Sagetap Campaign Compendium — The Wordsss Team

Every campaign Empathic has run (or staged) on Sagetap, in one place: the exact words in market, who they were aimed at, and what came back. Prepared July 21, 2026 for the brand/storytelling engagement. Verbatim buyer language lives in the companion notes (raw captures + applicant tracker); the highlights are quoted here.

How Sagetap works, in one paragraph: vendors publish campaigns (a title, a product description, and three qualifying questions). Enterprise buyers (“Sages”) apply by answering the questions; applications are free to receive, calls cost credits. The questions are therefore a survey instrument as much as a filter — every application banks verbatim buyer language whether or not a call happens.

The portfolio in one picture

One substrate — the AI agent session record — sold at three altitudes:

Altitude	Campaign	Register	Motion
Engineer (does the work)	Engineers	”Your reasoning disappears when you merge”	Bottom-up, free install
Engineering leader (owns the budget)	AI Spend & Value (staged)	“You can see what AI costs. Can you see what it built?”	Top-down, design partner
CTO / platform owner (owns the record)	CTOs / Institutional Memory	”Agent sessions are your company’s institutional memory”	Aggregate data + rare exceptional calls

The engineer campaign generates the traces; the leadership campaigns convert the value of those traces into pipeline. The earlier Engineering Leaders campaign (May) was the first probe and produced the two-archetype map that everything since is built on.

Campaign 1 — Engineering Leaders (May 2026; first probe, now superseded)

Title:

Engineering Leaders: See What Your AI Coding Agents Are Actually Doing

Product description (in market):

Engineering teams are shipping more code with AI agents but have zero visibility into what those agents actually did. Pathbase captures, structures, and shares agent session traces across Claude Code, Codex, and Gemini — so leaders can finally see what’s really happening.

Persona: Director/VP/C-suite engineering leaders at 100–5,000-person US companies, 20+ engineers on agentic tools, feeling the gap between “everyone seems more productive” and “we can’t measure it.”

Qualifying questions:

How does your engineering team currently use AI coding agents (e.g., Claude Code, Copilot, Cursor, Codex)? How widely adopted are they across the team, and how has that changed in the last 6 months?
What visibility do you have today into how your team is actually using AI agents? If your leadership asked “what’s the ROI on our AI tooling investment?” — what evidence could you point to right now?
How has the shift to AI-assisted development changed your team’s code review process, engineering quality practices, or how you coach and develop engineers? What’s working and what’s broken?

What it taught:

The two-archetype map. Applicants split into Archetype 1 (engineering effectiveness: “is AI making my team better?”) and Archetype 2 (governance/security: “is AI introducing risk?”). The title’s “see what your agents are doing” reads as an audit promise to security buyers — the campaign attracted more CISOs and IT directors than engineering leaders.
Star signals: JosephB (VP Product, Gen AI, insurance F500): “Macro numbers clearly indicate a positive ROI, but we have no clear visibility at the micro (individual agent / developer) level.” Joseph_1446: “AI can mask knowledge gaps while still producing functional code.” Kevin_5542 on AI PR review tools: “none have ‘solved’ the problem for us yet.”
Five calls completed (Brett, Kevin, Joseph_1446, JosephB, Nicholas origin); the JosephB pass (“feels like it just delivers information… without specific insights, suggested action items, or downstream process”) became the most consequential piece of feedback in the company’s positioning history — it drove the shift from information-led to decision-led framing.

Campaign 1b — Effectiveness revision (designed May 28; retired unlaunched)

A data-driven revision of Campaign 1 designed to repel the security archetype:

Proposed title:

Engineering Leaders: Is Your AI Investment Making Your Team Better — or Just Faster?

Why it was designed this way: “Better vs. faster” is an engineering-leadership frame; a security buyer frames concern as “safe vs. unsafe,” so the title filters by vocabulary. It was retired before launch when the portfolio consolidated: the hands-on-leader job moved to the Engineer campaign, and the leadership job was rebuilt around spend and value (Campaign 4).

Campaign 2 — Engineers (June 4, 2026; active — the flagship)

Title:

Engineers: Your Agent Did the Work — But the Reasoning Disappears When You Merge

Product description (in market):

Your AI agent session disappears when the PR merges. Pathbase captures the full conversation — every prompt, every dead end, every tool call — and packs it into a link you can share, resume, or hand off. Works with Claude Code, Codex, and Gemini. Try it at pathbase.dev.

Product summary (long form):

When engineers build with AI agents, the session that produced the code disappears when the PR merges. Git keeps the diff; the prompts, the approaches the agent tried and rejected, and the reasoning behind every decision are gone. On a team, that gap costs real time three ways: reviewers ask “why did you do it this way?” and the answer is buried in a conversation they can’t see, so reviews stretch from hours to days; work-in-progress can’t be handed off without re-explaining everything to a fresh agent session; and resuming your own work on another machine means starting the agent cold. Pathbase captures the full session and packs it into a link you can share, resume, or hand off — drop it in the PR and the reviewer asks the agent directly instead of interrupting the author.

Persona: Senior ICs, staff/principal engineers, tech leads, hands-on engineering managers on teams of 10+ where PRs are reviewed before merge.

Qualifying questions:

Which AI coding agents do you use day-to-day — and when you open a PR that one of them mostly built, what does the reviewer actually see? Can they tell what the agent tried and rejected, or just the final diff?
Have you ever needed to pick up an agent session on a different machine, or hand work-in-progress to a teammate who then had to re-explain everything to a new agent? How did you handle it?
Think about the PRs you ship or review: how many of the review comments are questions about why rather than what?

What it taught:

The why-comment number is real and replicable. Independent applicants at different companies: 70%+ (DJC), 60–70% (Wizard), 50% (Tad), 30–40% (Pavel, Femo), 30% (pumiki). The review-breakdown thesis has a measurable signature.
The manual workarounds ARE the product spec. “Pasting massive prompt histories into Slack or Jira tickets” (DJC); “we copy paste all the information into the session” (pumiki); “asking the agent to summarize everything into an md file and have another agent read this md file” (Tad).
Best outcomes to date: NicholasE (Engineering Manager, retail, 10,000+) — call scored 29/30, September deployment target, volunteered a design-partner model. pumiki (Sr. Eng Manager, finserv) — runs an autonomous Claude agent that pushes PRs with no human in the loop (“there’s no way to tell what the agent tried or rejected”); installing now. JosephB re-applied to this campaign after passing on the old product in May — the framing brought him back.

Campaign 3 — CTOs / Institutional Memory (June 12, 2026; active)

Title:

CTOs: Agent Sessions Are Becoming Your Company’s Institutional Memory — Are You Keeping Them?

Product description (in market):

Every agent session is a record of intent, alternatives, and decisions — and it disappears when the PR merges. Empathic captures that record across Claude Code, Codex, Gemini, and custom agents, and unifies it into one timeline of your engineering work: the data foundation of an AI-native company.

Purpose: deliberately a zero-marginal-cost experiment on the biggest register (“the data foundation AI-native companies are built on”). Default-decline on calls; the answers are the harvest. Accept only exceptional applicants.

Qualifying questions:

Roughly how many engineers at your company work with AI coding agents daily, and what happens to those sessions after the work ships? Are they retained anywhere today — vendor logs, internal storage, something you built yourselves — and if you ever needed to go back to one to understand why a change was made, could you?
Imagine a complete, searchable record of every agent session behind your codebase — the intent, the alternatives considered, the dead ends — joined to your PRs, deploys, and incidents. What is the first question you would ask it? And who or what else at your company would use that record: engineers onboarding, reviewers, auditors, or your own agents and models?
What would have to be true for agent session history to become a first-class system of record at your company? Specifically: where would the data need to live (your cloud, vendor-hosted, fully on-prem), what would need to be redacted or excluded before you’d allow capture by default, and is this something you would expect to build in-house or buy?

What it taught (July 2026 batch — the campaign’s thesis answered by its target buyer):

CloudDB (Head of Cloud Engineering, UK finserv, 10,000+; leads ~400 platform engineers): “~5000 engineers using agents daily. Session work, if stored by the harness, is local until deleted. Otherwise we typically lose this data.” On retention: “We store teams messages and emails today, this would fall into this category. Open to a product in this space if it fits our needs.” On review: “why does this line of code exist, what were the decisions, was a human involved.”
ggzuazo (Enterprise Architect, finserv, Brazil; “I’m the Final Decision Maker, holding the budget”): “Sessions are stored in Elastic for central storage… For audit reasons I need to search into my elastic repo… I would buy a solution like that. I have been using an elastic repository, but need something better.” Existence proof: an org already built the crude version and wants to buy the real one.
The requirements checklist wrote itself across three independent security-conscious respondents: your-cloud/VPC residency, pre-capture redaction of PII and secrets, data-classification mapping, hybrid/on-prem option.
Cost of the register: the institutional-memory framing also attracts pure security executives (the campaign’s decline-and-log gate handles them).

Campaign 4 — AI Spend & Value (staged July 1, 2026; launch-pending)

The leadership campaign rebuilt around the budget-owner’s job. Register: value first, cost as the denominator.

Title (recommended):

Engineering Leaders: You Can See What AI Costs. Can You See What It Built?

Alternates tested in design: “Heads of Engineering: When Finance Asks What the AI Budget Produced, What Can You Show Them?” · “Your AI Coding Spend Is a Real Budget Line Now. Can You Attribute It to the Work It Produced?” · “An Engineer Spent $20K This Month. Was That a Problem?”

Tagline:

Understand what your team builds with AI — starting with where the money goes.

Product description (staged):

You can see what AI costs, not what it built. Empathic captures the session record behind AI-assisted work — what the agent was asked, what it tried, where the tokens went — and we’re working with design partners to connect it across tools, PRs, and vendor spend so leaders can see what the budget produced.

Persona: Head/SVP of Engineering, VP/Head of AI, CIO, Head of Developer Experience at 200–10,000+ companies where agents are past pilot and the spend is a real budget line — the person who owes finance an answer.

Qualifying questions (staged):

How many developers in your org use AI coding agents day to day, and what’s the tool mix (Copilot, Claude Code, Cursor, Codex, custom agents)? Roughly what are you spending on AI inference per developer per month, and is that budget owned centrally or allocated to teams and business units?
When leadership or finance asks what your AI spend produced, what can you show them today, and what can’t you? Specifically: can you connect the spend to the work it shipped — which projects and PRs it produced — and can you tell a high-value session apart from thrash, retries, and abandoned attempts?
If you had the full record of every agent session tied to the work it produced — cost per merged PR, where tokens went to waste, which agent is best for which task — what’s the first decision you’d make with it? And who else in your org would need it: your finance partner, a VP, the CIO, team leads?

Why it exists: the value-gap answer kept arriving unprompted in other campaigns — “PR frequency metrics and lines of code, which we don’t trust, but neither tells the real story” (Tobie, applying to Campaign 1); “we calculated ROI by benchmarking time reduction” (RU); JosephB’s macro-vs-micro gap. The campaign gives that buyer their own front door. Its questions also build an aggregate dataset (spend-per-developer benchmark, budget structure, tool mix) that no one else has.

Results snapshot (as of July 21, 2026)

31 applications across the three live campaigns (14 leaders, 10 engineers, 7 CTO), 38 total connections.
5 calls completed; best outcome 29/30 (NicholasE — active deployment conversation). One pass (JosephB) converted into a re-application after the product went live.
Archetype mix: the recurring finding is that visibility language attracts security/governance buyers roughly 2:1 over engineering-effectiveness buyers. The strongest fits are found by initiative data (what the applicant is actually budgeted to buy), never by answer eloquence.
Spend signals collected: ~$50/mo to ~$500/wk per developer; adoption figures from 35 to ~5,000 engineers per org; multiple orgs at 50–100% agent adoption.

For the storytelling work: what the words did

The campaigns are a controlled experiment in which titles select audiences. Three findings with receipts:

“See what your agents are doing” reads as surveillance/audit — it pulled CISOs, IT directors, and compliance buyers who frame the need as oversight. The same product described as “your reasoning disappears when you merge” pulled hands-on engineers describing lived pain with numbers.
Buyers consistently reach for one noun we never used: “evidence.” “We lack structured evidence tying agent usage to measurable outcomes” (Joseph_1446); “a clean end-to-end evidence trail” (Wizard); “we need more evidence of reasoning, not only the code output” (Link); versus “anecdotal” (Weimin, TheCloudGuru). The word appears across all three buyer types, unprompted.
The deletion mechanism is described almost identically by everyone: “the session history… disappears from the engineering record” (Link); “the underlying conversation that produced the code disappears upon merging” (DJC); “nothing preventing deletion of work sessions, so there is not a written record” (Leopold_R); “the entire chain of reasoning is not in the PR logs” (pumiki). When a dozen buyers independently narrate the same mechanism, that’s the story’s spine.

Companion notes

sagetap-raw-captures — verbatim three-surface data per application
sagetap-applicant-tracker — the living roster, fit ratings, and chat activity
Campaign source notes: 2026-05-10-sagetap-onboarding · 2026-05-28-sagetap-campaign-revision · 2026-06-04-sagetap-engineer-campaign · 2026-06-12-sagetap-cto-platform-campaign · 2026-07-01-sagetap-ai-spend-value-campaign