Endgame · agentic context layer

Endgame makes Data 360 data agent-ready.

Data 360 centralizes the records. Endgame turns them into cited facts, linked to the accounts, people, and deals they concern — ready for Agentforce Coworker, Claude, or any agent on top.

9 / 9correct, cited, one turn — Endgame, every tier, first three questions
1 / 9clean answers from Data 360 alone
17in verification — results land here when sealed
Inside what Endgame is · how it works · twenty questions of evidenceDataset Northwind (synthetic)Status draft

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

01 / 09

01 TL;DR

What the bakeoff showed.

Endgame 9 / 9 correct, cited, one turn — every question, every model tier
Data 360 1 / 9 one clean answer in nine — on the same records, the same questions

“Data 360 successfully centralized the evidence. It did not turn that evidence into sales context.”

— Codex, an OpenAI model, reviewing the results

The hard part isn’t storing the records — it’s pulling the facts out of calls, emails and Slack and linking them to the accounts, people, and deals they concern. That work is a layer of its own. Data 360 centralizes the records. Endgame turns those records into agent context. Every question tested exactly that seam.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

02 / 09

02 The stack

An agent is only as good as its context.

Agentsconsume context Agentforce Coworker · Claude · any MCP client
data layer alone
1 / 9
no context layer
through the context layer
9 / 9
Endgame agentic context layer cited facts, linked to the accounts, people, and deals they concern cited, linked facts context graph · kept current
Data 360data layer The records, centralized — CRM · email · calls · Slack
Sourceswhere records originate Salesforce · email · calendar · calls · Slack

bakeoff agent: Claude Desktop · same records · same questions

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

03 / 09

03 What Endgame is

One living model of the sales world.

Endgame builds and maintains a context graph — a single model, per tenant, of every account, person, deal, and signal across the stack — that people and agents read, and teach.

Customer dataaccounts, people, deals, calls, email, Slack
Playbookshow the team sells
Semanticsthe team’s terms and metrics
Team & ownershipwho owns what, whose pipeline rolls up
Cited
Every fact carries its evidence — a claim in an answer traces to the exact email, call, or message it came from.
Consistent
Structure reads straight from one resolved core — the same answer for everyone, every session.
Teachable
“John no longer works at the company” — typed to Endgame mid-deal — becomes a proposed correction, confirmed, applied, reversible. Human intent outranks rules and models.

In the evaluation org, the graph resolved to 2,575 entities connected by 4,153 relationships, carrying 17,822 cited facts — built from the same records Data 360 saw.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

04 / 09

04 How it works

Four stages, all before the question.

Unify the record

Connectors bring CRM objects and the engagement record — calls, emails, Slack — into one store. In the evaluation dataset: 308 accounts, 1,395 contacts, 535 opportunities, 2,399 engagements, identical on both sides.

Resolve it into a graph

Endgame resolves every engagement to the accounts, people, and opportunities it involves — including engagements that arrive with no CRM identifiers, the way real calls and emails do. The evaluation org resolved to 2,575 entities connected by 4,153 relationships.

Extract facts, each with a citation

An extraction pass reads every transcript, email, and thread and produces discrete facts — 17,822 in the evaluation org, each anchored to the exact source record it came from, and every one carrying at least one link into the graph. A fact in an answer traces to a specific email, call, or message.

Maintain perspectives with context agents

Scheduled agents read the graph and keep account perspectives current — a headline briefing and the account’s active work streams, refreshed every half hour across all 215 accounts with open pipeline. The synthesis runs once, on a schedule, from the extracted facts — a specific claim checks against the facts behind it, and every user who asks gets the same answer.

At question time there is no archaeology left to do. When an agent calls an Endgame tool, Endgame retrieves across the graph and its semantic index, ranks against the ask, right-sizes to a token budget, and enforces the calling user’s permissions — the agent gets back pre-ranked, right-sized results. In the evaluation org, the production API returned 90 ranked facts for one account in 218 ms.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

05 / 09

05 How we tested

One dataset, both systems, twenty questions.

DatasetNorthwind — a synthetic B2B software vendor selling into the enterprise.
LabelsReal company names — Databricks, Eli Lilly, … — on fully synthetic records. Reads like real deal work.
ParityBoth services see exactly the same data. Apples-to-apples.
308
accounts, with hierarchy & open pipeline
1,395
contacts · 1,050 opportunity roles
535
opportunities, open and closed
2,399
engagements — 293 calls, 1,464 emails, 642 Slack

Two question sets, two objections

Q1–Q3 · the easy set

Three factual questions any system connected to the records should answer.

They settle the first objection: “why not just connect the agent to Data 360?” Results are final — the scorecard and the full runs are in this deck.
Q4–Q20 · the representative set

Seventeen questions at the working difficulty of real deal work.

They settle the second objection: “easy questions can’t show what the context layer adds.” In verification now; questions and per-tier results land here when sealed.

Models — the easy set ran on all three tiers

Strongest

Fable 5

Reasoning effort: High
Everyday

Sonnet 5

Reasoning effort: Medium
Fast & cheap

Haiku 4.5

Extended thinking: off

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

06 / 09

06 The scorecard

Twenty questions, one grid.

The easy set is settled: Endgame answers correctly at every tier; Data 360 lands one clean answer in nine. The representative set lands on this grid when sealed.

Data 360 Endgame
Fable 5F5Sonnet 5S5Haiku 4.5H4.5 Fable 5F5Sonnet 5S5Haiku 4.5H4.5
Renewalthe Databricks renewal date
Last contactwhen Eli Lilly last reached out
Economic buyerwho owns the expansion budget
Q4–Q20the representative set — seventeen questions
in verification — results land here when sealed

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

07 / 09

07 The easy set

Each question turned on which layer did the linking.

Every question had a correct answer in the data and a wrong answer sitting in the obvious field. An agent on the data layer read the field; an agent on the context layer got the linked fact.

Q1 · renewal date

The CRM close date said Sep 30. The customer’s June 10 email corrected it to Aug 31.

Endgame had extracted the correction as a fact linked to the account — every model tier answered Aug 31 in one turn and flagged the stale field.

Q2 · last contact

The activity rollup read June 2026. The last real inbound was an email of April 3, 2025 — a year of silence.

Endgame’s graph links each engagement to the account with its date and direction — the silence was directly visible.

Q3 · economic buyer

The contact roles omitted the buyer entirely. On a May 28 call she said: “I own the platform engineering budget.”

Endgame had extracted that statement as a cited fact — every tier named her and flagged the missing role.

The full runs — every Data 360 screenshot, every Endgame citation, and the two extra steps Data 360 needed — are in the appendix.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

08 / 09

08 The representative set

Harder questions, same grid.

The easy set shows a data layer alone missing questions the records can answer. It leaves a fair objection standing: easy questions can’t show what the context layer adds. The representative set exists to answer it — seventeen questions at the working difficulty of real deal work, run and scored the same way.

In verification

The questions and their per-tier results are being verified and sealed in a parallel workstream. When they land, this section carries the seventeen scorecard rows and worked runs — with screenshots and citations — for the questions that best show the seam.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

09 / 09

09 The bottom line

Alone, the data layer answered one in nine. The same records, through the context layer, answered nine.

Data 360 Endgame
Centralized the evidencePulled the records — emails, calls, Slack, CRM — into one place.
Turned it into sales contextLinked across the sources to the one resolved answer the question was actually asking for.

Both systems saw exactly the same data. The two rows are different jobs — and the stack that answers is the one that staffs both: the data layer centralizing the records, the context layer turning them into cited, linked, ready-before-the-question context, and every agent on top — Agentforce Coworker, Claude, anything — inheriting the answers.

For this test we loaded the records into Data 360 directly. In a live Salesforce deployment only CRM is centralized by default — email, calls, and Slack each require separate connectors, and email in particular stays largely outside queryable storage (Einstein Activity Capture writes it to AWS, with native EmailMessage sync only since Summer ’25 and historical backfill capped at 180 days).

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

A1 / A4

A1 Appendix · the easy set in full · Q1 renewal

“When is the Databricks platform subscription up for renewal?”

The right answer is Aug 31, 2026 — the customer corrects the date in a June 10 email. The obvious field — the CRM close date — still says Sep 30. The truth lives in the email, not the field.

Data 360Fable 5amber
Aug 31 correct + cited · via a 5-min research run

Reached “August 31, 2026 as the operative renewal deadline” only by auto-escalating into a five-minute, 31-source research run. No answer arrived in a single turn.

Data 360Sonnet 5red
Sep 30 wrong · stale CRM date

Reports “2026-09-30” — the Salesforce close date — as “the closest thing to a formal ‘renewal date.’” It never found Amara Mensah’s email correcting the date to Aug 31.

Data 360Haiku 4.5red
No answer couldn’t run the query

“I don’t have direct access to Databricks account information.” No query is run — it hands back three suggestions and asks which to try. No date.

Endgameevery modelgreen · every tier
Aug 31, 2026 correct + cited · single turn · corrects the CRM

All three tiers answer “August 31, 2026” in one turn, cite Amara Mensah’s June 10, 2026 email, and flag that the Salesforce close date of 2026-09-30 is the record she was correcting.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

A2 / A4

A2 Appendix · the easy set in full · Q2 last contact

“When did we last hear from Eli Lilly?”

The right answer is April 3, 2025 — silent since. The obvious field — the LastActivityDate rollup — reads June 2026, a year off; read alone, it calls a dead account active.

Data 360Fable 5red
June 2026 wrong · a year off

“Last touch: June 19, 2026 — that’s the Last Activity Date on both the account record and the one open opportunity.” In-flow, and a year off: the rollup field, read as the answer.

Data 360Sonnet 5red
June 2026 wrong · contradicts its own table

States “the most recent touch on record is June 19, 2026” off the EngagementAggregation rollup — while its own closed-won table, in the same reply, shows the real Apr 3, 2025 row.

Data 360Haiku 4.5red
No answer punted to the user

“No matching tools found,” three times over — and this run used the spelled-out phrasing. It offers manual Gmail / Slack / Calendar checks and asks which to try. No date.

Endgameevery modelgreen · every tier
April 3, 2025 correct + cited · single turn · on the plain question

Fable dates the last real inbound to April 3, 2025 — Ibrahim Sato’s email — and confirms the silence since; Haiku puts the last substantive exchange at April 2025 and recommends re-engaging. Sonnet returned the same April 3, 2025 answer (Fable and Haiku frames shown).

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

A3 / A4

A3 Appendix · the easy set in full · Q3 economic buyer

“Who is the economic buyer for the Databricks expansion?”

The right answer is Priyanka Sharma — she says on a May 28 call that she owns the platform-engineering budget. The obvious field — the opportunity contact roles — doesn’t list her at all. Fatima Khan is a FinOps gatekeeper, not the buyer.

Data 360Fable 5amber
Priyanka? correct, but wouldn’t commit

Names Priyanka as the “presumed economic buyer to validate, not a confirmed fact” — right name, correctly flagged as missing from the roles, but recommended only “once validated.” A hedge, not an answer.

Data 360Sonnet 5green
Priyanka correct + cited · the one clean win

The honest result: Data 360 is not zero-for-everything. It prints the committee table, then catches the gap — “your buying committee is missing its actual economic buyer.” Shown with full weight.

Data 360Haiku 4.5red
Fatima wrong person

“Update the Opportunity Contact Role for Fatima Khan to include ‘Economic Buyer.’” The FinOps gatekeeper, confidently recommended as a CRM edit; Priyanka never surfaces.

Endgameevery modelgreen · every tier
Priyanka Sharma correct + cited · flags the missing contact role

Every tier names Priyanka Sharma, quotes her own May 28 call — “I own the platform engineering budget” — flags the missing contact role, and rules out Fatima Khan as a FinOps gatekeeper. No hedge, no wrong person.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

A4 / A4

A4 Appendix · what it took to get Data 360 to answer

Two answers needed extra steps.

Two of Data 360’s better results took extra steps beyond asking the question. Those steps, documented.

Q1 · renewala five-minute research run instead of an answer

On the plain question, Data 360 on Fable 5 returned no answer in a single turn. It auto-escalated into a research run — 31 sources, 5 minutes 24 seconds — and only that run’s report names August 31, 2026 as the operative renewal deadline.

Q2 · last contacta re-worded question that spells out the channels

On the plain question, every Data 360 run read the LastActivityDate rollup as the answer — June 2026, a year off — with no signal the answer was wrong. Getting the right date took re-wording the question to spell out the channels — “who at Eli Lilly last contacted us directly: an email, a call, or a meeting?” Sonnet 5 then lands it: “the April 3 email … from Ibrahim Sato is the last time someone at Eli Lilly directly reached out.” Fable 5’s re-worded run reaches the right region but calls the gap “over three months” — for a silence of more than a year.

© 2026 Endgame Labs, Inc. · Confidential & Proprietary

Endgame · the Data 360 packet · © 2026 Endgame Labs, Inc. · Confidential & Proprietary