# Your agents handle the upkeep. You handle the strategy.

> Calven gives your product marketing team nine AI agents that keep competitor intel, win/loss, ICP, personas, positioning and messaging current, with the source attached to every change. You approve the drafts, and every team works from the version you approved.

- Source: https://calven.ai/solutions/product-marketing
- Last updated: 2026-10-05

---

Nine agents do the routine work. You spend your week on launches, positioning and the big calls.

## Set it up once. Every day after is yours.

Calven agents build your Universe from your company URL and keep it current. You decide what gets published, then use it at every step of your work, in your AI tool.

- **01 Build and own the Universe.** Agents draft every document and keep it current. You decide what gets published.
- **02 Use it every day.** Run launches, win/loss readouts and claim checks in your AI tool, straight from the Universe.

## Step 1: How do the agents build and keep your Universe current?

Like nine new hires who need no babysitting. They run their own backlog and bring you each update to approve.

1. **Build.** Agents research your company, market, CRM and calls.
2. **Review.** Agents turn it all into drafts for your review, each claim linked to its source.
3. **Publish.** You publish the official version. Every team and AI tool reads from it.
4. **Keep it current (ongoing).** When something changes, the agent drafts the update.

### Which documents do the agents build and keep current?

#### Research & Intelligence

- **Competitive Intelligence Agent:** Dossiers and battle cards, rewritten when a competitor changes its site, docs or pricing
- **Market Research Agent:** Market shifts and analyst reports, turned into opportunities
- **Voice of Customer Agent:** What buyers say about you and your rivals, in their words
- **Win/Loss Agent:** Why deals are won or lost, from a survey or AI interview after every closed deal

#### Target Audience

- **ICP Agent:** The traits that predict a win, and every open deal checked for fit
- **Personas Agent:** Personas built from real buyers, to test a message on

#### Positioning & Messaging

- **Positioning Agent:** Positioning mapped against competitors, differentiators scored by win rate
- **Messaging Agent:** Messaging for every persona and stage, with drift on your site and decks caught
- **Product Intelligence Agent:** A product profile kept current from your site, docs and releases

## Step 2: What can product marketing teams use Calven for?

Connect Calven MCP to Claude, ChatGPT or any MCP client. Every prompt starts from your published Universe, not a pasted doc.

### Launch messaging

Positioning, messaging and the launch kit for a new product or feature

Ask it:
- Build the demand case for [feature]: the quotes, lost deals and competitor moves behind it.
- Write the launch messaging for [feature] for [personas] in [segment], from this demand case: [paste it]
- Fact-check every claim in this launch messaging against the product brief: [paste it]
- Review this launch messaging as [personas] and tell me what they would not believe: [paste it]
- Build the launch kit for [feature]: one-pager, FAQ, talk track, launch email and page copy.
- Write the field brief for [feature]: what to say, what not to claim, and the answer to [competitors].
- How did [feature] land since [launch date]? Calls, deals and competitor reactions.
- Pick the launch tier for [feature] with an expected-value tree built on our own launch history.
- War-game how [competitor] answers our [feature] launch and pre-write the counter.
- Rank these launch claims with a MaxDiff panel of [personas] and tell me which to lead with: [paste the claims]

Advanced prompts: Pick the launch tier with an expected-value tree, War-game the competitor's answer to the launch, Rank the launch claims with a MaxDiff panel. Workflow prompts: Gather the evidence buyers asked for this, Draft the launch positioning and messaging, Fact-check every launch claim, See how each persona reads the messaging, Build the launch kit, Brief the field before launch, Read how the launch landed. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/launch-messaging.md

### Win/loss readouts

The monthly or quarterly readout on why deals are won and lost

Ask it:
- Build the headline win/loss read for [last quarter]: win rate, top reasons, what moved.
- Break down win/loss by competitor for [last quarter], with the buyers' words.
- Rank the product gaps that cost us deals in [last quarter], by pipeline lost.
- Score how we sold in [last quarter]: what buyers said about the sales experience.
- Assemble the [last quarter] win/loss readout from these sections: [paste them]
- Test whether our win-rate change in [last quarter] is real or noise, before leadership acts on it.
- Test which of our beliefs about what wins deals actually predicts a win: [paste the hypotheses]
- Replay last quarter's lost deals with [the missing integration] fixed and count how many flip.
- What is our competitive win rate for [window], and how did it move?
- Why did we lose to [competitor] this quarter? Quote the buyers.

Advanced prompts: Test whether the win-rate move is real, Find what predicts a win in closed deals, Replay the lost deals with one gap fixed. Workflow prompts: Build the headline win/loss read, Read win/loss by competitor, Rank the product gaps that cost deals, Score how we sold, Assemble the full readout. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/win-loss-readouts.md

### Competitive teardowns

A full read on one competitor, from dossier to the deals it touches

Ask it:
- Which competitor deserves the next teardown, based on [last quarter]?
- Build the full teardown of [competitor] for [last quarter].
- Compare our product and pricing with [competitor] side by side.
- Check the [competitor] battlecard against this teardown: [paste the teardown]
- Draft the board's competitive slide on [competitor] from the teardown.
- Model our next move against [competitor] as a payoff matrix and find the option that holds up whatever they do.
- Write the battlecard [competitor]'s PMM uses against us, then show me where it beats ours.
- Map every segment on a fight-or-walk 2x2 against [competitor] and give sales rules of engagement.
- What did [competitor] do in the last 90 days?
- What is our win rate against [competitor], and how did it move?

Advanced prompts: Build a payoff matrix for their next move, Red-team us as their PMM, Map where to fight them and where to walk. Workflow prompts: Pick the competitor for the next teardown, Build the full teardown, Compare product and pricing side by side, Check the battlecard against the teardown, Draft the board's competitive slide. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/competitive-teardowns.md

### Messaging audits

Check pages, decks and campaigns against the approved story, in the buyer's words

Ask it:
- Check this [page] for drift from the approved messaging for [persona]: [paste it]
- Does this asset use the words our customers use? Rewrite the lines that don't: [paste it]
- Fact-check this asset and read it as our personas: [paste it]
- Do reps pitch the approved story on calls? Compare [last quarter]'s calls with the pillars.
- Find the gaps in our messaging framework: personas, stages and objections it doesn't cover.
- Assemble the messaging audit from these check results: [paste them]
- Size an A/B test of the new headline against the current one on [page]: how long, and is it worth running?
- Build an eval set from our messaging that grades any future [homepage section] before it ships.
- Run a recall test: have [personas] read this page and tell me what they remember: [paste it]
- Is this headline on-message for [persona]: "[headline]"?

Advanced prompts: Size an A/B test for the new message, Build an eval set that grades future copy, Run a recall test on synthetic buyers. Workflow prompts: Catch drift from the approved messaging, Check the asset speaks like customers, Check the facts and the persona read, See whether the field pitches the story, Find the gaps in the messaging framework, Assemble the messaging audit. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/messaging-audits.md

### Claim checks

Fact-check every product, pricing and competitive claim before it ships

Ask it:
- Fact-check every product and pricing claim in this asset against the brief: [paste it]
- Check every claim this asset makes about [competitor] against the dossier: [paste it]
- Check that every quote and outcome claim in this asset has a source: [paste it]
- List the claims on [our pricing page] that have no proof behind them.
- Cross-examine the claims in this page as opposing counsel would: [paste it]
- Score every claim on [our website] in a risk register: evidence, staleness, exposure and reach.
- Grade the evidence behind these outcome claims, strongest to weakest: [paste the claims]
- Is "[claim]" in our product brief?
- Do we integrate with [tool]?
- What does the [tier] plan include, and what does it cost?

Advanced prompts: Cross-examine the claims as opposing counsel, Score every claim in a risk register, Grade the evidence behind outcome claims. Workflow prompts: Fact-check product and pricing claims, Check the competitive claims, Check the proof and outcome claims, List the unsupported claims on a site section. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/claim-checks.md

### Customer evidence packs

Quotes, proof points and stories grouped by theme, for any asset

Ask it:
- Assemble an evidence pack on [theme] for [persona]: quotes, proof points and stories.
- Pull the proof from deals we won against [competitor]: who, why, in their words.
- Build a language bank of the words customers use for [problem].
- Build objection handling for [persona] from what customers actually said.
- Map our proof against what costs us deals, and show where we have nothing to say.
- Rank these quotes on [theme] in a credibility tournament judged by [personas].
- Audit this evidence pack for selection bias: whose voice is over- or under-heard? [paste it]
- What are the top three pains customers named this quarter, with quotes?
- Give me a quantified-outcome quote about [topic].
- What did [account] say about [feature]?

Advanced prompts: Map proof coverage against what costs deals, Rank quotes in a credibility tournament, Audit the pack for selection bias. Workflow prompts: Assemble an evidence pack on a theme, Pull the proof from won deals, Build a customer language bank, Build objection handling from customer quotes. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/customer-evidence-packs.md

### Persona briefs

A one-page brief on a persona for anyone writing, selling or building for them

Ask it:
- Write a one-page brief on [persona] for [a new SDR] who has to sell to them.
- How much does [persona] being on a deal change our win rate and deal size?
- Find what we don't know about [persona] and where the evidence is thin.
- Check this campaign audience against our personas and ICP: [paste it]
- Map [persona]'s push, pull, anxiety and habit forces from customer quotes.
- Build a [persona] simulator in [segment] that new reps can practice discovery against.
- Test [persona] against who actually signed our deals in [the last four quarters].
- What is [persona] measured on?
- What are [persona]'s top pains, with quotes?
- What objections does [persona] raise, and what is our response?

Advanced prompts: Map the persona's switching forces, Train reps against a persona simulator, Test the persona against who actually signs. Workflow prompts: Write a persona brief for one reader, Measure the persona's effect on deals, Find what we don't know about the persona, Check a campaign audience against personas. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/persona-briefs.md

### Product collateral

One-pagers, solution briefs, feature-to-benefit copy from the product brief

Ask it:
- Draft a one-pager on [capability] for [persona] in [segment].
- Turn [feature] into value for [persona], in their words, with proof.
- Write the why-not-the-alternatives section of [asset] against [competitor].
- Review this one-pager for stale claims and off-message lines: [paste it]
- Write a demo storyline for [persona] in [segment], built on what wins.
- Build an ROI calculator for [capability] from outcomes our customers reported.
- Have feature-led and outcome-led versions of [asset] debate, and let [persona] judge.
- Sort these features with a Kano model for [segment] and tell me what to lead with: [paste the list]
- What does [feature] do, as the brief states it?
- Which integrations do we list for [category]?

Advanced prompts: Build an ROI calculator from customer outcomes, Settle feature-led versus outcome-led in a debate, Sort the features with a Kano model. Workflow prompts: Draft a one-pager or solution brief, Turn a feature into buyer value, Write the why-not-the-alternatives section, Review an existing asset, Write a demo storyline. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/product-collateral.md

### Battlecard distribution

Turn the battlecard into the formats each team uses

Ask it:
- Turn the [competitor] battlecard into a one-page rep cheat sheet.
- Pull the proof from customers who switched from [competitor] for the sales deck.
- Write this week's competitive update for the field.
- Brief a CSM on a renewal where [competitor] is in the account.
- Turn the [competitor] battlecard into spaced-repetition flashcards for reps.
- Backtest the [competitor] battlecard against [the last four quarters] of deals.
- What is the elevator pitch against [competitor]?
- What landmines do we set against [competitor]?
- When was the [competitor] battlecard last updated?
- What did [competitor] change since the battlecard was updated?

Advanced prompts: Turn the battlecard into spaced-repetition flashcards, Backtest the battlecard against real deals. Workflow prompts: Turn the battlecard into a rep cheat sheet, Pull the displacement proof for the deck, Write this week's competitive update, Brief a CSM on a competitive renewal. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/battlecard-distribution.md

### Positioning variants

Positioning adapted for a vertical, segment or launch

Ask it:
- Create a positioning variant for [vertical] from the core positioning.
- Stress-test this positioning variant against the core: [paste it]
- What evidence do we have for our positioning in [segment]? Win rate, quotes, proof.
- Converge on the [vertical] positioning variant with a Delphi panel of our personas.
- Score which of these segments earns its own positioning variant: [paste the list]
- Is [vertical] a priority or secondary vertical in our ICP?
- What do customers in [industry] say about [pain]?
- What is our win rate in [segment], with n?
- Which competitor do we meet most in [vertical]?
- Which trends does the market research agent track for [vertical]?

Advanced prompts: Converge on the variant with a Delphi panel, Score which segments earn their own variant. Workflow prompts: Create a positioning variant, Stress-test a variant against the core, Check the evidence for this context. Full prompts: https://github.com/calven-ai/calven-mcp/blob/main/use-cases/product-marketing/positioning-variants.md

10 use cases, full prompts: https://github.com/calven-ai/calven-mcp/tree/main/use-cases/product-marketing

## What changes when product marketing runs on Calven?

Less upkeep, one approved story, and no more answering the same question twice.

| Pain point | Without Calven | With Calven |
| --- | --- | --- |
| Hours back | Much of the week goes to updating cards, decks and docs by hand | Agents draft the update when a source changes; you review and approve |
| Reaction time | A competitor's move reaches you weeks later, through a lost deal | You see what changed, with the source, and a new battle card draft follows |
| Evidence | Positioning calls defended with a few interviews a quarter | Win/loss patterns from your closed deals, by reason and competitor |
| Buyer fit | ICP and personas built in a workshop and refreshed when someone has time | ICP and personas rebuilt from the deals you won and the buyers on your calls |
| Consistency | Found out when a rep pitches last year's story | Drift caught on your site and decks, with the approved line next to it |
| Fewer interruptions | Every team pings you for the latest version | Teams ask Calven in Slack or their AI tool and get the version you approved |
| Visible impact | Leadership sees what you shipped, not what it moved, so budget talks run on opinion | Win rate by competitor and whether reps say your lines on calls, before and after you ship |

## FAQ

### What is the Universe in Calven?

The published set of documents your agents build and keep current, each with its sources. Your product marketing team publishes it, and every team and AI answer reads from it.

### Will Calven replace product marketers?

No. Agents draft, and by default nothing ships until a product marketer approves it. Positioning, launches and what the field hears stay your team's call.

### What do our PMMs spend their time on once the agents run?

Deciding the positioning, running launches, coaching the field and making the case to leadership. They review drafts against sources instead of starting from a blank page.

### We already have a competitive intelligence tool. Why Calven?

Competitive intel is one of nine agents. Calven connects it to win/loss, calls, the ICP, positioning and messaging, so a competitor move reaches the battle card and the field from one source.

### We already run product marketing prompts in Claude or ChatGPT. What does Calven add?

Five things a prompt can't do on its own. Calven collects new evidence: the Win/Loss agent invites the buyer after every closed deal and runs the survey or AI voice interview itself, and research reaches the review sites a chat assistant can't. The agents work to proven frameworks for ICP, personas, positioning, messaging and battle cards, and flag the weak spots instead of writing whatever you ask for. Nine agents work as one team: when a competitor changes its site, its dossier updates, and when the positioning changes, the messaging is redrafted for your review. There's no code for your team to maintain when models change, connectors break or prompts drift. And over Calven MCP, every AI tool in the company reads the ICP, positioning and messaging your team approved, with sources.

### Can we control what the agents change?

Yes. By default every change arrives as a draft its owner accepts or leaves. You can let routine updates publish on their own; strategy documents still come back as drafts.

### How long until Calven is useful?

Day one: from your company URL, the agents deliver the first competitor dossiers. Connect your CRM and call recorder for an ICP built from real deals. Within a week you publish battle cards and messaging.

See also: [All teams](https://calven.ai/solutions.md) · [Calven MCP](https://calven.ai/mcp.md) · [Platform](https://calven.ai/platform.md)