Riseplan from Roast & Rise

Agentic Work Redesign Sprint

A 30-day sprint to reset team roles, task horizons, and routines for the era of agentic work.

Learn how to redesign work for teams using AI agents to handle delegated, long-running, and cross-functional tasks. Build a new operating model that makes room for parallel delegation, reusable instructions, modern review cycles, and the next level of team collaboration.

A warm, atmospheric editorial scene of a large, clean workspace where a long table is bathed in gentle, Renaissance-style sunlight. On the table, a set of abstract work artifacts - maps, ladders, boards, and stacks of neatly arranged documents - unfold in sequence, each glowing subtly from within. Human hands (no faces shown) are reaching in from the edges, discussing and reorganizing the artifacts, evoking collaboration and transformation. The color palette is Roast & Rise orange, burnt red, neutral light, and white with deep soft shadows and ample negative space. No visible text, logos, charts, or dashboards. — The new unit of work: beyond chat, toward intentional design. This sprint is a path through mapped change, illuminated one step at a time.

Course thesis

When AI agents shift the unit of work from chat turns to delegated, long-horizon tasks, teams need more than digital permissioning - they need to remake roles, redefine what gets reviewed, and build parallel delegation as a muscle. Small pilots, decisive maps, and tight review rhythms are the way in.

What you leave with

By the end, you'll have a decision-ready operating map and clear routines to run agentic work - plus the tools, checks, and templates to pilot, measure, and extend the new model with your team.

For

Team leads, managers, operators, founders, and mixed business/technical teams actively experimenting with or managing AI agents for complex, delegated work across departments.

Workflow

Redesigning team operating models and workflows to delegate and verify long-running, cross-functional tasks with AI agents. Includes mapping work, shifting roles, piloting parallel delegation, building reusable instructions, and updating manager/verification routines.

Change

Move from ad-hoc agent experiments to a shared, mapped, and reviewed operating model for agentic work, including new delegation workflows, reusable task components, shared review cycles, and measured outcomes.

What you can do

Use these as checks while you move through the plan.

Map current and target agentic workflows across your team or department.

Identify and plan for the biggest role, responsibility, and review shifts triggered by agentic work.

Pilot and evaluate parallel delegation of long-horizon tasks using agents.

Build and refine a shared library of reusable instructions and agent skills.

Set up team-level review and verification rhythms suitable for agent-powered work.

Draft and adapt a 30-Day Work Redesign Memo to record what changes, what holds, and how success is measured.

Chapters

Diagnose: Map Your Agentic Work Landscape

Identify where agentic delegation can unlock value in your team's real work. Create a before-and-after Agentic Work Map to reveal manual, bottleneck, and agent-ready processes. Set a strong foundation for role design and pilot experiments.

An editorial diagram on a warm-toned work surface: a before-and-after workflow map. On the left, an organic path with bottlenecks and human figures at key steps (all abstracted, no faces), each step indicated by tactile orange markers. On the right, a more parallel, luminous path where several steps branch and converge, icons for agents (abstract, not robots) fit into the flow, and critical review points remain highlighted for humans. The palette is orange, burnt red, and white. No text, numbers, charts, or UI elements. — Map the ground truth: manual, automated, and agent-ready. Clarity comes first. See where agents fit, and where human review must remain.

Why this matters in the workflow

Agentic work is not smarter chat. The work unit shifts from answers-in-turn to outcomes-on-deadline. If you map work the old way, agents fit nowhere. If you map it the new way, opportunities appear. You see parallel potential, review traps, and the hidden human steps that trip agents up. OpenAI's data shows the work moving from single chat turns to longer delegated tasks. The map makes that shift concrete.

The working model

You need an Agentic Work Map. It is not a process map or a RACI chart. It is a before-and-after, showing: - The real steps in one workstream (choose something cross-functional, long-horizon - think: onboarding, product launch, research review). - Where today's process is manual, human-bottlenecked, or fragmented by chat. - Where agentic delegation actually fits: tasks that can be fully described, handed off, reviewed for outcome, and run in parallel. - Potential gaps and verification points.

Quality checklist

Captures all key steps, including quiet handoffs and pain points

Clearly marks manual, automated, agent-ready, and unclear steps

Identifies at least two agentic delegation opportunities

Highlights required human review stages

Includes at least one partner or cross-functional review input

Revised agentic workflow is actually implementable (not wishful)

Edits or feedback from review process are included

Common mistakes

Mapping only tasks that annoy, skipping big blocks of process

Assuming agentic steps are minor automations, not outcome delegation

Failing to mark verification or review points

Skipping feedback from cross-functional partners

Overcomplicating the map; missing quick wins or partial shifts

Checkpoint

Can you show your Agentic Work Map to another team and clearly explain which parts are delegated to agents, which remain manual or human-reviewed, and why?

Exercise

Map Your Team's First Agentic Workstream

Steps:

Gather the team members closest to one cross-functional, repeatable workflow (e.g., onboarding, reporting, campaign planning).
Using the map template, list every manual, automated, and unclear step as it exists today.
For each step, note: - Who owns it - Bottleneck risks (waiting, errors, silos) - Current review/check handoffs
Identify at least two steps where an agent - using clear input and output criteria - could own the outcome with a human review point.
Mark where human review cannot be skipped (e.g., regulatory, final decisions).
Redraw the flow showing: - Which steps become agent-owned blocks - Which reviews or checks are required - Where parallelization (multiple agents) is feasible
Review your draft with a partner from another team or function who depends on the workflow outcome.
Capture edits and agree on the "before" and "agentic" version of the map.

Use this at work tomorrow

Pick one real workflow, fill in the Agentic Work Map, and review with a partner - see where agents can take ownership and where review is essential.

Design: The Role Shift Ladder

Redesign roles, responsibilities, and review points to fit a workflow where agents handle delegated, long-horizon tasks. Document the changes with a clear ladder, covering what shifts, what escalates, and what the team now needs to review or learn.

Editorial diagram of an abstract ladder structure spanning left (old roles/tasks) to right (new roles/tasks). Human silhouettes shift upward along rungs to new positions - shifting from hands-on to oversight, review, and direction. Agent steps are integrated into the ladder as differently colored rungs. Key escalation points and human review areas are subtly illuminated. Palette: Roast & Rise orange, burnt red, pale warm white. No visible text, job titles, or numbers. No generic charts or icons. — The organizational backbone bends, not breaks. The ladder reveals who moves up to review, who guides, and where escalation flows.

Why this matters in the workflow

When agents take on delegated, cross-functional, long-form work, the team structure bends. The review points and accountability lines that made sense in step-by-step human workflows don't match the shape of agentic work. Roles shift from production to review, oversight, error-tracking, and instruction design. Without a mapped ladder, confusion slows adoption and kills trust in the new model.

Recent studies show many failures (Apostolou et al., 2026) come from teams skipping the redesign of roles, oversight, or escalation.

The working model: The Role Shift Ladder

Quality checklist

Covers at least one real and relevant workflow, not a hypothetical.

Explicit about which old tasks shift to agent, and which human tasks now focus on review, direction, or escalation.

Documents both agent and human roles clearly, with review and escalation points visible.

States new skill or upskilling requirements needed for the new model.

Review rhythm is specified: who, how often, what criteria.

Blind spots or remaining risks are named.

Output is visible to the team or at least one cross-functional peer beyond private notes.

Common mistakes

Only describing agent tasks, not restating role shifts for humans.

Leaving review or escalation points vague - e.g., 'Manager checks output' with no cadence or method.

Not naming specific skill or knowledge gaps for the team.

Making the Role Shift Ladder privately, with no feedback from another function.

Turning exercise into a generic table with empty fields.

Checkpoint

Can you point to a published Role Shift Ladder for a real workflow, signed off or challenged by at least one peer or partner?

Exercise

Create Your Team's Role Shift Ladder

Follow these steps to build a real Role Shift Ladder for your team. 1. Pick one real workflow recently mapped for agentic delegation - at least one with cross-functional steps.

List current roles: For every person/role involved, write their main tasks and review points today.
List agentic redesign: For each, document what tasks shift to agents, and what each human does now (review, instruct, escalate, trust, etc).
Define new review rhythms: How often and by whom are agentic outputs checked? Where can things fall through the cracks?
List new skill/upskilling needs: Be specific.
Draw the Role Shift Ladder: Fill in the template below, making it visible to your team or at least one cross-functional partner for feedback.
Note blind spots, unknowns, or risks. Commit to publishing this output in your workspace or team folder.

Use this at work tomorrow

Draft and share a Role Shift Ladder for one cross-functional workflow. Make role, review, and escalation changes explicit - then circulate for input.

Practice: Parallel Delegation in Action

Pilot running multiple agents in parallel on real, cross-functional, long-horizon tasks. Capture actual workflow breakdowns, blockers, and convergence points using a Parallel Delegation Board.

A warm-lit tabletop scene with an open Parallel Delegation Board: rows of visually distinct 'cards' represent agent task runs, each showing progress, blockers, and results in tactile, colored tokens. Some cards sit cleanly in parallel while others show small breakages (misaligned, stacked, or flagged). Two human hands point to cards where blockers exist. Palette: orange, burnt red, white, and soft neutrals. No visible text, numbers, or UI elements; board structure is clear from layout and iconography only. — A live board - tasks, agents, blockages, lessons. The difference between theory and learning is what gets surfaced here.

Why this matters in the workflow

Agentic work unlocks speed and reach - but only when teams move beyond single-task automation to parallel, multi-agent experiments. OpenAI's research shows over 10% of users now manage 3+ agent sessions a week, but the majority of organizations haven't made this jump (see Apostolou et al. 2026). The risk: assuming sequential agent use means you're agentic-ready or that successful single pilots scale cleanly to parallel, cross-functional work.

Parallel delegation forces teams to see where work splits, collides, or grinds to a halt. Live, side-by-side runs surface real bottlenecks: instruction gaps, review collisions, unclear ownership, and agent handoff failures. It's less about tech and more about real process friction.

The working model

Quality checklist

Covers real cross-functional workflows, not toy demos.

At least two steps run in parallel, not sequentially.

Board tracks blockers, handoff quality, and agents used.

Debrief includes one or more actual surprises or breakdowns with more evidence than 'it worked'.

Team agrees on 2+ lessons or next steps to trial.

Common mistakes

Picking tasks that are inherently sequential, not parallelizable.

Running theoretical exercises, not live work.

Not tracking blocker details or handoff issues as they happen.

Declaring success after just one experiment.

Skipping the team debrief or not recording lessons learned.

Checkpoint

Have you piloted at least one live workflow using the Parallel Delegation Board and captured both breakdowns and next steps in your review?

Exercise

Run Your First Parallel Delegation Board

You will: - Pick a cross-functional, multi-step workflow your team actually owns this week. - Break the workflow into at least two steps that could run in parallel. - Assign each step to an agent (real-world tool or simulated if needed). - Use the Parallel Delegation Board template to track status, blockers, time, and handoff quality. - Run the experiment and debrief the results. ### Steps

Select a workflow (e.g. campaign launch, onboarding, reporting, contract review).
Break out all steps. Identify at least two steps that could happen at the same time.
Assign each task to an agent, prep clear instructions, and set expected outputs.
Track each run's progress on the Board: note status, blockers, handoff issues, and times.
After completion, review the Board as a team. Capture: What worked? What jammed? What surprised you?
List 2 lessons and the next change you'll try for the next pilot.

Use this at work tomorrow

Pick any two recurring tasks this week that could run side by side. Document them on the Parallel Delegation Board and test agent handoffs - real friction is the lesson.

Implement: Build Skills & Review Rhythms

How to operationalize agentic work by building a live, shared library of reusable agent task instructions, and setting up a recurring review and verification cycle for outputs. Teams will leave with concrete assets and routines that travel beyond solo pilots.

A central, open shelf structure on a warm-lit table holds visible, layered task instruction templates - each sheet slightly offset, glowing with active use. Beside the shelf, a small circle of seated humans and an abstract agent icon gather for a recurring review, with an illuminated clock symbolizing the ongoing rhythm. Palette: orange, burnt red, white, and light neutrals. No visible text, labels, or numbers. The scene focuses on clarity, cycle, and accessibility, not screens. — Instruction evolves, review recurs. Pilots become practice only when skills are shared, reviewed, and improved in daylight - not buried in chat.

Why this matters in the workflow

Piloted agentic delegation quickly stalls without shared assets or disciplined output review. Lone operators create brittle prompts, instructions disappear into chat logs, and teams fall back to manual failsafes. Reinvention becomes routine. The OpenAI/Codex surge showed that non-developer teams could run parallel agents only when clear task blueprints and review cycles were in place. Without them, error rates and output confusion quietly compound .

To scale reliable agentic work, you need two things: - A visible, maintained Skills/Instructions Library anyone can use and update - A review rhythm that fits new agent/human boundaries - catching errors, surfacing learnings, preventing silent rot

The working model

Quality checklist

Template covers clear steps from start to output, using live work details.

Each template names at least one real edge case or input error.

Templates are accessible to every delegated agent workflow user.

Owner/editor and next review date are defined.

Review cycle is visible - next meeting scheduled and invite sent.

At least one real output was reviewed and led to a template adjustment.

Common mistakes

Templates only reflect ideal cases, not failures or odd inputs.

Instructions live in personal docs or chats, not a team-accessible hub.

Nobody owns the template; updates stall or drift.

Review meeting checks only 'completion,' not quality or new exceptions.

Edge case updates are forgotten between review sessions.

Checkpoint

Can you point to your team's live (shared) Skills Library and scheduled review? Did your last agent output update a template or trigger a review action?

Exercise

Draft and Launch Your Team Skills Library & Review Cycle

1. Identify 2-3 recurring tasks being handled by agents (or nearly ready for delegation). ### 2. For each task, draft a Skills/Instructions template using the provided structure below. Use real outputs and failures to fill in edge cases. ### 3. Place the templates in a central, team-accessible location (Docs folder, Notion, etc). Announce and share edit/view permissions. ### 4. Schedule a 30-minute live review with agent task owners and key team members. Bring real agent outputs from the past week. ### 5. In the session: - Review each template and a matching output. - Note (and update for) at least one new edge case or error per template. - Assign an owner to update and oversee each template. - Set a recurring review rhythm, and calendarize the next session. ### 6. Capture surface-level feedback and share a 3-point summary with the whole team. You now have the building blocks of sustainable agentic work with visible checks.

Use this at work tomorrow

Draft and store your first agent task template, then calendar a review. Teams often skip explicit cycles. Do not let this stay in chat.

Measure: Decide and Adapt in 30 Days

Turn pilot outcomes into an explicit operating decision - expansion, pause, or adjustment. Capture proof points, lessons, risks, and next actions in a memo, reviewed with the team.

A focused, warm-lit tabletop with a physically prominent memo document at the center, visually collated with the earlier work artifacts - map, ladder, board, and skill templates - arrayed around it. A group of team hands (no faces) gathers for review and approval; the central memo is marked with a bold, illuminated orange circle as a decision point. Palette: orange, burnt red, white, restrained light. No visible text, signatures, stamps, charts, or screens. — End the endless pilot loop. The memo collects artifacts, lessons, and hard truths - making outcome, risk, and next actions visible and actionable.

Why this matters in the workflow

Pilots are only step one. Leaving experiments in limbo kills momentum and creates confusion about what changes, what doesn't, and why. Real transformation starts when you make the risks, results, and next actions visible - and force a decision. The 30-Day Work Redesign Decision Memo stops endless trial mode. It turns agentic experiments into operating knowledge and team-wide follow-through.

The working model

Borrowing from OpenAI's frontier pilots, but grounded in normal-company reality:

Quality checklist

Memo covers all major outcomes, proof points, and remaining risks (no missing sections)

Decision (go/pause/adjust) is unambiguous and justified by pilot data

Concrete next steps are assigned to named owners

At least two peers/managers/executives have reviewed and commented

Findings and action points are shared with all impacted team members

A fixed date is set for next review - no 'sometime later' or vague cycles

Risks and gaps are named honestly, not smoothed over

Common mistakes

Glossing over poor results or gaps to push for expansion

Keeping the memo in a private folder - not sharing with team

Missing reviewer signoff or skipping cross-functional feedback

Listing next steps with no owners or due dates

Not clearly stating the go/pause/adjust recommendation

Treating the pilot as finished - no next review scheduled

Checkpoint

Can you show a complete, reviewed Decision Memo - with a clear go/pause/adjust call, next steps, and next review date - for your agentic work pilot?

Exercise

Draft and Review Your 30-Day Work Redesign Decision Memo

Purpose: Lock in the learning and clarity. Distill the real outcomes, blockers, and what needs to change - then drive a go/no-go or adjust decision with your team. ### 1. Gather Data and Outputs - Collect the outcomes (quantitative and qualitative) from your agentic pilots: task speed, error rates, confusions, successful handoffs, surprises. - Bring together your previous outputs: Agentic Work Map, Role Shift Ladder, Parallel Delegation Board, and at least one reusable skill/instruction. ### 2. Fill Out the Decision Memo Template (see below) - Capture what worked, what failed, key proof points, risks, and your recommended path. ### 3. Run a Review with Key Stakeholders - Share your draft with at least your manager and one cross-team partner or exec. - Lead discussion with data and findings. Push for clear agreement or challenge. ### 4. Finalize and Publish - Edit the memo with feedback. Share with all affected team members and post action items publicly - calendar the next review date now. You can complete the draft memo in one working session - doable in 15 minutes if you prepare your data.

Use this at work tomorrow

Draft your pilot's 30-Day Decision Memo, share with two stakeholders, and turn feedback into your next operating plan.

30-day path

Week 1: Map agentic opportunities and gather real work examples.

Week 2: Run the Role Shift Ladder and plan skill/review pilots.

Week 3: Launch and track parallel agentic delegation pilots.

Week 4: Build skills/instructions library, run reviews, collect outcomes.

Final 2 days: Collate data, fill out Decision Memo, and hold review with team leads or execs.

Success signals

Number of team members participating in mapping and redesign.

Volume and effectiveness of parallel agent task pilots completed.

Quality and reuse rate of new skill/instruction templates.

Manager and team confidence in the updated review and verification routines.

Clarity and uptake of the 30-Day Decision Memo.

Reflection prompts

Where does this topic show up in real work?

What behavior should change first?

What evidence would prove this Riseplan worked?

Manager checklist

Choose one owner for the behavior change.

Use the exercise on live work.

Review the output before scaling the habit.

Decide what changes after 30 days.

Want this shaped around your company?

Risey can research your company foundation first, then build a version of this path around your real workflows, customers, and culture.

Start with your company