Discover
We map where your hours actually go, rank the bottlenecks worth fixing, and define the success metric before anything ships.
The Agentic 100 · Benchmark v1
A public benchmark for agentic work that moves revenue, margin, customers, and execution. Each pattern is selected against one operating contract: value with evidence, bounded by reliability, adoption, and human control.

From work to a working system.
Bottleneck → outcomeBenchmark workflow patterns
Business workflows lead the benchmark: revenue, finance, operations, customer success, people, product, data, risk, and executive decisions. Personal patterns remain available as entry points, but operating leverage is the center of gravity.
74 of 100 workflows
Sales calls start with twenty minutes of discovery that the website, CRM, and form response already answered.
Strong opportunities stall for days while someone reconstructs discovery notes into scope, outcomes, and pricing language.
The forecast looks healthy until close dates move, stakeholders disappear, and next steps go stale all at once.
Account research is either skipped or consumes an hour of tab-opening before every important conversation.
The bank balance is visible, but upcoming payroll, collections, renewals, and tax obligations are not assembled into one decision view.
Collections happen from memory, so old small invoices consume attention while large recoverable balances age silently.
Leaders see actual versus budget but spend the review meeting discovering what moved instead of deciding what to do.
Purchase requests arrive as persuasive messages without total cost, alternatives, owner, security review, or success criteria.
Critical processes live in one person’s muscle memory and become stale the moment a handoff is attempted.
New vendors start work before security, privacy, insurance, payment, ownership, and renewal details are complete.
Operators scan multiple dashboards looking for the few orders, projects, or handoffs that actually need intervention.
The weekly review is assembled manually from status messages, spreadsheets, tickets, and whatever leaders remember to mention.
Health scores are broad averages that turn red after the customer has already disengaged.
Quarterly reviews become slide-production exercises full of activity counts and screenshots instead of customer value.
At-risk accounts trigger generic “check in” tasks without a shared view of root causes, commitments, or decision makers.
Customer feedback is abundant but fragmented across calls, tickets, chats, and account-team memory.
Hiring starts from recycled job descriptions, so interviewers optimize for credentials instead of the outcomes the role must deliver.
Interview feedback arrives late, uses different standards, and converges around the loudest person in the room.
Onboarding is a folder of links and meetings, leaving new hires to guess what good looks like and who can unblock them.
Performance reviews overweight the last few weeks, manager memory, and polished self-advocacy.
Important decisions arrive as meetings and slide decks with unclear options, assumptions, owners, and reversal costs.
Leadership sees charts but spends the meeting debating definitions, causes, and whether a movement is even material.
Board preparation becomes a last-minute hunt across department decks, finance files, and inconsistent narratives.
Competitive intelligence is episodic, anecdotal, and quickly polluted by sales hearsay or stale web pages.
Campaign briefs begin with internal opinions instead of current customer language and performance evidence.
Strong webinars and interviews disappear after one use while the content team starts from a blank page.
Teams react to top-line traffic while high-intent page and form failures remain hidden.
Spend, acquisition cost, or tracking can move for days before a human notices.
Evergreen pages lose rankings and conversions gradually, making decay hard to distinguish from noise.
Every event contact gets the same follow-up regardless of conversation, fit, or intent.
Claims drift across web pages, decks, and campaigns faster than legal and evidence libraries are updated.
Partner launches stall while both sides negotiate audience, claims, responsibilities, and measurement in threads.
Creative tests produce winners and losers but the reasoning disappears into campaign platforms.
Teams share scattered competitor, customer, and category news without deciding what matters.
Loud requests outrank repeated customer problems because evidence is split across support, sales, and research.
Discovery interviews repeat known questions and fail to target the riskiest assumptions.
Release status is assembled manually and important launch dependencies surface at the last minute.
Responders repeatedly answer status questions while facts, hypotheses, and actions mix together.
Postmortems lose time reconstructing timelines and often blur observed events with hindsight.
Engineers spend triage time requesting versions, reproduction steps, account context, and duplicates.
Technical debt is discussed anecdotally, so urgency and business impact are impossible to compare.
Blocked dependencies and unclear acceptance criteria become visible only after commitments are missed.
Customer and internal documentation silently diverges from shipped behavior.
Experiments end with a chart but no durable decision, limitation, or follow-up.
Quarterly access reviews bury high-risk entitlements among thousands of expected permissions.
Licenses renew while inactive seats and duplicate tools remain invisible to budget owners.
Broken fields and stale pipelines create downstream confusion without a shared severity or owner model.
Teams use the same metric name for different logic, populations, and time windows.
API and schema changes surprise downstream teams because dependencies are scattered across code and tribal knowledge.
IT repeats safe diagnostic steps while response quality varies by agent and queue pressure.
Change meetings spend time gathering basics instead of evaluating risk, timing, testing, and rollback.
Backups appear healthy until a real restore is needed, while test evidence is incomplete or hard to retrieve.
Employees adopt AI tools faster than security, procurement, and data guidance can respond.
Analytics queues fill with solution requests that lack a decision, audience, deadline, or reusable definition.
Renewals, notice windows, reporting duties, and service commitments remain buried in signed documents.
Privacy requests arrive through multiple channels with missing jurisdiction, identity, system, and deadline context.
New policies are published, but affected systems, teams, controls, training, and evidence are not mapped.
Audit evidence is requested repeatedly because files lack control, period, owner, and source context.
Risk registers age between reviews while incidents, vendors, controls, and business conditions change.
Security teams repeatedly answer similar questionnaires while inconsistent language creates commercial and legal risk.
Failed controls enter a flat queue even when exposure, duration, affected data, and compensating controls differ.
Regulatory updates are shared without mapping applicability, effective dates, affected processes, or counsel review.
Conflict disclosures arrive as free text and lack the relationships, decisions, timing, and reviewer context needed.
Legal queues mix urgent commercial deadlines with vague questions and missing documents.
Vendor comparisons become feature matrices disconnected from requirements, total cost, risk, and adoption.
Teams notice renewals after notice windows pass and usage, owner, alternatives, and performance are unprepared.
Procurement spends cycles chasing business need, budget, data access, alternatives, and accountable ownership.
Vendor reviews rely on anecdotes while SLA misses, tickets, adoption, cost, and commitments sit in separate tools.
Evaluation teams miss caveats and non-answers while comparing long vendor responses under deadline.
Ownership, financial, security, service, and geographic changes emerge between scheduled vendor reviews.
Contracts end while accounts, integrations, retained data, invoices, and knowledge transfer remain open.
Business stakeholders struggle to see which redlines affect economics, operations, data, risk, and timing.
Negotiated savings are reported without separating avoided cost, reduced scope, timing, and realized spend.
Employees buy duplicate or unapproved tools because the approved catalog is hard to navigate.
From pattern to production
We map where your hours actually go, rank the bottlenecks worth fixing, and define the success metric before anything ships.
The workflow goes live inside the tools you already use, with your context, rules, and human checkpoints.
We measure what it returns. Weak workflows get cut; strong ones expand.
WORKFLOW_RECORD
source ______
score _______
version _____
feedback ____
The benchmark reward is weighted operating value × evidence confidence − unmitigated risk. Publication quality remains a separate gate, so polished writing cannot outrank a workflow that is safer, more repeatable, and more valuable in production.