Task Mapper

Map a workflow. Score every step.

GenAI tools can draft a process diagram from a paragraph, but one-shot models miss the nuances that decide whether automation pays off — judgment calls, exception paths, criticality, data ownership. The structured worksheet captures those; the editor scores each step deterministically. Pick a worked example below, edit to match your reality, export to PDF.

How to read the evidence chips

  • Green-dot chips are Tier 1 cases (peer-reviewed studies, NBER working papers, government findings). Blue-dot chips are Tier 2 (business press, corporate disclosures). Click any chip for the source detail.
  • Workflow headers carry a theoretical anchor (UTAUT, Kotter, DeLone & McLean, TAM) showing the management-research model the workflow maps to.
  • AI-assist time-saved figures with a green-dot chip are anchored to specific productivity studies. Estimates marked Illustrative are typical-report archetypes without a single anchoring case.
  • Step durations (e.g. "60 min to define a need") are illustrative archetypes throughout. The cited cases support magnitude and direction of AI gains, not per-task minutes.
Limits to keep in mind

Worked examples

Initiate a product acquisition

You have decided you need to buy a tool, service, or piece of software. From here, what does it actually take to get a purchase order issued?

Audience: Any individual contributor sponsoring a purchase, not procurement specialists.

Frequency: Most people do this 1-3 times per year, which is exactly why it is painful.

Anchored in

Estimated time, mapped end to end

Without AI

8 hr 15 min

With AI assist

3 hr 35 min

Time saved

~57%

Step times are illustrative archetypes, not measured per-task figures. AI time-saved estimates marked with a green chip are anchored to specific cited studies (BCG-Harvard, NBER, BBVA, ZoomInfo). Estimates marked "Illustrative" are typical reports without a single anchoring case. Magnitude and direction are supported by the cited cases; absolute minutes will vary with role, team, and AI fluency.

1

Articulate the need and write requirements

Write down the problem you are solving, who is affected, what the solution must do, and what would disqualify a vendor. This becomes the basis for every later step.

Systems

Internal wiki / Notion / ConfluenceYour own notes

Authentication

Standard SSO

Typical time

1 hr

Illustrative

Why this is tedious

  • It is tempting to skip this and start shopping, which causes rework later.
  • Most people never use a template, so requirements are inconsistent.

Where AI can help

Draft~25 min saved

Paste a rough description of the problem and ask for a structured requirements doc with sections for problem statement, must-haves, nice-to-haves, and disqualifiers.

Generate / check checklist~15 min savedIllustrative

Ask for the standard procurement-readiness checklist (data classification, integration constraints, regulatory considerations) so you do not forget categories.

2

Check the approved-vendor catalog

Before you go shopping, find out whether procurement already has an approved vendor that covers your need. If yes, the path is faster.

Systems

Coupa / Ariba / JaggaerInternal procurement portal

Authentication

Often a separate login from your main SSO; sometimes role-gated.

Typical time

30 min

Illustrative

Why this is tedious

  • Catalog search uses a different vocabulary than how you describe the need.
  • You may not have catalog read access and have to ask procurement first.

Where AI can help

Reformat / translate~10 min savedIllustrative

Convert your need into the catalog's vocabulary - ask for 5-10 alternative search terms a procurement system might use for the same capability.

Q&A on a document~15 min saved

Paste catalog descriptions for 3-5 candidates and ask which one most closely matches your written requirements.

Verify the match in the catalog itself before choosing.

3

Identify candidate vendors

If nothing in the approved catalog fits, find 3-5 vendors that plausibly meet your requirements, including the incumbents in the space.

Systems

Web searchGartner / G2 / ForresterPeer Slack channelsLinkedIn

Authentication

Standard SSO

Typical time

1 hr 30 min

Illustrative

Why this is tedious

  • Every vendor's marketing site sounds the same.
  • Hard to tell which vendors are in your size segment vs enterprise-only.
  • Analyst reports are paywalled.

Where AI can help

Lookup / synthesize~30 min savedIllustrative

Ask for a starting list of vendors in the category with one-sentence positioning for each, then verify each independently.

Vendor lists can be out-of-date or include defunct companies. Always verify on the vendor's own site.

Summarize~15 min saved

Paste a vendor's pricing or product page and ask for a 5-bullet summary against your specific requirements.

4

Compare options

Build a comparison across the candidates: capability fit, integration, pricing model, contract terms, security posture.

Systems

Vendor websitesTrust centersSales-shared decksDemo notes

Authentication

Standard SSO

Typical time

2 hr

Illustrative

Why this is tedious

  • Each vendor uses a different format for their pricing and terms.
  • Security questionnaires are buried in trust portals behind a click-through.

Where AI can help

Compare options~45 min saved

Paste descriptions of all candidates and ask for a comparison table along the requirement dimensions you defined in step 1.

Extract structured data~25 min saved

Paste a vendor's MSA or terms page and extract specific fields: data location, indemnification cap, termination notice period, AI training clause.

Treat extracted terms as a starting point. Legal still has to review.

5

Confirm budget headroom and cost code

Find the right cost center, confirm there is unspent budget in the relevant line, and identify the fiscal-year implications.

Systems

Workday / NetSuite / Oracle FinancialsFP&A dashboardEmail to your finance partner

Authentication

Frequently a separate finance portal login or VPN-gated.

Typical time

45 min

Illustrative

Why this is tedious

  • You are not sure which cost center to charge.
  • Reports are slow to load and have unintuitive filter UIs.
  • Ambiguity between calendar year and fiscal year deadlines.

Where AI can help

Generate / check checklist~15 min savedIllustrative

Ask for the questions you should ask your finance partner so the email is complete the first time and you do not need a second round.

Reformat / translate~10 min savedIllustrative

Convert a vendor's annual price into the monthly figure your dashboard uses, with proration for the fiscal-year remaining months.

Do not paste sensitive financial figures into a public AI tool. Use the enterprise-approved tool or paste only the math problem with placeholders.

6

Pin down timing and dependencies

Confirm the vendor's lead time, security review duration, legal review queue, and any fiscal-year freeze.

Systems

Vendor emailInternal security wikiLegal intake formFinance calendar

Authentication

Standard SSO

Typical time

30 min

Illustrative

Why this is tedious

  • Each gating function quotes a different review time and you have to chase them.
  • Q4 freezes catch people by surprise.

Where AI can help

Draft~15 min saved

Draft three short, specific emails (vendor / security / legal) asking each one for their committed review time so you can build a realistic timeline.

7

Submit the request in the procurement portal

Find the right form, fill in the 30-50 fields it expects, attach quotes / requirements / security review, and submit.

Systems

Procurement portalDocuSignEmail

Authentication

SSO plus, often, vendor-side authentication.

Typical time

1 hr

Illustrative

Why this is tedious

  • Field labels are written for procurement specialists, not requesters.
  • Required fields appear conditionally and you discover them only on submit.
  • The justification field expects a specific structure that is not documented.

Where AI can help

Draft~25 min saved

Draft the business-justification narrative from your requirements doc, in the structure procurement reviewers expect: problem, solution, alternatives considered, risk of inaction.

Q&A on a document~10 min savedIllustrative

Paste the form's field labels and ask which ones likely need backup documentation, so you have everything ready before opening the form.

8

Route for approvals and follow up

Manager, finance, security, legal, and sometimes data privacy each need to approve. Track status, nudge politely, answer questions.

Systems

Procurement portal status pageEmailSlack

Authentication

Standard SSO

Typical time

1 hr

Illustrative

Why this is tedious

  • Status pages do not say who is currently blocking.
  • Approvers ask for context already in the request.
  • It feels rude to nudge but everything stalls if you do not.

Where AI can help

Draft~15 min saved

Draft an approver-facing 4-bullet summary you can paste at the top of the request or into a follow-up email.

Draft~10 min saved

Draft a polite nudge email that references the specific approval that is pending and a concrete deadline reason.

Your process worksheet

Map a process you actually run. Each step gets a deterministic AI-suitability score so you know where automation pays off and where it does not. Saved to your browser; export the worksheet to PDF and bring it to a Diagnostic.

Reference: the ten fields the worksheet captures

The first six fields describe the step. The last four are the nuances that LLMs struggle with — the things a person who has actually done the work needs to write down before any automation makes sense.

1Step

What you actually do, in one sentence.

2Systems

Every tool, portal, doc, or inbox you touch.

3Authentication

Where the friction is - separate logins, role gates.

4Typical time

How long this step takes, honestly.

5Pain points

Why this is tedious for someone who does it rarely.

6AI-assist pattern

Which of the eight patterns might safely help, and what you would have to verify.

7Actor / role

Who actually does this step. Different roles bring different judgment.

8Decision or exception

The branch a script can't make on its own - the place LLMs are most likely to get it wrong.

9Data inputs

The documents and fields the step depends on. Structured data makes automation tractable.

10Criticality

If automation gets this wrong, what is the cost? High-criticality steps need a human in the loop.

The eight AI-assist patterns

Draft

Generate a first-draft of a document, email, or form field from rough notes.

Lookup / synthesize

Synthesize a starting answer from open sources you can verify.

Summarize

Condense a long document into the parts that affect your decision.

Compare options

Build a side-by-side comparison from descriptions you paste in.

Extract structured data

Pull structured fields (dates, amounts, names) out of unstructured text.

Generate / check checklist

Produce or evaluate against a checklist for a recurring task.

Reformat / translate

Reformat content between styles, formats, or audiences.

Q&A on a document

Ask questions of a long document instead of reading it end to end.

A note on safe use

Do not paste internal financial figures, employee data, customer data, or anything covered by NDA into a public AI tool. Use the enterprise-approved AI tool when you have one, or paste only the abstracted shape of the problem when you do not. The patterns above all work fine on the abstracted version.

Bring your worksheet to a Diagnostic

Export a saved process, send it over, and we will use it to focus a 60-minute scoping call. The conversation does what the worksheet can't.

See Diagnostic Review