HR TECH 2026 · VICTOR “VIC” AKOSILE
The Architecture. Why your HR pilots stall
CloudflareAI transformation & people strategy
Hold before you start. The formal conference title remains the principal title.
Prepared content is targeted at about 33 minutes. The duration is a planning estimate, not a measured performance. Begin with your prediction on the next slide. Use Right or PageDown to continue. The People org chartchanging. MY 2027 PREDICTION Our own work agent becomes a new front door to work.
I think the org chart of the People function is changing. My prediction for 2027 is that more of us will begin work through our own work agent, connected to specialized capabilities across the organization.
This is my prediction, not a statement that everyone already has an agent, or that HR will disappear. The work changes; human judgment, process ownership and accountability remain. The people in this room have a role in leading that change. One work agent. A connected system. A possible way an HRBP prepares for a talent review.
THE PERSON HRBP “Prepare my talent-review briefing.”
THE COORDINATOR My work agent Plans · calls permitted tools · assembles context
People analytics
Skills inventory
Internal mobility
Policy & approved guidance
The briefing comes together. The HRBP still makes the decision.
Concept illustration based on Vic’s supplied LinkedIn post; not a deployed-system diagram.
Imagine an HRBP getting ready for a talent review. They ask a work agent to prepare a briefing.
CLICK: the work agent coordinates specialized capabilities across people analytics, skills, internal mobility and approved guidance. This is a possible architecture, not a claim about a live deployment. Some capabilities may be tools or deterministic workflows rather than separate autonomous agents.
CLICK: the briefing comes together. The HRBP still makes the decision. Each contributing workflow needs ownership, permission boundaries, evaluation and an escalation path. A second org chart. Still human-led. 40 possibilities across 8 functions—not a list of what to build.
CHRO & executive leadership Direction · judgment · accountability
AI Transformation AI Product Manager Governance & Compliance Lead Agent Builder Technical Lead
Talent Sourcing Agent
Resume Screening Agent
Scheduling Agent
Candidate Q&A Agent
Offer Letter Agent
Onboarding Pre-boarding Agent
Day-1 Orchestration Agent
Buddy/Mentor Match Agent
30/60/90 Check-in Agent
New Hire Status Bot
Talent Mgmt & Dev Career Path Agent
Mobility Match Agent
Skills Inventory Agent
Conversation Prep Agent
Succession Mapping Tools
Business Partners HRBP Briefing Agent
Coaching Support Tool
Engagement Pulse Agent
Listening Synthesis Agent
Team Health Agent
People Operations Policy Q&A Agent
HR Service Desk Agent
New Hire Provisioning Agent
Case Triage & Routing
Investigation Doc Agent
Total Rewards Salary Banding Agent
Pay Equity Analysis Agent
Benefits Q&A Agent
Equity Education Agent
TR Statement Agent
People Analytics NL Query Agent
Attrition Risk Agent
Workforce Reporting Agent
Headcount Modeling Agent
Predictive Modeling Agent
HR Technology Data Quality Agent
Integration Health Agent
Agent Performance Monitor
Sandbox Agent
Documentation Agent
Shared foundation Knowledge & data Identity & access Audit & trust Integration hub
Vic Akosile · supplied “40 AI Agents” framework. Concept map, not a Cloudflare deployment list.
This map makes the possibilities tangible. It is not a deployment inventory and it is not a recommendation to create 40 agents.
The two rows preserve the original map’s employee-facing and people-infrastructure functions. Human direction and governance sit above the work; knowledge, access, audit and integrations support it. These are owned capabilities, not necessarily four new hires.
The point is to choose problems worth solving, allocate responsibility, and build capabilities that can last. Risk-sensitive uses such as screening, pay and employee relations require appropriate human oversight and validation. Offer a pause for a photograph; do not read all 40 entries. Three promises. 1
See value 2
See what we’re 3
Leave with a way A practical approach to redesigning workflows in your own organization.
Three promises. A better way to identify AI value beyond time saved. A practical look at what we are building and learning at Cloudflare. And an approach you can use by Monday to start redesigning workflows in your own organization.
Ask: would those three things make this time worthwhile? Pause briefly, then move into the LinkedIn invitation. Let’s keep the Share one useful takeaway.
I’ll send you the slides.
Help someone who isn’t in the room get value today.
Day-of LinkedIn post The post URL has not been supplied. Add the link in Setup
This QR is for the day-of LinkedIn post, not the resource form. Invite one takeaway or a question; questions can begin with Q:. Do not ask people to like, repost or praise the talk.
The promised slides need a real follow-up process. The opening QR opens the published HR Tech LinkedIn post. The closing QR opens builtinto.work/hrtech. Both are embedded in this portable file; QR generation is local.
Imagine your kitchen So you give everybody
Illustrative AI-generated kitchen image from the earlier deck.
Imagine your kitchen is driving you crazy. Not enough room. Everything you need seems to be on the wrong side. Two people try to make breakfast and both end up in front of the same drawer.
CLICK: you decide to transform it. So you give everybody in the house a power tool. Good tools, training, encouragement to experiment. Somebody discovers they can build shelves. Someone improves a cabinet.
There is learning and useful activity. But nobody has agreed on the design, what can move, who handles the plumbing, or who is responsible for the finished kitchen. And everybody still needs breakfast. This is an imagined scenario, not a personal renovation story. The tools The kitchen is still
Let the paradox land. The tools may be excellent. The people may be talented. Some improvements may be useful. But we have asked people to renovate a shared space while still living in it, without making room for the renovation.
CLICK: the kitchen is still difficult to live in. This is not an argument against the tools or against experimentation. It is a question about the change we expected them to produce. ?
We gave people AI. Pause. Ask the question naturally. It introduces the next movement; do not give the whole answer here. More people are using AI. U.S. employees using AI in their role at least a few times a year.
Gallup · Global Indicator: Artificial Intelligence · dated tracking table, May 2023–Aug 2026. https://www.gallup.com/699797/indicator-artificial-intelligence.aspx
Gallup’s published tracking table: at least a few times a year, 21% in May 2023, 27% in May 2024, 40% in May 2025 and 53% in August 2026. Weekly-or-more use rose from 11% to 32% between the first and final dates.
This is a U.S. employee-use measure, not the percentage of companies using AI and not a measure of financial return. The table is newer than some summary text on the page. We use the dated table values.
More people are using AI. That is a starting point. It does not establish what changed about the work. Avoid presenting a viral pilot-failure statistic as a universal failure rate; first ask what an experiment was intended to prove. ?
What value are we Pause. Ask the question naturally. It introduces the next movement; do not give the whole answer here. Using the tool is only the beginning. AI activity What do we do?
→ Workflow change What becomes different?
→ Value created What gets better?
→ Business outcome Why does it matter?
Connect the steps with evidence—not just arrows.
Leaders need to connect what people do with AI to an actual change in how work is done, the value that change produces, and an outcome the organization cares about.
These arrows are a hypothesis to test, not proof of cause and effect. Start with the desired outcome and examine the chain. If a tool produces more information, who uses it and what decision or action becomes better?
Transformation: a lasting change in how work gets done that improves what we can deliver and the experience of delivering it. The 7-layer AI value stack. Start with the change. Then decide how to measure it.
1 Efficiency Less time, cost or effort per task.
What happens to the capacity released?
2 Scale More volume with available resources.
Can the whole workflow handle more?
3 Quality More accurate, consistent, complete work.
What does “good” look like?
4 Experience People feel better served.
Does the work feel easier to navigate?
5 Risk Fewer failures and better defenses.
Which exposure have we reduced?
6 Capability Do something we could not do before.
What can we now deliver?
7 Financial impact Translate operating value into money.
Show the mechanism; include the costs.
Adapted for 16:9 from Vic Akosile’s supplied “The 7-Layer AI ROI Stack.”
Vic’s seven-layer framework: Efficiency, Scale, Quality, Experience, Risk, Capability, and Financial Impact. This is a practical framework, not a claim of a formally validated scale.
Time saved can be real value, but it is not automatically cash saved. It may become capacity, quality, or less strain. Scale requires looking beyond a faster task to the system’s bottlenecks. Capability means something people can now do—not simply a tool they own.
Financial impact translates demonstrated benefits into money and accounts for delivery, operating and review costs. Avoid adding overlapping benefits twice. Pause for a photograph; choose a few layers to explain rather than reading every cell. OfferGEN: follow the value. A concrete chain—not a claim that every benefit is already measured.
CHANGE THE WORK
Validated inputs. CREATE VALUE
Less editing. TEST THE OUTCOME
Accurate offers, Evidence to track Preparation time Rework Exceptions Errors
Supplied use-case portfolio, p.3; workflow narrative from Vic. Proposed value measures, not reported results.
Introduce OfferGEN, with GEN capitalized. The inventory names it OfferGen; the deck uses Vic’s preferred styling. It validates Greenhouse data and generates country-specific offer documents in batches. The supplied inventory labels it LIVE and TEAM-BUILT.
The desired value is less manual editing and rework, and more consistency. The business outcome to test is accurate offers ready sooner. These measurement suggestions are not measured results or quantified savings.
We will return to how it works. For now, notice that the tool’s output, the operating benefit and the desired business outcome are not the same thing. ?
What does this look like Pause. Ask the question naturally. It introduces the next movement; do not give the whole answer here. Four A’s. One working system. These are complementary mechanisms—not a ladder of sophistication.
A1 Assistants Help a person think, write,
A person directs the exchange. A2 Applications Give a defined job
The software need not use a model at runtime. A3 Automations Run defined steps
Rules make the path repeatable. A4 Agents Choose and carry out steps
Human boundaries and oversight remain. Use the simplest mechanism that does the work well.
Vic’s four-A framework. Technical reference: Anthropic, Building effective agents (2024). https://www.anthropic.com/research/building-effective-agents
Explain the four A’s in about 90 seconds. Assistants help a person interact. Applications provide a defined experience. Automations execute predefined paths. Agents use models to select actions and tools toward a goal, within boundaries.
Real solutions combine these. An application may contain an assistant, call deterministic automations and ask an agent to resolve a bounded exception. More autonomy is not automatically better.
The workflow/agent distinction is consistent with Anthropic’s engineering guidance; the four-A taxonomy is Vic’s presentation framework. A governed place to build. Cloudflare OS
Ask with context Build useful applications Share within governed boundaries
Screenshot supplied by Vic. Public context: https://blog.cloudflare.com/cloudflare-os/
Use the actual Cloudflare OS screenshot supplied by Vic. It is not a mockup. Explain the environment: asking questions with context, creating outputs, and building and sharing useful applications or gadgets.
Cloudflare’s public description combines a workspace, governance/security framework and a platform for building applications. Access controls and approvals still depend on the data and workflow; do not imply blanket permission for sensitive HR data.
Workers is part of the platform people can build on. Keep the technical explanation brief: the point is a governed way to do more than chat. Project Blue Instant workforce answers, built around governed definitions and access.
“What changed in our workforce this quarter?”
Ask in
→ Apply definitions
→ Calculate with SQL
→ Return an answer
Capability A question in the meeting becomes an answer people can use.
Vic’s prior Project Blue / SIOP description. Concept view—not a product screenshot or a measured-impact claim.
Blue is a workforce-intelligence example. Authorized users ask plain-language questions; governed definitions and access are applied; SQL calculates against approved data; the output is a table or chart plus explanation.
The question on this slide is illustrative. This is a concept view, not a screenshot or a claim about measured accuracy, adoption, savings or current rollout scale. The supplied use-case PDF does not list Blue.
Use Blue to show new capability and a changed interaction. Do not spend as long here as on onboarding. Further dashboard-control capabilities were discussed previously as future or ongoing; they are not represented as complete here. What we’re building—and exploring. 23 use cases across Compensation, Benefits and HR Operations.
Compensation7
Total Rewards Agent (Pave)
Promotion & Compensation Assistant (Skill)
Compensation Intelligence
Retention & Pay Equity Analytics Benefits6
Benefits & Payroll Reconciliation
Benefits Help Agent (Astra)
AI Content Audit and Migration
Benefits Vendor & PO Operations
Benefits Plan Audit Assistant HR Operations10
OfferGEN — Offer Letter Creation & Automation
ClearPath — Background Check Automation
Global Time & Leave Compliance
People Ops Audit & Process Center
Severance Document Automation
People Ops Learning Portal
Workday AI Integration Layer Supplied use-case portfolio, pp.1–3. Status labels reproduced as provided; this is not a count of live deployments.
A photograph-friendly inventory, preserving the supplied PDF’s categories, patterns, outcomes, maturity and build-model labels. These are mixed stages, not 23 live deployments.
There are 7 Compensation, 6 Benefits and 10 HR Operations entries. Do not classify all as agents: the source includes platforms, automations, assistants, analytics and an agent. OfferGEN is styled as Vic requested; the source spells it OfferGen.
Give the room a moment to take a photograph. Pick one or two examples only. We will zoom into the onboarding journey rather than narrate the entire inventory. Source: Frank-Conference-Board-AI-Use-Cases.pdf, pages 1–3. Project Nova The intended onboarding experience: ready for day one.
TARGET-STATE ILLUSTRATION THE EXPERIENCE One place to know
READY Defined and verified—not assumed.
Offer OfferGEN
Checks ClearPath
Tasks & access Connected systems
Welcome People & manager
Assistant: guide Application: show status Automation: execute Agent: coordinate within limits
Nova direction from Vic. OfferGEN, ClearPath and onboarding maturity: supplied portfolio p.3. Target-state illustration, not current architecture.
Nova is the broader onboarding transformation direction described by Vic. This is an illustrative target-state view, not a supplied current process map or a claim that every integration is deployed.
The supplied portfolio labels Onboarding Orchestration UNDERWAY / HYBRID. OfferGEN and ClearPath are LIVE / TEAM-BUILT in that source. Keep those distinctions visible in the narration.
Start with the desired experience and define readiness. The four A’s are mechanisms that can contribute: an assistant guides, an application provides visibility, automations carry out repeatable steps, and agents coordinate permitted work or exceptions. The human owner remains accountable. Start where it hurts. A big journey. A bounded first problem.
The larger journey Onboarding
A painful part Offer preparation Different roles. Departments. Geographies. Languages.
Improve a painful part now—without losing sight of the whole journey.
We did not need to finish all of onboarding before relieving a painful part of it. The complexity of offer preparation made it a bounded place to begin.
The applicant tracking system’s generic ability to generate an offer was not the whole problem. The many combinations of roles, departments, locations and languages created repeated validation, template editing and checking.
This is thinking in horizons without introducing another framework: deliver relief now while working toward the larger journey. Make sure the local improvement fits the direction, rather than becoming another disconnected tool. OfferGEN AI helped write the code. Defined rules generate the letter.
Before Validate ATS information Download a master template Remove what does not apply Check, edit and check again Redesigned Validate Greenhouse inputs Apply approved document rules Generate country-specific offers Keep review and exception handling LIVE · TEAM-BUILT · AUTOMATION Source status; no quantified impact claim.
Vic’s workflow account and supplied portfolio p.3. Display styling: OfferGEN; source spelling: OfferGen.
Tell the actual offer-preparation story. People were validating ATS data, downloading a master template, removing irrelevant content, and checking repeatedly.
CLICK: OfferGEN validates Greenhouse information and generates country-specific offer documents in batches. Defined rules—not a model improvising contract terms—determine the document behavior. Appropriate review and exceptions remain.
The source labels OfferGEN LIVE / TEAM-BUILT / AUTOMATION. Avoid adding unverified time-saved percentages or claiming all review is eliminated. Deterministic code still needs validation, testing, secure access and maintenance. Build reliable workflows. Let AI help where flexibility matters. Use defined operations where repeatability matters.
Human intent “Help me complete this work.”
INTERPRET Assistant Understand the request.
→ EXECUTE Defined Validate inputs.
→ VERIFY Review Check the outcome.
Permissions · logging · tests · approvals · an accountable owner
Conceptual pattern informed by Vic’s experience and Anthropic, Building effective agents. https://www.anthropic.com/research/building-effective-agents
Language models can translate human intent into text, including code. One practical benefit is building software that can then follow defined, tested rules.
This diagram is a design pattern, not the deployed architecture of every Cloudflare example. Flexible language interaction can sit above repeatable operations. Agents may call those operations within explicit permissions rather than improvise every step.
Do not promise perfect reliability. Both generated code and model-driven actions need testing, controls, monitoring and human escalation. The design choice should follow the task and the failure consequences. Four contributions. One workflow. Illustrated through the work of creating and using OfferGEN.
Navigator Clarifies the outcome.
Practitioner Knows the work.
Builder Connects systems.
Amplifier Helps people use it.
OFFERGEN
Not four mandatory hires. Not everybody has to be a builder.
Vic’s Navigator / Practitioner / Builder / Amplifier framework; illustrative application to OfferGEN.
Keep the four contribution roles. Apply them to the same example so they do not become an unrelated taxonomy. The mapping is illustrative; do not imply it is a verified staffing chart for OfferGEN.
A navigator clarifies what matters and gets decisions made. A practitioner understands the work and exceptions. A builder connects systems and tests the software. An amplifier helps colleagues use it, learn and improve. People can contribute in more than one way.
This is an opportunity for the audience to see themselves. The person who knows why a handoff fails is essential, even if they never write code. ?
Who has the capacity Pause. Ask the question naturally. It introduces the next movement; do not give the whole answer here.
Do the work. Learn AI. Reinvent how the work gets done.
Same calendar. Ask: how many of you could do your current job and comfortably take on another person’s job as well? Pause. If a hand goes up: we should talk afterward. Keep it warm.
We assumed domain experts with tools would naturally redesign individual, functional and cross-functional work. But those people were already carrying the existing work.
CLICK: same calendar. Return to the kitchen: the people who know it best are still making breakfast. Transformation is work; it needs capacity. Transformation capacity. The time, people, expertise and authority deliberately available to change how work gets done.
Reduce
Stop or simplify Protect
Make time—with Add
Embed or borrow An offsite concentrates capacity. It does not create free capacity.
Define transformation capacity plainly. Can the team step back, understand the workflow, redesign it, and implement the change while sustaining essential work?
Three honest options: reduce existing demands, protect time with coverage, or add/embed/borrow delivery support. Usually the answer is a combination.
An offsite is a deliberate concentration of capacity; it is not a substitute for capacity after the meeting. If the team returns to an unchanged workload with no ownership or time to implement, the design can still stall.
The builder left. Useful tools create
Experience described by Vic. Image is an AI-generated illustration, not a real employee’s workstation.
We are now using tools or agents built by people who no longer work at the company. The builder moved on; the work and the dependency stayed. This can be evidence of a useful contribution, and it creates responsibility.
Someone must understand the purpose, permissions, failure modes and maintenance needs; decide what changes, and retire it when appropriate. Do not invent a named departed employee or a specific incident.
Use the SaaS-freeze joke as a bridge: a freeze on new SaaS purchases encouraged building. I may have lost several dinner invitations by saying that at HR Tech. But building comes with maintenance, risk and ownership. Buying can be the right answer too. There is more than build versus buy. Five ways to organize delivery capacity.
Closer to the problem Farther from the problem →
Self-serve builders The people with the problem build it.
BEST WHEN The hard part is knowing what to build; breakage is limited.
WATCH FOR Sprawl. Orphaned tools become load-bearing.
Embedded engineers Engineers work within your function.
BEST WHEN Domain knowledge and engineering need to stay together.
WATCH FOR Support ownership and the cost of ongoing capacity.
Central build team One shared team builds across functions.
BEST WHEN Work is standard, specified and shared.
WATCH FOR Queues, lost context and shadow work.
Borrowed engineers Engineers join from another team.
BEST WHEN A scoped build, clear stakes and a real end date.
WATCH FOR Side-of-desk allocation and the handoff afterward.
Vendor model An outside company builds and runs it.
BEST WHEN Standard work, heavy load and a mature market.
WATCH FOR Fit, integration, oversight and knowledge retained inside.
Choose a mix—not a winner. Is the scarce resource problem knowledge, technical craft, or sustained capacity?
Adapted from Vic’s supplied “Designing Your AI Delivery Model.” Five models and problem-proximity framing preserved.
Use the supplied five-model framework. Self-serve, embedded, central, borrowed and vendor models are options to combine; this is not a maturity ladder.
Closeness can bring motivation and context. Greater distance can bring specialist capacity and craft, but those are trade-offs, not guarantees. Ask whether the hard part is knowing what should change or building and sustaining the change.
The source’s warnings are risks to consider, not inevitable outcomes. Different workflows may warrant different models. Do not argue that all functions should build or that all vendor products are a poor fit.
This landscape explains how leaders can create the support needed to transform work. It replaces a separate build/buy/wait lecture. ?
How do we get started? Pause. Ask the question naturally. It introduces the next movement; do not give the whole answer here.
A workflow Protected time.
Vic’s facilitation approach. Illustrative AI-generated workshop image.
The most reliable way I have found to create this capacity intentionally is a workflow-transformation offsite. It brings the right people together around a consequential workflow, with preparation before and delivery ownership after.
This is not a brainstorming day about all the exciting things AI could do. It is a working session to define the outcome, challenge the current process, design a better one, and decide what will happen next.
The image is illustrative, not a photograph of a specific Cloudflare workshop. There are many suitable formats; do not promise every offsite produces deployed enterprise software. Find the pain. Redesign the work. Put it to work. BEFORE THE ROOM Find Interview the people doing and receiving the work. Gather recent examples and a baseline. Choose a bounded problem and an owner. Leave with: a clear brief. INSIDE THE ROOM Redesign Define the outcome and what “done” means. Work backward. Challenge assumptions. Walk through the process and exceptions. Leave with: a tested process design. AFTER THE ROOM Put it Assign delivery, approvals and operating ownership. Implement and test the bounded change. Measure, review and choose the next step. Leave with: an owned delivery plan. Vic’s workflow-transformation-offsite method; synthesis of the approved talk.
This is the method worth photographing. Before the room, interview people on both sides of the workflow and collect recent examples, not just opinions. Choose a boundary small enough to examine.
Inside the room, define the outcome and completion criteria. Ask what has to be true, challenge inherited assumptions and compare with the existing process. Simulate normal and exception cases.
Afterward, assign implementation, approvals, testing, operating ownership and a review date. The simulation tests the process design; technical testing still has to follow. Improvement may require no AI at all.
The method is available to teach. Facilitation value lies in the preparation, the questions, resolving decisions and supporting the handoff—not withholding the steps. What actually has to be true? Start with the outcome. Then challenge the route to it.
DESIRED OUTCOME
A new employee can begin meaningful work Required The access needed for
Chosen A named person who helps
Inherited A chain of emails used
Keep the commitment. Question the mechanism.
Illustrative onboarding example from the approved talk; not a statement of Cloudflare policy.
The example is illustrative. Not every step is a hard requirement. Some are deliberate choices about the experience and values we want to express. Others are simply the mechanism inherited from previous constraints.
We may still want a buddy because welcoming someone matters. That does not mean the same sequence of manual emails must remain. Ask what truly has to be ready now, what we choose to provide, and which mechanism could be simpler.
Be careful not to eliminate a real legal, security or policy requirement merely because it is inconvenient; involve the responsible owner. “Then somebody Who? Checks what? What if it’s wrong?
During the session, walk through the process rather than jumping over the awkward parts. When someone says “then somebody checks it,” stop.
CLICK: Who? CLICK: Checks what? CLICK: What if it is wrong?
Ask how the person knows it is their turn, what information they need, and how an exception moves forward. Simulate missing data, a changed decision, an unavailable owner and a system problem. This is one place facilitation turns a good idea into something implementable. Leave with decisions—not just sticky notes. The workflow brief One workflow · one accountable owner · a review date
OUTCOME What will be different?
RELIEF What disappears, simplifies,
BOUNDARIES What stays human?
DELIVERY Who changes the process?
OWNERSHIP Who supports and improves it
EVIDENCE Baseline → test → review.
This is a photograph-worthy summary of the output. We are not aiming for a wall of sticky notes that no one touches after lunch. We need decisions, a process design and an accountable implementation plan.
Name what becomes better, what work is removed, what remains human, who has decision rights, who implements, who operates it and what evidence will be reviewed.
Some changes can be implemented without touching AI. Others require integration, configuration or a build. The blueprint should survive the handoff to the people doing that work. Choose a workflow. Take a moment. Make the problem concrete.
1 A recurring problem people can describe with a recent example.
2 A boundary small enough to test—and an owner who can act.
3 A result worth improving, with evidence of what happens today.
4 What happens if the new approach gets it wrong?
Bring that workflow. Bring the people who know it.
Give the audience a short private reflection. Think of one recurring problem with a recent example, a named owner, enough evidence to understand today, and a boundary small enough to test. What result is worth measuring? What happens if the new approach is wrong?
“Use AI in onboarding” is too broad. “Managers repeatedly miss three pieces of information before a start date” gives us something to investigate. This example is illustrative.
Allow 30–45 seconds of silence. The deck does not auto-advance; hold for the room. Take the answer into the resource invitation. Take the method Slides. Free templates.
Start a better conversation
Work-email resource signup.
Resource signup page The final page URL has not been supplied. Add the link in Setup
The closing QR is a separate destination from the LinkedIn post: a resource signup page with slides, free templates and guidance on leading a workflow-transformation offsite. Vic wants work emails only and optional interest in learning or arranging an offsite, with priority workflows.
The page and its final URL are not provided or built by this deck. Press L to configure it before presenting. Do not claim form collection is live until the destination is actually deployed.
Briefly mention facilitated offsites: preparation, bringing the right people together, and working through a consequential workflow. Then finish the talk rather than ending with a sales pitch.
The kitchen For the people who have
Illustrative AI-generated image from the earlier deck.
Return to the kitchen. The measure is not simply how many people learned to use the drill. That learning is useful, but the kitchen needs to work for the people who live there.
Our colleagues have their version of breakfast tomorrow: candidates waiting for an answer, new employees trying to begin, managers trying to make a decision. They still have work to do.
Close the three promises: value beyond time saved, real examples of what we are building and learning, and a method to begin redesigning workflows. You know the work.You can help Give people the tools—and the capacity to transform.
The people who understand the work are already there. They know what gets copied, what gets checked again, what gets stuck and what they keep doing because nobody has had the time or authority to change it.
Their knowledge is a place to begin. Give them tools, learning, time, support and the authority to design a better way together. They do not all have to become builders.
Pause. Thank the audience. The next slide is an optional Q&A holding screen with the resource QR. Do not rush the final line. Let’s talk about What is painful?
Victor “Vic” Akosile Cloudflare · Built Into Work Resource signup page Add its URL in Setup.
Optional Q&A holding slide. The resource QR repeats so people can scan without interrupting your closing words. Questions can also be in the LinkedIn post.
No auto-advance. The rehearsed material ends before Q&A. Use the remaining time in the 45-minute session for discussion. 
