# AI Writing Tool Selection Card Use this card to compare two AI writing tools for **one recurring writing job**. Evaluate the complete path to acceptable final work, not just the first draft. > **Core rule:** Choose the tool that makes the finished work easier, not the tool that makes the first 30 seconds look impressive. --- ## 1. Define the writing job **Writing job:** ____________________________________________________________________ **Finished artifact:** ____________________________________________________________________ **Starting material:** ____________________________________________________________________ **What should remain human-led?** ____________________________________________________________________ ____________________________________________________________________ **Where does the current process create the most friction?** ____________________________________________________________________ ____________________________________________________________________ ### Current manual baseline **Approximate time to acceptable final work:** ____________________________________________________________________ **What usually requires correction or review?** ____________________________________________________________________ ____________________________________________________________________ --- ## 2. Set the boundaries **Information the tool may use:** ____________________________________________________________________ ____________________________________________________________________ **Information that must remain outside the tool:** ____________________________________________________________________ ____________________________________________________________________ **Evidence or sources that must remain verifiable:** ____________________________________________________________________ ____________________________________________________________________ **Required reviewer or approval:** ____________________________________________________________________ **Who owns the final result?** ____________________________________________________________________ --- ## 3. Select two finalists ### Finalist A **Tool:** ________________________________________________ **Plan or configuration:** _________________________________ ### Finalist B **Tool:** ________________________________________________ **Plan or configuration:** _________________________________ Before testing, verify any product facts your decision depends on using current vendor documentation. - [ ] Plan availability - [ ] Usage limits - [ ] File support - [ ] Integrations - [ ] Workspace or administrative controls - [ ] Data terms - [ ] Pricing - [ ] Other workflow-critical features: __________________________ --- ## 4. Score WORK Fit Score each finalist from **0 to 3** on each dimension. ### W: Work Can the tool handle the actual writing job? - **0:** Poor fit for the real job - **1:** Useful in places, but major reconstruction remains - **2:** Handles most of the defined job - **3:** Strong fit for the job and finished output **Finalist A:** ____ / 3 **Finalist B:** ____ / 3 ### O: Oversight Can a person realistically review the output and remain accountable for the finished work? - **0:** Reliable review is impractical - **1:** Heavy checking with weak controls - **2:** Review is manageable with explicit checks - **3:** Review and accountability fit naturally **Finalist A:** ____ / 3 **Finalist B:** ____ / 3 ### R: Repeatability Does this recurring job benefit from reusable setup, instructions, context, templates, or integrations? - **0:** Mostly one-off work - **1:** Little reusable value - **2:** Recurring work benefits from reusable setup - **3:** Highly repeatable workflow **Finalist A:** ____ / 3 **Finalist B:** ____ / 3 ### K: Knowledge Can the required approved information be supplied safely and reliably? - **0:** Required context cannot be used appropriately - **1:** Context is incomplete or cumbersome - **2:** Required context can be supplied with manageable effort - **3:** Approved knowledge fits cleanly into the workflow **Finalist A:** ____ / 3 **Finalist B:** ____ / 3 ### WORK Fit total **Finalist A:** ____ / 12 **Finalist B:** ____ / 12 > A WORK Fit score creates a shortlist. It does not override a hard failure. --- ## 5. Check for hard failures | Hard-failure check | Finalist A | Finalist B | |---|---|---| | Requires information that should not enter the tool | [ ] No [ ] Yes | [ ] No [ ] Yes | | Important claims can be traced to adequate evidence | [ ] Yes [ ] No | [ ] Yes [ ] No | | Required human review remains in the workflow | [ ] Yes [ ] No | [ ] Yes [ ] No | | Someone is accountable for the final result | [ ] Yes [ ] No | [ ] Yes [ ] No | | Consequential actions remain under required approval | [ ] Yes [ ] No | [ ] Yes [ ] No | | Output is practical to inspect and verify | [ ] Yes [ ] No | [ ] Yes [ ] No | ### Hard-failure result **Finalist A:** - [ ] Continue - [ ] Restrict or redesign before testing - [ ] Remove from consideration **Finalist B:** - [ ] Continue - [ ] Restrict or redesign before testing - [ ] Remove from consideration --- ## 6. Run the same four tests Use comparable instructions, approved source material, output expectations, and review standards. ### Test 1: Ordinary case Can the tool handle a normal example of the job? **Finalist A notes:** ____________________________________________________________________ ____________________________________________________________________ **Finalist B notes:** ____________________________________________________________________ ____________________________________________________________________ ### Test 2: Incomplete case What happens when important information is missing? **Finalist A notes:** ____________________________________________________________________ ____________________________________________________________________ **Finalist B notes:** ____________________________________________________________________ ____________________________________________________________________ ### Test 3: Conflicting case What happens when instructions or sources conflict? **Finalist A notes:** ____________________________________________________________________ ____________________________________________________________________ **Finalist B notes:** ____________________________________________________________________ ____________________________________________________________________ ### Test 4: Stop-or-escalate case Does the tool recognize when it should stop, ask for clarification, or leave the decision to a person? **Finalist A notes:** ____________________________________________________________________ ____________________________________________________________________ **Finalist B notes:** ____________________________________________________________________ ____________________________________________________________________ --- ## 7. Count the cleanup Do not stop the evaluation when the first draft appears. Track the work required to reach an acceptable final result. | Cleanup category | Finalist A | Finalist B | |---|---|---| | Factual repair | | | | Missing requirements | | | | Tone or clarity repair | | | | Source verification | | | | Formatting and transfer | | | | Failed handoffs | | | | Maintenance | | | ### Time to acceptable final work **Current manual process:** _________________________________ **Finalist A workflow:** ____________________________________ **Finalist B workflow:** ____________________________________ --- ## 8. Check the cost of switching Would adopting either finalist require: - [ ] Rebuilding reusable instructions - [ ] Recreating context - [ ] Changing review procedures - [ ] Training people - [ ] Duplicating or moving files - [ ] Changing integrations - [ ] Maintaining another workspace - [ ] Changing approval or security procedures - [ ] Other: ______________________________________________ **Notes:** ____________________________________________________________________ ____________________________________________________________________ --- ## 9. Make the decision Select one outcome: - [ ] **Adopt narrowly:** Use the selected tool for the tested writing job. - [ ] **Revise and retest:** Fix a specific problem and repeat the same trial. - [ ] **Restrict:** Limit the inputs, outputs, audience, or authority given to the tool. - [ ] **Require human approval:** Keep a person at the required final approval point. - [ ] **Reject:** Choose another candidate or keep the current process. **Selected tool, if any:** ____________________________________________________________________ **Why?** ____________________________________________________________________ ____________________________________________________________________ ____________________________________________________________________ **What would cause this decision to be reviewed again?** ____________________________________________________________________ ____________________________________________________________________ Possible triggers include changes to the product, plan, workflow, data, integrations, audience, volume, or approval process. **Review date:** ___________________________________________ **Decision owner:** ________________________________________ --- ## Final check - [ ] I know exactly which writing job this decision covers. - [ ] I know what should remain human-led. - [ ] I know what information may and may not enter the tool. - [ ] Important claims can be checked against appropriate evidence. - [ ] I know which failures would eliminate or restrict the tool. - [ ] I counted review and cleanup work, not just drafting speed. - [ ] I know what happens if the tool stops working or becomes unsuitable. **Choose the tool that makes the finished work easier, not the tool that makes the first 30 seconds look impressive.**