Shiken

Work Intelligence · Enterprise Intelligence Suite

Plan for AI at the task level

Decompose every role into tasks. Score AI exposure with the Stanford Human Agency Scale. Forecast where to augment, automate and hold human-in-the-loop — and capture the capacity that comes with it.

Become a design partnerSee the full suite

Design-partner programme open now. Enterprise GA: H2 2026.

Trusted by 30k+ people from top universities & businesses across the world

Nestlé
University of Cambridge
Merck
NHS
University of Oxford
Electrolux Group
University of Pennsylvania
MIT
Nestlé
University of Cambridge
Merck
NHS
University of Oxford
Electrolux Group
University of Pennsylvania
MIT
Nestlé
University of Cambridge
Merck
NHS
University of Oxford
Electrolux Group
University of Pennsylvania
MIT

How it works

Decompose. Score. Forecast.

Three layers that turn “what does AI mean for our roles?” into a quarter-by-quarter capacity plan.

Decompose work to the task

A role is too coarse for AI planning. Shiken decomposes every role into its component tasks, drawing on org charts, JD libraries, recorded work and ESCO mappings.

  • Task-level inventory per role
  • Frequency and time-weighted
  • Anchored to real recorded work

Score AI impact, not vibes

Every task is scored against the Stanford Human Agency Scale — Human, Augmentable, Automatable — using a structured rubric that maps to the latest research and your tooling.

  • Stanford Human Agency Scale rubric
  • Tooling-aware scoring (Copilot, GPT-5, etc.)
  • Confidence-banded by evidence depth

Forecast capacity by quarter

Project the augmentable and automatable hours per BU, region or function. Plan the capacity gain — and the upskilling that captures it — before the all-hands.

  • BU-level capacity forecasts
  • Skill gap to capture augmentation
  • Linked back to Skills Intelligence

The rubric

Built on the Stanford Human Agency Scale

Three buckets that map to action. Confidence-banded by evidence, refreshed as the tooling landscape evolves.

Human

Stays human-led

Judgement-heavy, relational, or regulated work where human-in-the-loop is mandatory or strongly preferred.

Examples

  • ·Compassionate clinical decisions
  • ·Compliance attestations
  • ·High-stakes 1:1 coaching

Augmentable

Human + AI, together

Work where AI gives material lift but the human still makes the call. This is where most upskilling pays off.

Examples

  • ·Sales discovery prep
  • ·Marketing brief writing
  • ·Code review and design

Automatable

Hand off to AI

Structured, high-volume, low-judgement tasks that can be fully automated — freeing humans for the work that needs them.

Examples

  • ·Meeting recap delivery
  • ·L1 ticket triage
  • ·Standard report generation

Who it's for

For the leaders being asked “what's our AI plan?”

CIO & Chief AI Officer

Move from PoC to portfolio

Stop running disconnected AI pilots. Build a portfolio view of where AI lifts capacity across the org — and which roles need investment to capture it.

  • AI portfolio dashboard by BU
  • Capacity-gain forecasts in hours and £/$
  • Augmentation-ready skill profile gaps

Sample output

Finding

47 disparate AI tools in use, only 3 mapped to capacity plan — concentration unclear

Recommendation

Consolidate to 8 tools, sequence by capacity gain, build governance for the rest

Forecast capacity gain

~2,400 hours/quarter in 12-month plan

CHRO & Workforce Planning

Reshape roles before reshaping headcount

When you can see which 28% of a role is automatable, you can redesign the work, not just the org chart. Capture the gains and avoid the layoff narrative.

  • Role redesign playbooks
  • Internal mobility into Augmentable roles
  • Capacity-aware hiring plans

Sample output

Finding

12% of Customer Support tasks fully automatable, 41% augmentable — re-design before restructuring

Recommendation

Role redesign: Tier-1 specialists with AI co-pilots, route automatable to async AI

Forecast capacity gain

~31% capacity expansion without headcount change

L&D leaders

Target training where AI multiplies it

Train the skills that compound with AI. Shiken connects Augmentable task scores to skill profile gaps to recommend the highest-leverage learning paths per role.

  • Augmentation-aligned learning paths
  • AI fluency embedded by function
  • ROI measured in capacity, not completion

Sample output

Finding

AE cohort missing AI-pair skills for 6 of top 8 Augmentable tasks

Recommendation

Targeted AI-pair-programming microlearning + roleplay rollout, 2-week cadence

Forecast capacity gain

~6.4 hours/rep/week if cohort hits L3 by Q2

Enterprise Intelligence Suite

Three questions every L&D leader needs to answer

One data foundation, three flagship products — needs, skills and work. Each answers a question the others can't.

Needs Analysis

“What training do they need right now?”

Diagnose what to train, fast

A guided 3-stage agent workflow — plan the audience, interview them with multi-channel polls, and roll up the priorities into board-ready needs reports.

  • •Audience-first planning with org context
  • •Polls, drip and conversational surveys
  • •Auto-generated needs reports with citations
See Needs Analysis

Skills Intelligence

“What skills does our workforce have today?”

See the skills you actually have

A continuous skills engine that infers competencies from real work — meetings, roleplays, learning, code, tickets — and reconciles them with self-declared and HRIS data.

  • •Inference from work signals, not surveys
  • •Validation loop with explainable evidence
  • •Role-fit and gap analysis at scale
See Skills Intelligence

Work Intelligence

“What work do they do, and where does AI fit?”

Plan for AI at the task level

Decomposes roles into tasks, scores AI exposure with the Stanford Human Agency Scale, and forecasts where to augment, automate or hold human-in-the-loop.

  • •Task-level AI impact scoring
  • •Augmentable vs automatable forecasts
  • •Capacity plans aligned to org goals

You're here

Built with Responsible AI guardrails

Decision-support, not deterministic. Audit trails on every score. Works-council friendly. EU AI Act ready. Customers using Work Intelligence stay in control of how outputs feed into people decisions.

Enterprise-ready security & ethics

EU AI Act

Ready

Stanford HAI

Aligned

Audit logs

Every score

Opt-out

Per employee

Work Intelligence FAQs

How does Shiken score AI impact?
Each task in a role is scored against the Stanford Human Agency Scale — Human, Augmentable, Automatable — using a structured rubric that considers task complexity, judgement, regulation, available tooling, and your organisation's recorded work. Confidence is banded by depth of evidence. The rubric updates as new model capabilities and tools land.
Where does the task data come from?
Three sources, layered: (1) public role-and-task libraries (O*NET, ESCO occupational mappings), (2) your own job descriptions and role architectures, (3) inferred tasks from recorded work — meetings, tickets, code, learning. The deeper your Shiken adoption, the sharper the task model.
Is this just about replacing people with AI?
No. The default lens is augmentation, not automation. Most Shiken customers run Work Intelligence to redesign work for human-AI teams — improving capacity, throughput and quality without headcount cuts. Augmentable scoring is intentionally the largest bucket in the rubric.
How does this connect to Skills Intelligence?
Tightly. When Work Intelligence identifies an Augmentable task that needs an AI-pair-programming skill, Skills Intelligence checks whether the people in that role have it. If not, the system recommends the learning path that captures the augmentation — and measures the lift over time.
How accurate is the forecast?
Forecasts are explicitly confidence-banded — we show the range, not a single number. Accuracy improves materially with Shiken adoption depth: 30 days of meetings and learning data is enough to anchor most forecasts, 90 days makes them tight.
Is this ethical / legal in our jurisdiction?
Work Intelligence outputs are advisory and explainable, not deterministic. Customers in EU/UK use it as a decision-support tool alongside works councils and consultation processes. Audit trails, opt-outs and role-redesign playbooks are built in. We can share our Responsible AI framework on request.

Plan for AI like an asset, not a vibe

Work Intelligence is in active design-partner development. Join the customers building the playbook for AI capacity planning.

Apply to design-partner programmeSee Skills Intelligence