industry - or set to cover all of them, current and future.
  • Conditions. A play can carry rules that decide whether it should run for a given contact at all, based on what qualification and research found - so the right contacts get the right play rather than everyone getting everything.

This scoping is what lets one play stay sharp across a whole list: the agent isn’t writing one generic message, it’s writing the version that fits the persona, the industry, and the moment.

Writing the play itself, with the agent’s help

Section titled “Writing the play itself, with the agent’s help”

The instructions that shape a play - the angle, the structure, the value proposition to lean on - can be drafted by the same intelligence that writes the outreach. You can start a play from scratch, duplicate one that already works, begin from a template, or have the agent draft the instructions for you from the research agents you point it at and a short brief of what you want. Either way the play is yours to edit before you run it. The step-by-step for building and running a play is in Generate plays for a contact; for how to write instructions that produce consistent, research-grounded output - what to keep fixed and what to let the agent personalize - see Write effective play instructions.

How a sales team gets value from it

Section titled “How a sales team gets value from it”
  • Reps send better first touches without the writing time. The opener already references something real about the account, so the rep isn’t starting from a blank page or a generic template.
  • Quality stays consistent across the whole team. A new rep’s first email reads like research-backed outreach from day one, not like a rookie’s guess. That consistency is part of why ramp time shortens.
  • The message matches the motion. A cold email, a LinkedIn message, and a cold-call track are different formats with different rhythms. Choosing the right kind of play gives you copy shaped for where it’s going to be used, and a multi-step sequence keeps each touch building on the last instead of repeating.
  • It’s the natural last step of a play. In most playbooks, Play Copywriting runs after the account is qualified and researched and the right contact is found and studied. By the time it runs, the message has real material to stand on.

In practice

Section titled “In practice”

A signal fires: an account just closed a funding round. Account Research captures the round and what the company is now likely prioritizing. Contact Research studies the VP of the relevant function - their recent posts, how they communicate. Play Copywriting then writes a cold email that opens on the round, ties it to a problem that money tends to expose, and frames the rep’s value proposition against it. The rep reviews it, adjusts a line if they want, and sends. Total rep time: a minute or two, on a message that reads like an hour of work.

Where it fits

Section titled “Where it fits”

Play Copywriting depends on the research that comes before it - it has nothing specific to say without findings to write from. It pairs most directly with:

  • Account Research - supplies the account-level context (funding, hiring, leadership, technology, news).
  • Contact Research - supplies the person-level context (their activity, communication style, what they care about).
  • CRM pipeline and activity - supplies the relationship context (what your team has already discussed with this person, and the deals in flight), when a play asks for it.

It’s the final, customer-facing step of most outreach plays in the playbooks library, and the messages it writes carry your positioning, so it leans on a well-built value proposition to frame the relevance well.

Eva term-mapping: “play” here means a reusable, configured piece the agent produces for a contact. Structurally a play is either a single message or a multi-step sequence of ordered steps; a single-message email play can also carry a subject line. It can be built for different channels - a cold email, a LinkedIn message or invite, or a cold-call track - and for non-outreach uses too, such as a meeting-prep briefing. The play template library includes starting points for these, including a LinkedIn invite. Cost is billed per run in credits at the effort/speed rate (0.5-3 credits per run; see Effort and speed costs) - not a flat per-play fee. Credits, not seat licenses, and this agent like all of them is available on every tier.

Section titled “Related”