FY27 GTM Engineering
We're giving teams more AI tools, agents, and automation, so they can focus on the work that actually moves the business. The goal is to simplify and focus: fewer manual steps, fewer context switches, fewer things falling through the cracks.
Q2 Close · Jul 2026 · Cross-Functional Leadership
Q3 · Aug to Oct 2026
Four objectives, thirteen key results. Anchored to Simply Sendoso: Easier to Buy, Easier to Use, Easier to See Value, Easier to Operate. Every key result names a deliverable.
One place where Sales, BDR, and CS see everything about an account and what to do next.
How fast and how safely we produce and run agents. The leading indicators of next quarter's output.
What shipped, how much the agents already live improved, and how much of the fleet other teams run themselves.
From ticket-driven manual work to agent-run execution, plus visibility into total customer touch volume.
The Agent Roster
Ten agents and two apps live in production, plus the Q3 build queue.
Sprint Delivery Roadmap
Agents by launch month, with playbook and dashboard workstreams running across the quarter.
Q2 to Q3
What we shipped May to July, and what Q3 unlocks.
GTM Engineering: Building the "more" in "do more with more"
v2.4.0 · Aug 2026 · Austin · GTM Engineering
The Foundation
Before an agent can ship in hours, the platform underneath it has to exist. The groundwork came first, and it's what now lets every new agent stand up fast, governed, and accurate by default.
Medallion architecture · dev workspace · in progress
Q1 · Feb to Apr 2026
Two pulled to future state. Everything else shipped or rolled into Q2.
Proved Fivetran → Snowflake → Databricks architecture. New agents stand up in days, not weeks.
Account + Contact scoring deployed to AE team. Performance validated. Lead routing live April 3.
Visual lifecycle mapping with agent-consumable API. React Flow + Cloudflare. Auto-deployed.
Product data schema ready and updated in Databricks. AI and agent consumption ready.
Account, Contact, Opportunity objects documented at field level. AI/Databricks consumption ready.
End-to-end pipeline from agent idea intake through scoring, review, and sprint execution. Operational and repeatable.
Also shipped: Ticket Estimation System · Marketo Health Audit Workbook · Documentation Framework · Outbound Orchestration Consolidation · Sales Proposal Deck Automation
Q2 · May to Jul 2026
Four objectives, sixteen key results. Anchored to Simply Sendoso: Easier to See Value, Easier to Buy, Easier to Operate. Status as of Jul 26. 11 of 16 delivered.
Replace manual CX work with agents and automated value motions, measured by retention and expansion impact.
Better conversion, higher-quality pipeline. Supports $2.5M Q4 new ARR.
From reactive ticket-runners to autonomous execution.
Build the engine that ships AI safely at scale.
The Vocabulary
Two different things. Decoupling them matters.
Like Salesforce or Gong. You log in, you prompt it, you get work done. Connects to your existing systems.
Nobody's scared of a tool.
Autonomous workflows in the background. Each maps to a job function. Agents have skills: discrete capabilities.
Agent Program
Every agent idea goes through a weighted scoring rubric, not gut feel.
P0 (4.0+): Build now P1 (3 to 3.99): Build soon
P2 (2 to 2.99): Backlog P3 (<2): Parking lot
Archive
Context and framework slides moved out of the main Q3 flow, plus concepts from earlier versions of this deck. Still relevant, just not in the narrative.
Claude is a tool people use. Agents are programs we build. Why decoupling the two matters.
The weighted scoring rubric and the intake-to-production pipeline behind every agent we build.
The framework, security, data layer, and structure that let a new agent stand up in hours.
Before/after/future stories for starter provisioning and Webflow self-service. The original "imagine a world where" framing.
Databricks One → Claude pockets → First Agents. Sequencing logic for the original rollout.
Three-tier autonomy framework: automated day one (internal), human-in-the-loop, graduated autonomous. Still the model we use.
All prior versions accessible via the Versions button (bottom-right).